SPIE Defence, Security+Sensing
25 - 29 April 2011
Orlando World Center Marriott Resort & Convention Center
Orlando, Florida, USA
 

Hyperspectral Optics and Systems request a quote

Long-wave infrared (8 to 14 μm) hyperspectral imager based on an uncooled thermal camera and the traditional CI block interferometer 

Paper 8012-108 of Conference 8012
Date: Friday, 29 April 2011

Author(s): Dario Cabib, Moshe Lavi, Amir Gil, CI Systems (Israel) Ltd. (Israel)


Since the early '90's CI has been involved in the development of FTIR hyperspectral imagers based on a Sagnac or similar type of interferometer. CI also pioneered the commercialization of such hyperspectral imagers in those years. After having developed a visible version based on a CCD and a 3 to 5 micron infrared version based on a cooled InSb camera, it has now developed an LWIR version based on an uncooled infrared camera for the 8 to 14 microns range. The system has applications in gas cloud imaging among others. In this paper we will present the design and performance of the system.

Compact high-resolution VIS/NIR hyperspectral sensor 

Paper 8032-31 of Conference 8032
Date: Tuesday, 26 April 2011

Author(s): Timo Hyvärinen, Esko Herrala, Specim Spectral Imaging Ltd. (Finland)


This paper presents an extremely compact and high performance push-broom hyperspectral imager operating in the VIS/NIR region (380 to 1000 nm). The imager weighs only 1.4 kg, and has a format optimized for installation in small UAV payload compartments and gimbals. It features high light throughput, negligible keystone and smile distortion, 1300 spatial pixels and image rate of 200 Hz. A higher resolution version with 2000 spatial pixels runs at up to 120 images/s. The camera achieves, with spectral sampling of 5 nm, an outstanding SNR of 800:1, orders of magnitude higher than any current compact VIS/NIR imager.

Visible/near-infrared hyperspectral sensing of solids under controlled environmental conditions 

Paper 8018-20 of Conference 8018
Date: Wednesday, 27 April 2011

Author(s): Bruce E. Bernacki, Norman C. Anheier, Jr., Albert Mendoza, Bradley G. Fritz, Timothy J. Johnson, Pacific Northwest National Lab. (United States)


We describe the use of a wind tunnel for conducting controlled passive hyperspectral imaging experiments. Passive techniques are potentially useful for detecting explosives, solid-phase chemicals and other materials of interest from a distance so as to provide operator safety. The Pacific Northwest National Laboratory operates a wind tunnel facility that can generate and circulate artificial atmospheres to control lighting, humidity, temperature, aerosol burdens, and obscurants. We will present recent results describing optimized sensing of solids over tens of meters distance using both visible and near-infrared cameras, as well as the effects of certain environmental parameters on data retrieval.

Toward integration of AOTF-based hyperspectral imager in visual surveillance applications 

Paper 8048-25 of Conference 8048
Date: Tuesday, 26 April 2011

Author(s): Sergiy Fefilatyev, Univ. of South Florida (United States); Ronald G. Rosemeier, Brimrose Corp. of America (United States)


Such characteristics as small form-factor, portability, and low-cost have made AOTF-based hyperspectral imagers attractive for use in many applications. This paper explores several aspects of the use of AOTF-based hyperspectral imagers in visual surveillance. We present the implementation of the low-cost miniaturized hyperspectral imaging device based on AOTF-filter. The techniques of calibration, image acquisition, and hyperspectral data processing for such device are shown. In experimental part we report on the results of experiments to discriminate materials in hyperspectral images of static outdoor scenes and discuss the extension of such application to certain dynamic scenes by integrating it with conventional surveillance equipment.

Visualization of hyperspectral images using bilateral filtering with spectral angles 

Paper 8050-70 of Conference 8050
Date: Tuesday, 26 April 2011

Author(s): Jai-Hoon Lee, Ayoung Heo, Won-Chul Choi, Seo Hyun Kim, Dong-Jo Park, KAIST (Korea, Republic of)


In this paper, a new bilateral filter with spectral angles and a visualization scheme for hyperspectral images are presented. The conventional bilateral filter used to be implemented using a position vector and the intensity value at each pixel in the scene. Since hyperspectral image data can provide a spectrum vector which has hundreds of bands at each pixel, we propose a bilateral filter by using spectral angles. This bilateral filter with spectral angles can be used for extracting and preserving the spectrum edges of the hyperspectral image. The visualization scheme for hyperspectral images exploiting the bilateral filter with spectral angles has been also proposed. The simulation results show that the proposed scheme facilitates the anomaly detection and classification of objects in the hyperspectral scenes.

Generalized fusion: a new framework for hyperspectral detection 

Paper 8048-1 of Conference 8048
Date: Monday, 25 April 2011

Author(s): Peter Bajorski, Rochester Institute of Technology (United States)


The purpose of this paper is to introduce a general type of detection fusion that allows combining a set of basic detectors into one, more versatile, detector. The new approach shown in this paper is especially promising in the context of recent geometric and topological approaches that produce complex structures for the background and target spaces. We show specific examples of generalized fusion and present some results on false alarm rates and probabilities of detection of fused detectors. We show that Alan Schaum's continuum fusion is a special case of generalized fusion.

Log-linear Laplacian ratio (LLLR) algorithm for spectral detection using laboratory signatures 

Paper 8048-3 of Conference 8048
Date: Monday, 25 April 2011

Author(s): Brian J. Daniel, Alan P. Schaum, U.S. Naval Research Lab. (United States)


The potential of a new class of detection algorithms is demonstrated with the publically available RIT test data set. The continuum fusion (CF) methodology is applied to an affine target subspace model, which assumes that uncertainty in prediction of in-situ signature spectra from laboratory spectra is mostly confined to a one-dimensional region. The new algorithm results from imposing a CF methodology on a conventional GLRT-based algorithm. Performance is enhanced in two ways. First the Gaussian clutter model is replaced by a Laplacian distribution, which is not only more realistic in its tail behavior but, when used in a hypothesis test, also creates decision surfaces more selective than the hyperplanes associated with linear matched filters. Second, a log-Laplacian fusion flavor is devised that further increases the selectivity of decision surfaces to the point where outliers are also rejected.

Algorithm for detecting anomaly in hyperspectral imagery using factor analysis 

Paper 8048-4 of Conference 8048
Date: Monday, 25 April 2011

Author(s): Edisanter Lo, Susquehanna Univ. (United States)


Hyperspectral imaging is particular useful in remote sensing to identify a small number of unknown man-made objects in a large natural background. An algorithm based on factor analysis for detecting such anomalies in a high-dimensional data set from hyperspectral imagery and its performance in comparison with conventional algorithms are presented in this talk. Under the factor model, each observable component of the background pixel is postulated to be a linear function of a few unobservable common factors with unknown factor loadings plus a single latent specific variate. The covariance of the pixel is assumed to be in factored form which is a product of the loading matrix and its transpose plus the diagonal covariance matrix of the specific variates. The anomaly detector is defined to be the Mahalanobis distance of the resulting residual between the pixel and its predicted value. Experimental results using Visible-Near-Infra-Red hyperspectral imagery are presented.

Extension and implementation of model-based hyperspectral change detection 

Paper 8048-5 of Conference 8048
Date: Monday, 25 April 2011

Author(s): Joseph Meola, Air Force Research Lab. (United States)


Within the hyperspectral community, change detection is a continued area of interest as it provides an avenue for detecting subtle CC&D targets in complex environments. Complicating the problem of change detection is the presence of shadow differences and parallax/misregistration error between the scenes which often produce the appearance of change. The change detection problem can be formulated using a physical model describing the illumination reaching the sensor on separate occasions. Here the model-based approach is extended to include spatial information present in the scene to help with the problems associated with misregistration/parallax and to help improve shadow estimates associated with the model.

Hierarchical image segmentation for context-dependent anomalous change detection 

Paper 8048-6 of Conference 8048
Date: Monday, 25 April 2011

Author(s): James Theiler, Lakshman Prasad, Los Alamos National Lab. (United States)


The challenge of finding small targets in big images lies in the characterization of the background clutter. The more homogeneous the background, the more distinguishable a typical target will be from its background. One way to homogenize the background is to segment the image into distinct regions, each of which is individually more homogeneous, and then to treat each region separately. In this paper we will report on experiments in which the target is an anomalous change, and the segmentation strategy is a hierarchical tree-based scheme.

Change detection using mean-shift and outlier-distance metrics 

Paper 8048-7 of Conference 8048
Date: Monday, 25 April 2011

Author(s): Joshua D. Zollweg, Rochester Institute of Technology (United States); David B. Gillis, U.S. Naval Research Lab. (United States); Ariel Schlamm, David W. Messinger, Rochester Institute of Technology (United States)


Change detection with application to wide-area search seeks to identify where interesting activity has occurred between two images. Since there are many different classes of change, one metric may miss a particular type of change. Therefore, it is potentially beneficial to select metrics with complementary properties. With this idea in mind, a new change detection scheme was created using mean-shift and outlier-distance metrics. Using these metrics in combination should identify change more completely than either individually. An algorithm using both metrics was developed and tested using registered sets of multi and hyperspectral imagery.

Large scale micro-Fabry-Perot optical filter arrays 

Paper 8054-4 of Conference 8054
Date: Monday, 25 April 2011

Author(s): Ali A. Abtahi, Aerospace Missions Corp. (United States); Peter B. Griffin, Stanford Univ. (United States); Ricky J. Morgan, Usha Raghuram, Aerospace Missions Corp. (United States); Francisco Tejada, Sensing Machines (United States); Frida S. Vetelino, Aerospace Missions Corp. (United States)


Fabry-Perot filter arrays have been fabricated comprised of six million individual filters using standard semiconductor processing techniques. The current 3000 x2000 array consists of 5x5 subarrays in which each of the nine micron wide Fabry-Perot filters in the subarray has a different color response. The 5x5 subarray is replicated to create a 600x400 matrix of 5x5 micro Fabry-Perot filter subarrays. This Fabry-Perot matrix has been integrated with a commercially available panchromatic 6 megapixel CCD focal plane array to create a 25 color hyperspectral camera with 600x400 imaging pixels. Near- UV, visible and NIR filter arrays have been fabricated.

Anomaly detection of man-made objects using spectro-polarimetric imagery 

Paper 8048-11 of Conference 8048
Date: Monday, 25 April 2011

Author(s): Brent D. Bartlett, Ariel Schlamm, Carl Salvaggio, David W. Messinger, Rochester Institute of Technology (United States)


In the task of automated anomaly detection, it is desirable to find regions within imagery that contain man-made structures or objects. In the task of automated anomaly detection, it is desirable to find regions within imagery that contain man-made structures or objects. The task of separating these signatures from the scene background and other naturally occurring anomalies can be challenging. This task is even more difficult when the spectral signatures of the man-made objects are designed to closely match the surrounding background. As new sensors emerge that can image both spectrally and polarimetrically, it is possible to utilize the polarimetric signature to discriminate between many types of man-made and natural anomalies. In this work, an anomaly detection scheme is implemented which makes use of the spectral Stokes imagery collected of a real scene to find man-made objects.

Smart compression using high-dimensional imagery 

Paper 8063-10 of Conference 8063
Date: Monday, 25 April 2011

Author(s): Dalton S. Rosario, U.S. Army Research Lab. (United States)


We present a method for the disadvantaged user (Warfighter remotely carrying low bandwidth devices), featuring "smart" compression of high dimensional imagery from passive hyperspectral (HS) sensors. The method uses the application of anomaly detection closer to the sources, transmitting only the essential information (spectral anomalies) to the users for further analysis. The method's uniqueness relies on a binomial distribution model for spectral sampling. Its advantages over existing methods, includes (i) no prior imagery segmentation requirement, (ii) little sensitivity to its free parameter, and (iii) no prior knowledge of target scales. Experimentation results using HS imagery are promising for smart compression.

Target detection using multiple hyperspectral imagers and physics-based models 

Paper 8048-13 of Conference 8048
Date: Monday, 25 April 2011

Author(s): Emmett Ientilucci, John P. Kerekes, Rochester Institute of Technology (United States); Arnab Shaw, Gitam Technologies (United States)


The use of multiple hyperspectral imagers will be explored with applications to target detection.

An automated method for identification and ranking of hyperspectral target detections 

Paper 8048-14 of Conference 8048
Date: Monday, 25 April 2011

Author(s): William F. Basener, Rochester Institute of Technology (United States)


The basic process of target detection is to apply a detection filter to a hyperspectral image to produce a detection plane for each target. We will present a new method for target detection that includes additional spatial processing, multiple detection and identification metrics such as F-Test, ACE, unmixing and sub-pixel spectral visualization to build a more complete understanding of the image. The result is a draft detection report of the objects in the image ranked according to the confidence of the identification of each object. This method can be used for faster ground processing as well as on board processing, and the detection reports are much smaller than the image files enabling fast communication to users.

Enhancement of flow-like structures in hyperspectral imagery using anisotropic diffusion 

Paper 8048-15 of Conference 8048
Date: Monday, 25 April 2011

Author(s): Maider Marin-McGee, Miguel Velez-Reyes, Univ. de Puerto Rico Mayagüez (United States)


In this work, we are studying nonlinear anisotropic diffusion filtering for enhancement of flow-like structures, or coherence enhancement, in hyperspectral and multispectral imagery. Anisotropic diffusion is commonly used for edge enhancement by promoting diffusion in the direction of highest fluctuation of the contrast average within a neighborhood. For CE, the diffusion is promoted along the direction of lowest fluctuation in the neighborhood to account for the coherence of the structures in the image. This paper presents the theoretical development for the coherence enhancement algorithm using a diffusion PDE. Examples using hyperspectral and multispectral imagery are presented.

Image mapping spectrometry: a novel hyperspectral platform for rapid snapshot imaging 

Paper 8048-21 of Conference 8048
Date: Tuesday, 26 April 2011

Author(s): Tomasz S. Tkaczyk, Rice Univ. (United States)


This paper presents the Image Mapping Spectrometry a new snapshot hyperspectral imaging platform. Its applications span from basic science microscopy implementations through endoscopic diagnostics and reach to remote sensing use. The IMS replaces the camera in a digital imaging system, allowing one to add parallel spectrum acquisition capability and to maximize the signal collection. As such the IMS allows obtaining full spectral information in the image scene instantaneously at rates of 100 frames/second or higher. This presentation provides fundamentals of IMS operations based on image mapping, describes examples of designs and demonstrates the platform flexibility for use in numerous applications.

A Fabry-Perot interferometer with a spatially variable resonance gap employed as a Fourier transform spectrometer 

Paper 8048-22 of Conference 8048
Date: Tuesday, 26 April 2011

Author(s): Paul G. Lucey, Univ. of Hawai'i (United States); Jason Akagi, Spectrum Photonics, Inc. (United States)


We demonstrate a Fourier transform spectrometer (FTS) using a stationary Fabry-Perot interferometer with the gap between its partially reflecting layers varying orthogonal to the optical axis to produce a gradient in optical path different at a detector. The gradient produces a period fringe pattern that can be analyzed with standard FTS techniques. The device has some limitations in spectral resolution owing to the influence of incidence angle on the Fabry-Perot interferometer and these are quantified. Experiments in the visible and IR demonstrate the feasibility of this method for spectroscopy.

Estimation of the attenuation coefficient of the water body using polarimetric observations 

Paper 8030-2 of Conference 8030
Date: Tuesday, 26 April 2011

Author(s): Alberto Tonizzo, Tristan Harmel, Amir Ibrahim, Alex Gilerson, Samir Ahmed, The City College of New York (United States)


The degree of polarization (DOP) of the underwater light field in oceanic waters is related to the single scattering albedo of suspended particles which in turn represents the ratio of the scattering coefficient to the attenuation coefficient. The validity of the above approach for the whole visible spectral range was recently confirmed by us using experimental data obtained with our recently developed underwater polarimeter. This then opens up the possibility for estimation of attenuation coefficients from measurements of the Stokes components of the upwelling underwater light field which is not possible from unpolarized measurements of the remote sensing reflectance. Results of simulations using vector radiative transfer code are compared with below and above water experimental observations to assess the validity of the results.

The enhanced MODIS airborne simulator hyperspectral imager 

Paper 8048-23 of Conference 8048
Date: Tuesday, 26 April 2011

Author(s): Daniel Guerin, Ted Graham, John Fisher, Brandywine Optics, Inc. (United States)


The NASA Enhanced MODIS Airborne Simulator Hyperspectral Imager (EMAS-HSI) is designed to augment the resolution and monitor the radiometric stability of the existing MODIS Airborne Simulator (MAS). The system is designed for missions on the ER-2 and Global Hawk platforms. EMAS-HSI is a push-broom system that uses two Offner spectrometers to cover the 380-2400 nm spectral range, sharing the FOV of an all-reflective telescope with at 50° full field-of-view. The EMAS-HSI system performance trades optimize system performance in the spectral regions used by the multi-spectral MODIS satellite, with land, cloud, atmospheric and water bands given the greatest deference.

An interference microfilter array with tunable spectral response for each pixel 

Paper 8048-24 of Conference 8048
Date: Tuesday, 26 April 2011

Author(s): Frida E. Strömqvist Vetelino, Ali A. Abtahi, Aerospace Missions Corp. (United States); Peter B. Griffin, Stanford Univ. (United States); Ricky J. Morgan, Usha Raghuram, Aerospace Missions Corp. (United States); Francisco Tejada, Sensing Machines (United States)


A MEMS standing wave spectrometer is turned into a wavelength tunable band-pass filter by the addition of a reflective coating. It results in the standing wave filter (SWF), a miniaturized Fabry-Perot band-pass filter with a semi-transparent detector that can be incorporated into a pixel-tunable focal plane array, suitable for hyperspectral imaging applications. The performance of the SWF is optimized with thin film optics modeling and FDTD simulations. The SWF concept is taken from an ideal device to a design that can be fabricated. The limiting factors of the SWF are discussed. A comparison between design and fabricated components is included.

Standoff identification and quantification of flare emissions using infrared hyperspectral imaging 

Paper 8024-29 of Conference 8024
Date: Tuesday, 26 April 2011

Author(s): Kevin C. Gross, Air Force Institute of Technology (United States); Simon Savary, Telops (Canada); Pierre Tremblay, Univ. Laval (Canada); Jean-Philippe Gagnon, Vincent Farley, Martin Chamberland, Telops (Canada)


There is growing interest in measuring gaseous emissions to understand their environmental impact. It is thus desired to identify and quantify such emissions, ideally from standoff distances. AFIT and Telops have performed several field experiments, using the Telops Hyper-Cam infrared hyperspectral imager to perform identification and quantification of gaseous emissions from various pollution sources. Recent experiments have focused on turbulent gaseous emissions from sources of great interest from the environmental protection community, such as emergency flares. It is of interest to understand the flare emissions under varying operating conditions. This paper presents the first results of flare emission measurements with the Hyper-Cam.

Hyperspectral processing in graphical processing units 

Paper 8048-27 of Conference 8048
Date: Tuesday, 26 April 2011

Author(s): Michael E. Winter, Edwin Winter, Technical Research Associates, Inc. (United States)


With the advent of the commercial 3D video card in the mid 1990s, we have seen an order of magnitude performance increase with each generation of new video cards. While these cards were designed primarily for visualization and video games, it became apparent after a short while that they could be used for scientific purposes. These Graphical Processing Units (GPUs) are rapidly being incorporated into data processing tasks usually reserved for general purpose computers. It has been found that many image processing problems scale well to modern GPU systems. We have implemented four popular hyperspectral processing algorithms (N-FINDR, linear unmixing, Principal Components, and the RX anomaly detection algorithm). These algorithms show an across the board speedup of at least a factor of 10, with some special cases showing extreme speedups of a hundred times or more.

Modeling of pixel edge effects in a novel micro-filter array for the visible spectrum 

Paper 8014-1 of Conference 8014
Date: Tuesday, 26 April 2011

Author(s): Frida E. Strömqvist Vetelino, Ali A. Abtahi, Aerospace Missions Corp. (United States); Peter B. Griffin, Stanford Univ. (United States); Ricky J. Morgan, Usha Raghuram, Aerospace Missions Corp. (United States)


The modeling of a novel hyperspectral filter array for the visible spectrum, constructed of an array of micron sized Fabry-Perot band-pass filters, is presented. Each filter forms a squared cavity pixel, less than 10 µm wide, resonating at a different wavelength than the neighboring pixels. To study pixel edge effects and pixel cross-talk, 2D FDTD simulations were carried out. Extensive modeling was done for a cavity array with several pixels, and sloped cavity edges were compared to vertical ones. Comparisons of the peak power and spectral bandwidth were made between a finite pixel cavity and a cavity of infinite extent.

A thermal infrared hyperspectral imager for small satellites 

Paper 8044-24 of Conference 8044
Date: Tuesday, 26 April 2011

Author(s): Sarah T. Crites, Paul G. Lucey, Robert Wright, Univ. of Hawai'i (United States)


The Thermal Hyperspectral Imager (THI) is a sensor funded by the NASA EPSCOR (Experimental Project to Stimulate Competitive Research) program and fits into the niche for low-cost, short-lived experimental missions created by the growth of the small satellite market. THI is a low-mass, power efficient thermal hyperspectral imager integrated with a pressure vessel to enable the use of COTS components. THI is based on a Sagnac interferometer, uses a 320x256 microbolometer array, and will collect data at thermal infrared wavelengths in 230-meter pixels with 20 wavenumber spectral resolution from a 400-km Earth orbit.

Evaluation of the GPU architecture for the implementation of target detection algorithms for hyperspectral imagery 

Paper 8048-28 of Conference 8048
Date: Tuesday, 26 April 2011

Author(s): Blas Trigueros-Espinosa, Miguel Velez-Reyes, Nayda G. Santiago-Santiago, Univ. de Puerto Rico Mayagüez (United States)


Target detection in hyperspectral imagery involves processing of large volumes of data, which require hardware platforms with high computational power. In this work, we study the use of Graphics Processing Units (GPUs) as a computing platform for the implementation of target detection algorithms. The GPU implementation was done using the Compute Unified Device Architecture (CUDA) of the NVIDIA GPUs and compared with a multi-core CPU-based implementation. The detection accuracy of the implemented algorithms was evaluated using a set of phantom images simulating traces of different materials on clothing as models for detection of traces of explosives.

Parallel implementation of nonlinear dimensionality reduction methods using CUDA in GPU architecture 

Paper 8048-29 of Conference 8048
Date: Tuesday, 26 April 2011

Author(s): Romel Campana, Vidya B. Manian, Univ. de Puerto Rico Mayagüez (United States)


Manifold learning are important techniques to preserve a nonlinear structure and the objects geometry of nonlinear high-dimensional data in the lower dimension.Manifold learning algorithms are very slow (high computational algorithms) and time consuming in estimating the solution. The goal of this work is to parallelize the three most important manifold learning algorithms to reduce the dimensionality of the hyperspectral images for subsequent application in object segmentation. These three methods are ISOMAP, Local Linear Embedding and Laplacian Eigenmap. The parallelization consists of implementing the bottleneck parts like k-nearest neighbor, shortest path for geodesic distance, Graph Laplacian and other features in the Compute Unified Device Architecture (CUDA) in GPU developed by NVIDIA.

AOTF hyperspectral microscopic imaging for foodborne pathogenic bacteria detection 

Paper 8027-6 of Conference 8027
Date: Tuesday, 26 April 2011

Author(s): Bosoon Park, Jaya Sundaram, Gerald W. Heitschmidt, Seung Chul Yoon, Kurt C. Lawrence, William R. Windham, U.S.D.A. Agricultural Research Service (United States)


The objective of this research is to develop a hyperspectral microscopic imaging (HMI) method to evaluate spectral characteristics of foodborne bacteria. The HMI system consists of a Nikon upright microscope, acousto-optic tunable filters (AOTF), high performance cooled CCD camera, and bright-filed and dark-field illumination. The HMI system was used to scan Salmonella typhimurium with different dilutions. The hyperspectral microscopic images were collected at the wavelength ranges from 450 to 850 nm. In this paper, the AOTF-based hyperspectral microscope imaging method to characterize optical properties of Salmonella typhimurium to apply for rapid detection of foodborne pathogen will be presented.

Real-time georeferencing for an airborne hyperspectral imaging system 

Paper 8048-31 of Conference 8048
Date: Tuesday, 26 April 2011

Author(s): Thomas O. Opsahl, Trym V. Haavardsholm, Atle Skaugen, Ingebrigt Winjum, Norwegian Defence Research Establishment (Norway)


We describe the georeferencing part of FFIs real-time hyperspectral demonstrator system. Using a highly efficient representation of the digital elevation model and raytracing methods from modern computer graphics we are able to georeference high resolution images in real time. By adapting the calculations to match the ground resolution of the digital terrain model, the cameras field of view and typical flight altitude, the method has potential to provide real time georeferencing of even HD video at 60Hz on a DEM with 5 meter resolution when a graphics processor unit is used for processing.

Identification and mapping of night lights signatures using hyperspectral data 

Paper 8048-32 of Conference 8048
Date: Tuesday, 26 April 2011

Author(s): Fred A. Kruse, Naval Postgraduate School (United States); Christopher D. Elvidge, National Oceanic and Atmospheric Administration (United States)


This research demonstrates the use of imaging spectrometer (hyperspectral) data to identify, characterize, and map urban lighting based on spectral emission lines unique to specific lighting types. Spectral features extracted from ProSpecTIR hyperspectral data of Las Vegas, Nevada were compared to measurements made with an Analytical Spectral Devices spectroradiometer. Specific types identified included blue and red neon, high pressure sodium, and metal halide lights. There were also indications spectral mixing or variants of these specific light types. The nature and distribution of lights were used as a surrogate for measurement of urban development.

Advances in hyperspectral LWIR pushbroom imagers 

Paper 8032-32 of Conference 8032
Date: Tuesday, 26 April 2011

Author(s): Hannu Holma, Antti-Jussi Mattila, Timo Hyvärinen, Specim Spectral Imaging Ltd. (Finland)


Two designs of hyperspectral imagers have been under extensive development: one utilizing a microbolometer and another with an MCT FPA. The design and implementation of the high performance, extremely compact imager with MCT FPA and 8 to 12 um spectral range has been completed. The performance with 84 spectral bands and 384 spatial samples has been experimentally verified and NESR of 18 mW/(m2srum) at 10 um wavelength for 300 K target has been achieved. This results SNR of more than 500. The second design based on microbolometer FPA was introduced in 2009. Its improved design has now been finalized with sensitivity improved by a factor of 3 and SNR by 15%.

Validation of technique to hyperspectrally characterize the lower atmosphere with limited surface observations 

Paper 8038-7 of Conference 8038
Date: Tuesday, 26 April 2011

Author(s): Robb M. Randall, Steven T. Fiorino, Michelle F. Gerling, Adam D. Downs, Air Force Institute of Technology (United States)


This paper demonstrates the capability of AFIT/CDE's Laser Environmental Effects Definition and Reference (LEEDR) model to accurately characterize the meteorological parameters and radiative transfer effects of the atmospheric boundary layer with only surface observations of temperature, pressure, and humidity. The LEEDR model is a fast-calculating, first principles, worldwide surface to 100 km, atmospheric propagation and characterization package. This research compares the LEEDR vertical profiles created from input surface observations to actual observations from balloon launches, aircraft, and satellites. Additional comparisons are made to vertical profiles derived from short range numerical weather forecasts.

Development of narrow-band fluorescence indices for the detection of aflatoxin contaminated corn 

Paper 8027-12 of Conference 8027
Date: Tuesday, 26 April 2011

Author(s): Haibo Yao, Zuzana Hruska, Russell Kincaid, Ambrose E. Ononye, Mississippi State Univ. (United States); Robert L. Brown, Deepak Bhatnagar, Thomas E. Cleveland, U.S.D.A. Agricultural Research Service (United States)


Corn contaminated with aflatoxin is toxic to domestic animals as well as humans and thus is of major concern to the food and feed industry. Currently, aflatoxin detection and quantification methods are based on analytical tests. These tests require the destruction of samples, and can be costly and time consuming. This paper describes how narrow-band fluorescence indices were developed for the detection of aflatoxin contamination in corn. It is anticipated that the results would be helpful in the development of real-time detection instrumentation for the corn industry.

Analysis of multispectral signatures of shot 

Paper 8019-33 of Conference 8019
Date: Tuesday, 26 April 2011

Author(s): Mariusz Kastek, Rafal Dulski, Tadeusz Piatkowski, Henryk Madura, Jaroslaw Barela, Henryk Polakowski, Military Univ. of Technology (Poland)


The paper presents some practical aspects of sniper IR signature measurements. Description of particular signatures for sniper shot in typical scenarios has been presented. The measurements were made at field test ground. High precision laboratory measurements were also performed. Several infrared cameras were used during measurements to cover all measurement assumptions. The registration was made in NWIR, SWIR and LWIR spectral bands simultaneously. The infrared cameras have possibilities install optical filters for multispectral measurement. Exemplary sniper IR signatures for typical situation were presented. During the experiments in laboratory and test field was used LWIR imaging spectroradiometer HyperCam. The signatures collected by HyperCam were useful for determination of spectral characteristics of shot.

Aflatoxin contaminated chili pepper detection by hyperspectral imaging and machine learning 

Paper 8027-14 of Conference 8027
Date: Tuesday, 26 April 2011

Author(s): Musa Atas, Yasemin Yardimci Cetin, Alptekin Temizel, Middle East Technical Univ. (Turkey)


Mycotoxins are the toxic secondary metabolites of certain filamentous fungi. They have been demonstrated to cause various health problems in humans, including immunosuppression and cancer. Chili pepper is also prone to aflatoxin contamination during harvesting, production and storage periods. Hyperspectral and multispectral imaging are becoming increasingly important for rapid and nondestructive testing for the presence of such contaminants. We propose a compact machine vision system which employs a neural network with inputs from hyperspectral images for detection of aflatoxin contaminated chili peppers. Feature selection scheme is compared with an information-theoretic approach. It demonstrated robust performance with higher classification accuracy.

A Raman chemical imaging system for detection of contaminants in food 

Paper 8027-38 of Conference 8027
Date: Tuesday, 26 April 2011

Author(s): Kaunglin Chao, Jianwei Qin, Moon S. Kim, U.S.D.A. Agricultural Research Service (United States)


Raman chemical imaging technique combines Raman spectroscopy and machine vision to visualize the composition and structure of a target, and it offers great potential for food safety research. Commercially available systems generally perform Raman measurements at a microscopic level, and consequently cannot easily meet the requirements for evaluating whole surfaces of individual food items. A bench-top point-scanning Raman chemical imaging system was designed and developed in the laboratory for food safety inspection. This work demonstrates that Raman scattering information can be useful for mapping spatial distribution of constituents in complex food systems.

Generalized accelerated hyperspectral, and multiframe algorithm for nondestructive micro-electromechanical systems (MEMS) microscope metrology 

Paper 8056-35 of Conference 8056
Date: Tuesday, 26 April 2011

Author(s): Wojtek J. Walecki, Fanny Szondy, Sunrise Optical LLC (United States)


We have constructed a system employing accelerated Richardson Lucy algorithm for three dimensional mapping of the thin membranes in Micro Electro-Mechanical Systems (MEMS) pressure sensing devices. System is collecting data at several wavelengths bands. Several frames representing image of the device allow combining multi-frame spectral, and spacial information. Our algorithm uses this information together with prior information from thin film model of membranes, and Baysian model for point spread function the microscope to obtain the enhanced spacial resolution image, and the enhanced thickness maps of measured membranes.

Hyperspectral band selection using statistical models 

Paper 8048-67 of Conference 8048
Date: Tuesday, 26 April 2011

Author(s): Jochen M. Maerker, Alfons Ebert, Wolfgang Middelmann, Fraunhofer-Institut für Optronik, Systemtechnik und Bildauswertung (Germany)
No abstract available

Hyserspectral imaging for nondestructive quality and maturity evaluation in tomatoes 

Paper 8027-36 of Conference 8027
Date: Tuesday, 26 April 2011

Author(s): Sukwon Kang, National Academy of Agriculture Science (Korea, Republic of); Moon S. Kim, U.S.D.A. Agricultural Research Service (United States); Kangjin Lee, National Academy of Agriculture Science (Korea, Republic of)


The fresh-market tomatoes are one of the major vegetables in the world. Color in tomato (Lycopersicon esculentum) is one of the important external characteristic to assess ripeness and shelf-life of tomato. Usually, the degree of maturity has been estimated by human graders comparing the tomato color to a chart that classify fresh tomatoes into six maturity stages based on the USDA standard classification. This tomato maturity classification often results into errors due to human subjectivity, visual stress and tiredness. Color camera has been used to classify the tomato but it is not easy to define the six maturity stage based on color. Hyperspectral imaging system was used to find the relationship between the tomato maturity and hyperspectal reflectance images. Also, hyperspectal reflectance images were used to evaluate the quality and maturity in tomatoes.

Noise reduction of hyperspectral images by using joint bilateral filter 

Paper 8048-68 of Conference 8048
Date: Tuesday, 26 April 2011

Author(s): Ayoung Heo, Jai-Hoon Lee, Eun-Jin Choi, Won-Chul Choi, Seo Hyun Kim, Dong-Jo Park, KAIST (Korea, Republic of)


In this paper, we propose a new noise reduction method for hyperspectral images by using the joint bilateral filter. The Gaussian range kernel of the joint bilateral filter is applied to a sharp-edged image. In this proposed method, the sharp-edged image is constructed by the weighted summation of all bands of a hyperspectral image cube. Since the obtained sharp-edged image has high-frequency details, the joint bilateral filter plays a role not only to reduce noise but also to preserve the edge. We have evaluated the performance of the proposed denoising method on the hyperspectral imaging systems which we have developed for visible and near-infrared spectral regions. Simulation results show that the proposed method outperforms the conventional approaches.

Subpixel target detection and enhancement in hyperspectral images 

Paper 8048-70 of Conference 8048
Date: Tuesday, 26 April 2011

Author(s): Kailash C. Tiwari, Military Engineering Services (India)


Hyperspectral data due to its higher information content afforded by higher spectral resolution is increasingly being used for various remote sensing applications including information extraction at subpixel level. Whenever an object /class gets spectrally resolved but not spatially, mixed pixels result. Spectral unmixing models are used to recover components of a mixed pixel which output both the endmember spectrum and their corresponding abundance fractions but do not provide their spatial distribution. A new inverse Euclidean distance based super-resolution mapping method has been presented that achieves subpixel target detection by adjusting spatial distribution of abundance fraction within a pixel.

Miniaturization of a SWIR hyperspectral imager 

Paper 8020-1 of Conference 8020
Date: Wednesday, 27 April 2011

Author(s): Christopher P. Warren, William R. Pfister, Detlev M. Even, Arleen Velasco, Joseph Naungayan, Selwyn M. Yee, David S. Breitwieser, NovaSol (United States)


A new approach for the design and fabrication of a miniaturized SWIR Hyperspectral imager is described. This design uses the Offner design form, and solid optical blocks for light propagation, providing excellent, low distortion imaging. The microHSI SWIR Hyperspectral sensor is capable of operating in the 850-1700 nm wavelength range. The blazed diffraction grating was embedded in the glass blocks, and resulted in a high diffraction efficiency. This spectrometer can support slit lengths of up to 25.6 mm. The application of skin detection is discussed; and test results are shown for matched filter skin detections in the SWIR wavelength region.

Small unmanned aerial system high performance payload 

Paper 8020-2 of Conference 8020
Date: Wednesday, 27 April 2011

Author(s): Ricky J. Morgan, Ali A. Abtahi, Usha Raghuram, Frida E. Strömqvist Vetelino, Aerospace Missions Corp. (United States)


A unique, hyperspectral imaging plane "on-a-chip" developed for deployment as a High Performance Payload (HPP) on a micro or small unmanned aerial vehicle is described. HPP employs nanophotonics technologies to create a focal plane array with very high fill factor fabricated using standard integrated circuit techniques. The spectral response of each pixel can be independently tuned and controlled over the entire spectral range of the camera. While the current HPP is designed to operate in the visible, the underlying physical principles of the device are applicable and potentially implementable from the UV through the long-wave infrared.

Fast and accurate image recognition algorithms for fresh produce food safety sensing 

Paper 8027-15 of Conference 8027
Date: Wednesday, 27 April 2011

Author(s): Chun-Chieh Yang, Moon S. Kim, Kuanglin Chao, U.S.D.A. Agricultural Research Service (United States)


The research reported the development of image recognition algorithms to detect fecal pollution on fresh produce using hyperspectral line-scan images. The algorithms were developed to satisfy the requirements of fast operation and calculation as well as accurate detection and sensing performance. The algorithms could be easily installed and calibrated to manage the machine vision system. With the algorithms, the line-scan machine vision system can be applied to the real-world food processing line to ensure food safety.

Real-world noise in hyperspectral imaging systems 

Paper 8020-3 of Conference 8020
Date: Wednesday, 27 April 2011

Author(s): Richard L. Wiggins, Lovell E. Comstock, Jeffry J. Santman, Corning NetOptix (United States)


It is well known that non-uniform illumination of a spectrometer changes the measured spectra. Laboratory calibration of hyperspectral imaging systems is careful to minimize this effect by providing repeatable, uniform illumination. In actual hyperspectral measurements the real world images result in non-uniform illumination. We define the resulting variation as real-world noise and we compare real-world noise to other noise sources. Both in-flight performance and calibration transfer between instruments degrade significantly because of real-world noise.

Hyperspectral imaging technique for determination of pork freshness 

Paper 8027-16 of Conference 8027
Date: Wednesday, 27 April 2011

Author(s): Yankun Peng, Leilei Zhang, China Agricultural Univ. (China)


Freshness of pork is an important quality attribute. The specific objectives of this research were to develop a hyperspectral imaging system to predict pork freshness. Hyperspectral scattering images were collected from the pork surface at the range of 400-1100 nm. The spectral scattering profiles at individual wavelengths were fitted by a three-parameter Lorentzian distribution (LD) function; and, reflectance spectra were extracted from the scattering images. A prediction model for pork freshness was established by using a combination of TVB-N, pH and color parameters. It could give a good prediction with r = 0.90 and SEP = 5.05 for pork freshness.

Improved classification using image data fused via nonlinear dimensionality reduction 

Paper 8050-49 of Conference 8050
Date: Wednesday, 27 April 2011

Author(s): Colin C. Olson, Jonathan M. Nichols, K. Peter Judd, Frank Bucholtz, U.S. Naval Research Lab. (United States)


We present a process for fusing multiple sensor modalities that leverages nonlinear dimensionality reduction. In particular, diffusion map is used to embed high-dimensional images (or features from those images) as low-dimensional manifolds in an embedding space. Images of the same or similar scenes taken with different sensors are individually mapped into the space. Once embedded, the manifolds derived from each sensor are aligned and fused. Thus, image registration and fusion are accomplished in the same step. We present results pertaining to two sensors, one capturing visible wavelengths, the other infrared. Improved classification results are found using the fused images.

Detection of fruit fly infestation in pickling cucumbers using hyperspectral imaging 

Paper 8027-19 of Conference 8027
Date: Wednesday, 27 April 2011

Author(s): Renfu Lu, Agricultural Research Service (United States); Diwan P. Ariana, Michigan State Univ. (United States)


Fruit fly infestation is a serious problem in some pickling cucumber producing regions. Currently, processors have to rely on humans to detect and remove fruit fly-infested cucumbers. In this research, hyperspectral images in an integrated mode of reflectance (450-740 nm) and transmittance (740-1000 nm) were acquired from normal and fruit fly-infested pickling cucumbers. Mean spectra calculated for each pickling cucumber were used for classification of the cucumbers. Hyperspectral transmittance imaging mode achieved an overall classification accuracy of 87.8%, compared with 75.4% from human inspection. The research demonstrated the usefulness of hyperspectral imaging for detection of fruit fly infested pickling cucumbers.

Multiclass sub-pixel target detection using functions of multiple instances 

Paper 8048-41 of Conference 8048
Date: Wednesday, 27 April 2011

Author(s): Alina Zare, Univ. of Missouri-Columbia (United States); Paul Gader, Univ. of Florida (United States)


The Functions of Multiple Instances (FUMI) method learns target prototypes from data points that are functions of both target and non-target prototypes. In this paper, a multi-class case of FUMI is considered where, given data points which are convex combinations of a target prototype and several non-target prototypes. The Multi-class Convex-FUMI (C-FUMI) method learns the target and non-target signatures, the number of non-target signatures, and determines the proportions of the all prototypes for each data point. For this method, training data need only binary labels and the specific target proportions for the training data are not needed. In the case of hyperspectral image analysis, this provides a method for multi-class sub-pixel target detection when the spectral signatures of the target classes are unknown.
 

Dried fruits quality assessment by hyperspectral imaging 

Paper 8027-23 of Conference 8027
Date: Wednesday, 27 April 2011

Author(s): Silvia Serranti, Giuseppe Bonifazi, Univ. degli Studi di Roma La Sapienza (Italy)


Dried fruits products, such as hazelnuts and almonds, present different market values according to their quality. Such a quality is usually quantified in terms of freshness of the products, as well as presence of contaminants (pieces of shell, husk, small stones) and defects, mould and decays. Reflectance spectra of selected dried fruits of different quality and characterized by the presence of different contaminants and defects have been acquired by a laboratory device equipped with two hyperspectral imaging systems working in two different spectral ranges: visible-near infrared field (400-1000 nm) and near infrared field (1000-1700 nm). The spectra have been processed and results evaluated adopting both a simple and fast wavelength band ratio approach and a more sophisticated classification logic based on principal component (PCA) analysis.

Joint segmentation and reconstruction of hyperspectral images from a single snapshot 

Paper 8048-47 of Conference 8048
Date: Wednesday, 27 April 2011

Author(s): Peter Qiang Zhang, Robert J. Plemmons, Wake Forest Univ. (United States); David J. Brady, David Kittle, Duke Univ. (United States)


This work describes numerical methods for the joint reconstruction and segmentation of a hyperspectral image cube from a single snapshot taken by a coded aperture snapshot spectral imager (CASSI). For this highly underdetermined inverse problem, we seek a particular form of solution that assumes spectrally homogeneous segments in the two spatial dimensions, and greatly reduces the number of unknowns, often turning the underdetermined system into an overdetermined. The proposed method generalizes popular active contour segmentation algorithms such as the Chan-Vese model and also enables one to jointly segment and reconstruct the hyperspectral cube. The results are illustrated on simulated and real data.

Improved real-time processing of hyperspectral imaging data 

Paper 8017-44 of Conference 8017
Date: Wednesday, 27 April 2011

Author(s): Robert Schweitzer, Matthew P. Nelson, Robert J. D'Agostino, Patrick J. Treado, ChemImage Corp. (United States)


Sensor systems that can rapidly detect explosives at standoff distances in operationally relevant sensor configurations are achieving a state of robustness and reliability. ChemImage has developed algorithms and software strategies that are the foundation of a Real Time Toolkit (RTTK) that currently supports data from Raman, LIBS, SWIR, and RGB sensors. The RTTK takes advantage of multiple sensors, spectral and spatial information, multiple scenes allowing the use of persistence based algorithms, and the use of software techniques that take advantage of advances in multi-core computer processing. This presentation will describe several of these algorithmic advances.

Stand-off detection of explosive particles by imaging Raman spectroscopy 

Paper 8017-45 of Conference 8017
Date: Wednesday, 27 April 2011

Author(s): Markus Nordberg, Hanna Ellis, Anneli Ehlerding, Henric Oestmark, Torgny Carlsson, Swedish Defence Research Agency (Sweden)


Explosive particles from a fingerprint have been detected and identified at stand-off distanced using multispectral imaging Raman scattering. Fingerprints containing particles of DNT, TNT and ammonium nitrate were placed on a brick at a distance of 12 m, and image sequences measured at different Raman shift were recorded. The images sequence was processed for each pixel and the spectral data where compared with reference spectra. By using false color coding the pixels were marked with different colors corresponding to the detected substances in the fingerprint.

A new deblurring morphological filter for hyperspectral images 

Paper 8048-51 of Conference 8048
Date: Wednesday, 27 April 2011

Author(s): Ezz E. Ali, Military Technical College (Egypt)


In this paper, we introduce a new method to deblurr the hyperspectral images keeping edges as sharp as possible. Motivated by the success of threshold decomposition, gradient-based operators are used to detect the locations of edges followed by an adaptive morphological filter to sharpen these detected edges. Experimental results demonstrated that the performance of the proposed deblurring filter is promising for hyperspectral images in target detection applications.

Fusion of hyperspectral and ladar data for autonomous target detection 

Paper 8064-7 of Conference 8064
Date: Wednesday, 27 April 2011

Author(s): Andrey V. Kanaev, Thomas J. Walls, U.S. Naval Research Lab. (United States)


Robust fusion of data from disparate sensor modalities can provide improved target detection performance over those attainable with the individual sensors. We have developed a novel fusion algorithm enabling detection of difficult targets when the HSI data is simultaneously collected with ladar data. As a part of fusion processing we have also developed an algorithm for automatic co-registration of ladar and HSI imagery, based on the maximization of mutual information, which can provide accurate, sub-pixel registration even in the case when the imaging geometries for the two sensors differ.

Implications of model mismatch and covariance contamination on chemical detection algorithms 

Paper 8048-54 of Conference 8048
Date: Wednesday, 27 April 2011

Author(s): Dimitris Manolakis, Steven E. Golowich, MIT Lincoln Lab. (United States); Sidi Niu, Vinay K. Ingle, Northeastern Univ. (United States)


In this paper we investigate the impact of these factors on the performance of chemical plume detection algorithms. The analytical investigations are limited to the classical matched filter detector. However, using a plume-embedding procedure to embed plumes into real backgrounds, we can study the performance of the matched filter and various other detectors (for example, the widely used adaptive cosine estimator) by estimating their receiver operating characteristic (ROC) curves. Preliminary theoretical and experimental results show that a limited amount of background data, spectral heterogeneity, and background corruption by plume may lead to significant performance degradation. Therefore, understanding the impact of these issues and developing robust practical algorithms for their minimization or avoidance is critical to the successful deployment of systems that protect the warfighter.

Performance limits of LWIR gaseous plume quantification 

Paper 8048-55 of Conference 8048
Date: Wednesday, 27 April 2011

Author(s): Steven E. Golowich, Dimitris Manolakis, MIT Lincoln Lab. (United States)


The central parameter in the quantification of chemical vapor plumes via remote sensing is the mean concentration-path length (CL) product, which can lead to estimates of the absolute gas quantity present. The goal of this paper is to derive Cramer-Rao lower bounds on the variance of an unbiased estimator of CL in concert with other parameters of a general non-linear radiance model. These bounds offer a guide to feasibility of CL estimation that is not dependent on any given algorithm. In addition, the derivation of the bounds yields great insight into the physical and phenomenological mechanisms that control plume quantification.

Remote quantification of smokestack total effluent mass flow rates using imaging Fourier-transform spectroscopy 

Paper 8018-39 of Conference 8018
Date: Wednesday, 27 April 2011

Author(s): Jacob L. Harley, Kevin C. Gross, Air Force Institute of Technology (United States)


An infrared (1.5-5.5 µm) imaging Fourier-transform spectrometer (IFTS) was used to estimate industrial smokestack total effluent mass flow rates (kg/hr) by combining spectrally-determined species concentrations with flow rates estimated via analysis of sequential images in the raw interferogram cube. At a stand-off distance of 350 m, 200 hyperspectral images were collected on a 128 x 64 pixel sub-window (11.4 x 11.4 cm^2 per pixel) at high spectral resolution (0.5/cm). Strong emissions from H2O, CO2, CO, SO2, and NO were observed in the spectrum, and concentrations will be retrieved and compared with in situ measurements. The turbulent nature of the flow field results in instantaneous fluctuations in scene radiance; these fluctuations lead to brightness patterns which are captured in the DC-level imagery. A simple analysis of sequential imagery will be presented which enables an estimation of the flow velocity.

Multi- and hyperspectral scene modeling 

Paper 8048-56 of Conference 8048
Date: Thursday, 28 April 2011

Author(s): Christoph C. Borel, Ronald F. Tuttle, Air Force Institute of Technology (United States)


Often it is prohibitive or even impossible to obtain hyper-spectral data over real targets with existing sensors and under a number of conditions. In this paper we describe how a public domain raytracer with its own scene description language (POVRAY) can be used to model multi- and hyper-spectral scenes in the visible and also thermal. The advantage of using POVRAY is that the scene can be rendered using various rendering options from simple Gouraud type shading, single bounce raytracing, multiple bounce raytracing, radiosity and photon-mapping.

An empirical estimate of the multivariate normality of spectral image data 

Paper 8048-59 of Conference 8048
Date: Thursday, 28 April 2011

Author(s): Ariel Schlamm, David W. Messinger, Rochester Institute of Technology (United States)


Historically, much of spectral image analysis revolves around assumptions of multivariate normality. If the background spectral distribution can be assumed to be multivariate normal, then algorithms for anomaly detection, target detection, and classification can be developed around that assumption. However, as the current generation of sensors typically have higher spatial and/or spectral resolution, the spectral distribution complexity of the data collected is increasing and these assumptions are no longer adequate, particularly image-wide. A new empirical method for accessing the multivariate normality of a hyperspectral distribution is presented here.

Next generation signature-based hyperspectral detection: a challenge to atmospheric modelers 

Paper 8040-11 of Conference 8040
Date: Thursday, 28 April 2011

Author(s): Alan P. Schaum, Brian J. Daniel, U.S. Naval Research Lab. (United States)


A new class of hyperspectral algorithms has been developed for detection based on a re-flectance signature. These promise performance levels superior to state-of-the-art meth-ods employed in real systems, by creating selective decision surfaces that can be sculpted to mitigate the usual plague of ubiquitous outliers. The new class of detectors is based on an affine target subspace model and a continuum fusion interpretation of the generalized likelihood ratio test. The challenge to atmospheric modelers is to create a method for pre-dicting, from a given reflectance spectrum, a low-dimensional radiance subspace lying closer to the sensed target spectrum than the target is to the whitened clutter mean.

Interactive visualization of hyperspectral images on a hyperbolic disk 

Paper 8048-60 of Conference 8048
Date: Thursday, 28 April 2011

Author(s): Adam A. Goodenough, Ariel Schlamm, Rochester Institute of Technology (United States)


We look at developing an interactive, intuitive hyperspectral visualization and analysis tool based on using a Poincare disk as a window into a high dimensional spectral space. The Poincare disk represents an infinite, two-dimensional hyperbolic space such that distances and areas increase exponentially as you move farther from the center of the disk. By projecting N-dimensional data into this space using a non-linear, yet relative distance metric preserving projection (such as the Sammon projection), we can simultaneously view the entire data set while maintaining natural clustering and spacing. The disk also provides a means of interacting with the data for classification, analysis and instruction.

Adaptive hyperspectral sensing with carbon nanotubes 

Paper 8058-26 of Conference 8058
Date: Thursday, 28 April 2011

Author(s): Harold Szu, U.S. Army Night Vision & Electronic Sensors Directorate (United States); Yin-Lin Shen, Kenneth H. Ou, The George Washington Univ. (United States); Reinhardt Kit, Air Force Office of Scientific Research (United States)


Adaptive sensing is possible to achieve a compressive sensing, when we reverse the direction of Einstein photo-electric effect of Nano Solar Cells for imaging. Each pixel will be designed as a fireman staircase, of which each run is made of Carbon Nanotubes (CNT) at a different diameter. Saito-Wallace bandgap formula may be understood as de Broglie matter wave around the circumference. Thus, the band gap may be re-derived as follows: ε_BG=C_Fermi P=C_Fermi h/λ=C_graphene h/πd, and λ=2πR=πd of the CNT diameter , where use is made of Geim and Novoselov result (2010 Nobel Laureates) that single wall CNT enjoys a ballistic propagation C_Fermi identically to one thousandth of the speed of light in the single sheet grapheme C_graphene=〖10〗^(-3) C_o. We control the grid field effect of CNT to turn current signal on or off. We evaluate the dark current, the polarization, the quantum efficiency and the SNR

Metrics for the selection of frequency bands from hyperspectral data for image fusion and sensor development 

Paper 8064-15 of Conference 8064
Date: Thursday, 28 April 2011

Author(s): Jack E. Fulton, Jr., Naval Surface Warfare Ctr. Crane Div. (United States)


The application of imagers in security is to provide a clear warning of potential threats to the end users. Hyperspectral imagers (HSI) are not used in security applications due to the high cost and the need for extensive processing. A proposed set of objective and subjective metrics along with fusion techniques for specific applications is presented. The selection criteria create a basis set of frequencies to be used in a fieldable, threat specific, affordable imager.

Hyperspectral antireflective coatings for infrared windows 

Paper 8016-26 of Conference 8016
Date: Thursday, 28 April 2011

Author(s): Donald E. Patterson, Byron G. Zollars, Steve M. Savoy, Nanohmics (United States)


Using conical "moth-eye" structures, a hyperspectral antireflective coating is being developed for use with ZnS (Cleartran) infrared windows. In this work, we are using the emerging technique of imprint lithography to create moth‐eye structures on the surface of Cleartran windows with transverse scales from 200‐300 nm and with aspect ratios >10. The surface features, in conjunction with a conformal protective coating of amorphous AlN, can serve as anti‐reflection surface treatments spanning the wavelength range from the visible through the long‐wave infrared. Cleartran windows with imprinted moth‐eye structures can potentially be used in numerous aerospace applications.

High-spatial resolution hyperspectral spatially adaptive endmember selection and spectral unmixing 

Paper 8048-64 of Conference 8048
Date: Thursday, 28 April 2011

Author(s): Kelly Canham, Ariel Schlamm, William F. Basener, David W. Messinger, Rochester Institute of Technology (United States)


Spectral unmixing results in hyperspectral imagery are dependent on the number of estimated endmembers. Previous statistical and geometric approaches have been developed to estimate the number of endmembers using the global dataset, which do not take into consideration local area endmember variability. Here, the number of endmembers is estimated by using a spatially adaptive approach. Each pixel is unmixed using locally identified endmembers, and global abundance maps are generated by classifying the locally derived endmembers. Comparisons are made to established unmixing methodologies using multiple high-spatial resolution hyperspectral datasets and the residual unmixing error.

Spectral variations in HSI signatures of thin fabrics for detecting and tracking of dismounts 

Paper 8040-15 of Conference 8040
Date: Thursday, 28 April 2011

Author(s): Jared Herweg, Rochester Institute of Technology (United States) and Air Force Institute of Technology (United States); John P. Kerekes, Emmett Ientilucci, Rochester Institute of Technology (United States); Michael T. Eismann, Air Force Research Lab. (United States)


This work extends the understanding of the induced spectral variation in dismount spectral signatures in cluttered environments. The goal of this work was to isolate the spectral reflectivity of highly transmissive targets independent of the background. Using a linear mixing model, the effects of reflective backing materials on the signature of a thin fabric are presented. Also, an issue with tracking a pedestrian from full illumination into the shadow is considered. Reflectance factor signatures were measured using target reflectivity measured both in the lab and in the field to assess spectral variability and detectability.

Kernel-based weighted abundance constrained linear spectral mixture analysis 

Paper 8048-65 of Conference 8048
Date: Thursday, 28 April 2011

Author(s): Keng-Hao Liu, Englin Wong, Univ. of Maryland, Baltimore County (United States); Chein-I Chang, Univ. of Maryland, Baltimore County (United States) and National Chung Hsing Univ. (Taiwan)


This paper presents a Kernel-based Weighted Abundance Constrained-LSMA (KWAC-LSMA) which includes Least Squares-based Linear Spectral Mixture Analysis (LS-LSMA), Fisher's LSMA (FLSMA), Weighted Abundance Constrained-LSMA (WAC-LSMA) and Kernel-based LSMA as its special cases. In order to demonstrate utility of the KWAC-LSMA multispectral and hyperspectral experiments are conducted for performance analysis.

MRi dual-band MWIR imaging FTS 

Paper 8014-35 of Conference 8014
Date: Thursday, 28 April 2011

Author(s): Louis M. Moreau, Claude B. Roy, Stéphane Lantagne, Florent Prel, Christian A. Vallieres, ABB Analytical Measurement (Canada)


MRi is an imaging version of the ABB Bomem MR Fourier-Transform spectroradiometer. This field instrument generates spectral datacubes in the MWIR and LWIR. It is designed to be sufficiently fast to acquire the spectral signatures of rapid events. Overview of the instrument capabilities will be presented. Test results and results from field trials for a configuration with two MWIR cameras will be presented. That specific system is dedicated to the characterization of airborne targets. The two MWIR cameras are used to expand the dynamic range and simultaneously measure the spectral signature of the coldest and warmest elements of the scene.

Crude oil and refined petroleum product detection on terrestrial substrates with airborne imaging spectroscopy 

Paper 8040-20 of Conference 8040
Date: Thursday, 28 April 2011

Author(s): C. Scott Allen, George Mason Univ. (United States); Mark P. S. Krekeler, Miami Univ. (United States)


One of the most prominent portions of oil spill response is mapping spill extent. Yet, the most common method of detecting oil in a crisis remains visual spotting. Employing spectral libraries for material identification, imaging spectroscopy supplements traditional techniques by providing more accurate petroleum detection and discrimination from water on terrestrial backgrounds. This effort applies a new hydrocarbon-substrate spectral library to airborne imaging spectroscopy data from the Hurricane Katrina disaster in 2005. Future efforts anticipate applying the same methods to data from the Deepwater Horizon incident.

Formatting research and development sensors for data interoperability and fusion with GIS 

Paper 8053-10 of Conference 8053
Date: Thursday, 28 April 2011

Author(s): Karmon M. Vongsy, Air Force Institute of Technology (United States); Eric Cincotta, ITT Corp. Geospatial Systems (United States); Tom Jones, ITT Visual Information Solutions (United States)
No abstract available

Investigation of the potential use of hyperspectral imaging for stand-off detection of person-borne IEDs 

Paper 8017-69 of Conference 8017
Date: Thursday, 28 April 2011

Author(s): Catherine C. Cooksey, David W. Allen, National Institute of Standards and Technology (United States)


Advances in hyperspectral sensors and algorithms in numerous fields of research have opened up new possibilities and may also improve the detection of person-borne IEDs. While portions of the electromagnetic spectrum, such as the x-ray and terahertz regions, have been investigated for this application, the spectral region of the ultraviolet (UV) through shortwave infrared (SWIR) (250 nm to 2500 nm) has received little attention. The purpose of this work was to investigate what, if any, potential there may be for exploiting the spectral region of the UV through SWIR for the detection of hidden objects under the clothing of individuals. The optical properties of both common fabrics and threat objects were measured. The approach, measurement methods, and results are described in this paper, and the potential for hyperspectral imaging is addressed.

A novel infrared hyperspectral imager for passive standoff detection of explosives and explosive precursors 

Paper 8018-59 of Conference 8018
Date: Thursday, 28 April 2011

Author(s): Jean-Marc Theriault, Eldon Puckrin, Hugo Lavoie, Francois Bouffard, Defence Research and Development Canada (Canada); Paul Lacasse, AEREX avionique inc. (Canada); Alexandre Vallières, Vincent Farley, Martin Chamberland, Telops (Canada)


The passive standoff detection of vapors from particular explosives and precursors emanating from a location under surveillance can provide early detection and warning of illicit explosives fabrication. DRDC Valcartier recently initiated the development and field-validation of a novel R&D prototype, MoDDIFS (Multi-Option Differential and Imaging Fourier Spectrometer) to address and solve this security vulnerability. The proposed methodology combines the clutter suppression efficiency of the differential detection approach with the high spatial resolution provided by the hyperspectral imaging approach. This consists of integrating an imaging capability of the Hyper-Cam advanced IR imager developed by Telops with a differential CATSI-type sensor. This paper presents the MoDDIFS sensor methodology and first investigation results that were recently obtained.

Kernel and stochastic expectation maximization fusion for target detection in hyperspectral imagery 

Paper 8055-25 of Conference 8055
Date: Thursday, 28 April 2011

Author(s): Mohamed I. Elbakary, Mohammad S. Alam, Univ. of South Alabama (United States)


In this paper, we present a new algorithm for target detection using hyperspectral imagery. The proposed algorithm is inspired by the outstanding performance of nonlinear RX-algorithm and the robustness of the stochastic expectation maximization (SEM) algorithm. The traditional technique of using SEM algorithm for target detection in hyperspectral imagery is associated with dimensionality reduction of the input data using binning or principal components analysis (PCA) algorithm. To facilitate detection of the target by using the entire targets information and simultaneously reducing the computational burden on SEM algorithm, we propose a new scheme for data reduction based on using Kernels. The proposed scheme for fusion the kernel with SEM algorithm has been tested using real life hyperspectral imagery and the results show superior performance compared to alternate algorithms.
 

Multi-field-of-view hyperspectral imager 

Paper 8020-39 of Conference 8020
Date: Thursday, 28 April 2011

Author(s): Lovell E. Comstock, Richard L. Wiggins, Corning NetOptix (United States)


There is increasing interest in imaging spectrometers working in the SWIR and LWIR wavelength bands. Commercially available detectors are not only expensive, but have a limited number of pixels, compared with visible band detectors. Typical push broom hyperspectral imaging systems consist of a fore optic imager, a slit, a line spectrometer, and a two dimensional focal plane with a spatial and spectral direction. To improve the spatial field coverage at a particular resolution, multiple systems are incorporated, where the "linear fields of view" of the systems are aligned end to end. This solution is prohibitive for many applications due to the costs of the multiple detectors, coolers, spectrometers, or the space, weight, or power constraints. Corning will present a cost effective solution utilizing existing detectors combined with innovative design and manufacturing techniques.

QUEST hierarchy for hyperspectral face recognition 

Paper 8029B-60 of Conference 8029B
Date: Monday, 25 April 2011

Author(s): David Ryer, U.S. Air Force (United States); Trevor J. Bihl, Kenneth W. Bauer, Air Force Institute of Technology (United States); Steven K. Rogers, Air Force Research Lab. (United States)


A face recognition methodology employing an efficient fusion hierarchy for hyperspectral imagery (HSI) is presented. A Matlab-based graphical user interface (GUI) has been developed to aid processing and to display results. Adaptive feedback loops are incorporated to improve performance thru the reduction of candidate subjects in the gallery as well as the injection of additional probe image samples. Algorithmic results and performance improvements are presented as spatial, spectral, and temporal effects are considered in this Qualia Exploitation of Sensor Technology (QUEST) motivated methodology.

Selecting training and test images for optimized anomaly detection and material identification algorithms in hyperspectral imagery through robust parameter design 

Paper 8048-12 of Conference 8048
Date: Monday, 25 April 2011

Author(s): Frank M. Mindrup, Trevor J. Bihl, Kenneth W. Bauer, Air Force Institute of Technology (United States)


There are numerous anomaly detection and material identification algorithms proposed for hyperspectral imagery. Robust parameter design (RPD) techniques have been applied to some of these algorithms in an attempt to choose robust settings capable of operating consistently across a large variety of image scenes. Previous research developed a framework for optimizing anomaly detection in HSI by considering specific image characteristics as noise variables. Typically, the characteristics available in sets of images do not provide orthogonal noise designs assumed in RPD. This paper describes a method for selecting hyperspectral image training and test subsets yielding consistent RPD results.

GPGPU-based real-time conditional dilation for robust target detection in multispectral and hyperspectral imagery 

Paper 8048-71 of Conference 8048
Date: Tuesday, 26 April 2011

Author(s): James P. Morgenstern, Vision4ce LLC (United States)


A significant topic in many image processing systems is the derivation of a threshold to enable the detection of targets, the detection of classes of objects which are different than the background clutter or the automated analysis of the output of spectral filters and/or anomaly filters. In many cases the background signals are uni-modal and the estimation of a robust threshold is a straightforward problem with known solution. There are some cases where the signals of interest have local contrast against their immediate surroundings but the application of a global threshold over the entire image produces poor results. In such cases an adaptive or local threshold operator offers a more robust solution. One particular local threshold function is the conditional dilation [originally due to Serra] which produces a second image by a series of dilations but conditioned on not exceeding the signal levels in the original. In the limit this second image becomes a threshold surface where only locally contrasty areas or objects remain after application of the threshold. Algorithms have been introduced which enable use of conditional dilation in realtime systems by reducing the unbounded series of dilations to a small, fixed number of operations. In the present work we present an adaptation of this algorithm to a GPGPU device which enables highly parallel version of the algorithm subject to the unique architecture constraints of the GPGPU.

Anomaly detection in hyperspectral imagery using stable distribution 

Paper 8049-31 of Conference 8049
Date: Tuesday, 26 April 2011

Author(s): Suat Mercan, Univ. of Nevada, Reno (United States); Mohammad S. Alam, Univ. of South Alabama (United States)


In hyperspectral imaging applications, the background generally exhibits a clearly non-Gaussian impulsive behavior, where valuable information stays in the tail. In this work, we propose a new technique, where the background is modeled using the stable distribution for robust detection of outliers. The outliers of the distribution can be considered as potential anomalies or regions of interests (ROIs). We effectively utilize the stable model for detecting targets in impulsive hyperspectral data. To decrease the false alarm rate, it is necessary to compare the ROI with the known reference using a suitable technique, such as the Euclidian distance. This representation compensates a drawback of the Gaussian model, which is not well suited for describing signals with impulsive behavior. In addition, thresholding is considered to avoid misclassification of targets. Test results using real life hyperspectral image datasets are presented to verify the effectiveness of the proposed technique.

Course: Multispectral and Hyperspectral Image Sensors

Date: Wednesday, 27 April 2011

Instructor(s): Terrence S. Lomheim, The Aerospace Corp. (United States)


This course will describe the imaging capabilities and applications of the principal types of multispectral (MS) and hyperspectral (HS) sensors. The focus will be on sensors that work in the visible, near-infrared and shortwave-infrared spectral regimes, but the course will touch on longwave-infrared applications. A summary of the salient features of classical color imaging (human observation) will also be provided in an appendix.

Object classification using discriminating features derived from higher-order spectra of multi- and hyperspectral imagery 

Paper 8048-37 of Conference 8048
Date: Wednesday, 27 April 2011

Author(s): Karen N. Zachery, Jiangying Zhou, Yuwei Liao, Teledyne Scientific & Imaging, LLC (United States)


This paper describes a novel approach for the detection and classification of man-made objects using discriminating features derived from higher-order spectra (HOS) of multi- and hyperspectral signals. Our proposed algorithm exploits the fact that HOS is insensitive to symmetrically distributed noise (e.g., Gaussian, uniform); exhibits the capability of detecting and characterizing nonlinear structures in spectral signature and is invariant to translation, rotation, and scaling. By exploiting these HOS properties, we have devised a robust method for classifying man-made objects that are affected by different noise distributions and the presence of spectrally similar signatures (confusers) as well as variable signal-to-noise ratios.

Peach maturity/quality assessment using hyperspectral imaging-based spatially resolved technique 

Paper 8027-20 of Conference 8027
Date: Wednesday, 27 April 2011

Author(s): Haiyan Cen, Renfu Lu, Fernando A. Mendoza, Diwan P. Ariana, Michigan State Univ. (United States)


In order to develop an effective optical system for maturity/quality assessment of peaches, it is important to understand their optical absorption and scattering properties as related to the physiological states. The objective of this research was to measure the absorption and scattering properties of peaches for their maturity and quality assessment. A optical property measuring instrument was used in this study. Five hundred peaches, harvested at four different dates in 2010, were used in the experiment. Measurements for the optical properties and maturity/quality indices were performed on the same day of harvest. Spatially-resolved hyperspectral images were first acquired from each sample followed by the reference measurements. An inverse algorithm was used to extract the spectra of absorption and reduced scattering coefficients of peaches at 500-1,000 nm. Predictive and classification models relating the measured optical properties to maturity/quality indices were established.

Hyperspectral anomaly detection using sparse kernel-based ensemble learning 

Paper 8048-52 of Conference 8048
Date: Wednesday, 27 April 2011

Author(s): Prudhvi Gurram, Heesung Kwon, U.S. Army Research Lab. (United States)


In this paper, the principle of Sparse Kernel-based Ensemble Learning (SKEL) is extended to hyperspectral anomaly detection to obtain Sparse Kernel-based Anomaly Detection (SKAD). In SKAD, a one class classifier based on support vector data description (SVDD) is used as a sub-classifier. Each sub-classifier first finds the most compact enclosing hypersphere of the local background spectra within the corresponding randomly selected spectral subspace. Optimal sparse weighting of the kernels that minimizes the volume of the enclosing ball of the combined kernel is then obtained by optimizing the kernel weights under an L-1 constraint. The optimal hypersphere defines the support of the local normalcy data and the pixels with spectral signatures outside the hypersphere are considered outliers/targets.

Effect of random measurements on the performance of classical hyperspectral target detection algorithms 

Paper 8048-53 of Conference 8048
Date: Wednesday, 27 April 2011

Author(s): Yi Chen, The Johns Hopkins Univ. (United States); Nasser M. Nasrabadi, U.S. Army Research Lab. (United States); Trac D. Tran, The Johns Hopkins Univ. (United States)


In this paper, we study the effect of random measurements of spectral pixels on the performance of hyperspectral imagery (HSI) target detection. The N-dimensional spectral pixels are projected onto an M-dimensional measurement space, where M is much smaller than N, using some measurement matrix whose entries are usually i.i.d. Gaussian or Bernoulli random variables. The classical target detector algorithms are then directly applied to the M-dimensional measurement vectors to detect the targets of interests. Through extensive experiments on several real HSI, we demonstrate the minimal compression ratio M/N under various types of random projections that are necessary to achieve detection performance comparable to that obtained by exploiting the original N-dimensional pixels.

Course: Target Detection Algorithms for Hyperspectral Imagery

Date: Thursday, 28 April 2011

Instructor(s): Nasser M. Nasrabadi, U.S. Army Research Lab. (United States)


This course provides a broad introduction to the basic concept of automatic target and object detection and its applications in Hyperspectral Imagery (HSI). The primary goal of this course is to introduce the well known target detection algorithms in hyperspectral imagery. Examples of the classical target detection techniques such as spectral matched filter, subspace matched filter, adaptive matched filter, orthogonal subspace, support vector machine (SVM) and machine learning are reviewed. Construction of invariance subspaces for target and background as well as the use of regularization techniques are presented. Standard atmospheric correction and compensation techniques are reviewed. Anomaly detection techniques for HSI and dual band FLIR imagery are also discussed. Applications of HSI for detection of mines, targets, humans, chemical plumes and anomalies are reviewed.

The target implant method for predicting target difficulty and detector performance in hyperspectral imagery 

Paper 8048-57 of Conference 8048
Date: Thursday, 28 April 2011

Author(s): William F. Basener, John P. Kerekes, Rochester Institute of Technology (United States); C. Eric Nance, Raytheon Intelligence & Information Systems (United States)


In this paper we apply a method of inserting target spectra in real hyperspectral images for the purpose of determining top performing algorithms for a given image and target, and the relative difficulty for detection of targets in a given image with a given detector. Comparisons of predictions from this method to detection performance on real target pixels showed that the target implant method provides accurate relative predictions in terms of both target difficulty and detector performance, but reliably predicting the actual number of false alarms for a given target at a given fill fraction is difficult or impossible.

Dynamic dimensionality reduction for hyperspectral imagery 

Paper 8048-58 of Conference 8048
Date: Thursday, 28 April 2011

Author(s): Haleh Safavi, Keng-Hao Liu, Chein-I Chang, Univ. of Maryland, Baltimore County (United States)


This paper introduces a new concept of dynamic dimensionality reduction (DDR) which considers the dimensionality to be retained, p as a parameter so that it can adapt its value to meet various applications. It is quite different from the commonly used DR, referred to as static dimensionality reduction (SDR) with the p fixed at a constant value regardless of applications. In order to materialize the DDR another new concept, referred to as progressive DR (PDR) is also developed so that the DR can be performed progressively with dimensionality varying the value of p.

Simultaneous sparse recovery for unsupervised hyperspectral unmixing 

Paper 8048-62 of Conference 8048
Date: Thursday, 28 April 2011

Author(s): Dzung T. Nguyen, Yi Chen, Timothy S. Han, Trac D. Tran, The Johns Hopkins Univ. (United States)


Unsupervised Endmember Extraction and Unmixing in Hyperspectral Images (HSI) is often done using iterative algorithms which use a greedy suboptimal approach of collecting one endmember at a time. We propose a method which does the extraction and unmixing problem concurrently by solving a simultaneous sparse recovery problem. This approach is able to give a global optimum solution while requiring no prior knowledge of the representing signatures or the intrinsic dimension of the HSI. Our proposed algorithm uses the l1-l2 norm to promote simultaneous sparsity of abundance vectors while imposing important non-negativity and sum-to-one constraints. Preliminary results are competitive with other methods in terms of correctness of extracted endmembers and abundances.

Joint sparsity for target detection 

Paper 8048-63 of Conference 8048
Date: Thursday, 28 April 2011

Author(s): Yi Chen, The Johns Hopkins Univ. (United States); Nasser M. Nasrabadi, U.S. Army Research Lab. (United States); Trac D. Tran, The Johns Hopkins Univ. (United States)


In this paper, we propose a joint sparsity model for target detection in hyperspectral imagery. Hyperspectral pixels within a small neighborhood in the test image are simultaneously represented by a linear combination of a few common training samples, but weighted with a different set of coefficients for each pixel. The joint sparsity model automatically incorporates the inter-pixel correlation within the hyperspectral imagery by assuming that neighboring pixels usually consists of similar materials. The sparse representations of the neighboring pixels are obtained by simultaneously decomposing the pixels over a given dictionary consisting of background and target training samples. The recovered sparse coefficient vectors are then directly used for determining the label of the test pixels. Simulation results on several real hyperspectral images show that the proposed algorithm outperforms the classical target detection algorithms.

An adaptive algorithm for subpixel target detection using the spectral information divergence measure 

Paper 8049-14 of Conference 8049
Date: Monday, 25 April 2011

Author(s): Wesam A. Sakla, U.S. Dept. of Defense (United States); Adel A. Sakla, Univ. of South Alabama (United States)
No abstract available

Hyperspectral and multispectral above-water radiometric measurements to monitor satellite data quality over coastal area 

Paper 8030-1 of Conference 8030
Date: Tuesday, 26 April 2011

Author(s): Samir Ahmed, The City College of New York (United States); Robert Arnone, U.S. Naval Research Lab. (United States); Curtiss O. Davis, Oregon State Univ. (United States); Alex Gilerson, Tristan Harmel, Soe Min Hlaing, Alberto Tonizzo, The City College of New York (United States); Alan Weidemann, U.S. Naval Research Lab. (United States)


The Long Island Sound Coastal Observational platform (LISCO) near Northport, New York, has been recently established to support satellite data validation. LISCO has both multispectral SeaPRISM and hyperspectral HyperSAS radiometers for ocean color measurements. LISCO offers the potential for improving the calibration and validation activities of current and future Ocean Color satellite missions, as well as for satellite intercomparisons and spectral characterization of coastal waters. Results of measurements made by both the multi- and hyper-spectral instruments, in operation since October 2009, are presented, evaluated and compared with MODIS and MERIS ocean color satellite data and with hyperspectral imagery provided by the HICO satellite mission.

Chemical agent detection with low-resolution scanning FTIR sensors 

Paper 8018-41 of Conference 8018
Date: Wednesday, 27 April 2011

Author(s): Eric R. Larrieux, Dimitris Manolakis, MIT Lincoln Lab. (United States); Francis M. D'Amico, U.S. Army Edgewood Chemical Biological Ctr. (United States)


Typical standoff sensors for chemical warfare agent detection utilize passive imaging spectroscopy in the long wave infrared (LWIR) atmospheric window (8 - 13um). Low-resolution scanning spectrometers provide a small number of spectra by sampling the area surrounding a chemical plume. The limited amount of background training data and their spatial-temporal nonstationarity pose a unique challenge to the development of algorithms that exploit these data. The purpose of this paper is to analyze data from the JSLSCAD and low-resolution Aerospace scanning FTIR sensors to investigate the effects of limited background training data, background nonstationarity, and registration on the performance of chemical detection algorithms.

Characterization of turbulence in smokestack plumes via imaging Fourier-transform spectroscopy 

Paper 8048-10 of Conference 8048
Date: Monday, 25 April 2011

Author(s): Jennifer L. Massman, Kevin C. Gross, Air Force Institute of Technology (United States)


An imaging Fourier transform spectrometer was used to collect hyperspectral imagery of a coal-burning smokestack in the midwave infrared (1.5-5.5 µm). The instrument was positioned approximately 350 meters from the stack exit, giving each pixel a field of view (FOV) of approximately 11.4 cm of the plume. The instrument collected hyperspectral images on a 128 x 128 pixel sub-window at a spectral resolution of 20/cm. Approximately 5000 data cubes were collected in 30 minutes. When acquiring interferograms of a turbulent source, however, rapid fluctuations in radiance due to sudden temperature changes in the plume introduce scene change artifacts (SCA) and corrupt the spectra. Sorting an ensemble of interferograms (AC+DC) into quantiles prior to Fourier transformation minimizes SCAs. This method enables unbiased spectral retrievals of concentrations and temperature and reveals information about the temperature distribution.

Influence of aerosol estimation on coastal water products retrieved from HICO images 

Paper 8030-4 of Conference 8030
Date: Tuesday, 26 April 2011

Author(s): Karen W. Patterson, Gia M. Lamela, U.S. Naval Research Lab. (United States)


The Naval Research Laboratory has been developing the Coastal Water Signatures Toolkit (CWST) to estimate water column constituents, depth and bottom type from hyperspectral imagery using a look-up table approach. To succeed, the remote sensing reflectances (RRS) must be accurate which means the atmospheric correction must be accurate. Varying the user determined aerosol thickness in the Correction of Coastal Ocean Atmospheres software results in magnitude changes to the RRS and thus, CWST retrievals. This is an illustration of CWST retrieval variability from Hyperspectral Imager for the Coastal Ocean images due to inaccurate aerosol estimation during atmospheric correction.

Evaluating carotenoid changes in tomatoes during postharvest ripening using Raman chemical imaging 

Paper 8027-2 of Conference 8027
Date: Tuesday, 26 April 2011

Author(s): Jianwei Qin, Kuanglin Chao, Moon S. Kim, U.S.D.A. Agricultural Research Service (United States)


Evaluating carotenoid content in tomatoes can be used for monitoring their ripeness. This research was aimed to assess carotenoid changes in tomatoes during postharvest ripening using Raman chemical imaging technique. A benchtop point-scanning Raman chemical imaging system was developed to acquire hyperspectral images from tomatoes at different ripeness stages. Raman spectra of pure carotenoid standards were measured as references. A hyperspectral image classification method was developed to identify the carotenoids on the cross sections of the tomato fruits. Raman chemical images were created to visualize quantity and spatial distribution of the carotenoids at different ripeness stages of the tomatoes.

Course: Introduction to Optical and Infrared Sensor Systems

Date: Friday, 29 April 2011

Instructor(s): Joseph A. Shaw, Montana State Univ.-Bozeman (United States)


This course provides a broad introduction to optical (near UV-visible) and infrared sensor systems, with an emphasis on systems used in defense and security. Topics include both passive imagers and active laser radars (lidar/ladar). We begin with a discussion of radiometry and radiometric calculations to determine how much optical power is captured by a sensor system. We survey atmospheric propagation and phenomenology (absorption, emission, scattering, and turbulence) and explore how these issues affect sensor systems. Finally, we perform signal calculations that consider the source, the atmosphere, and the optical system and detector, to arrive at a signal-to-noise ratio for typical passive and active sensor systems. These principles of optical radiometry, atmospheric propagation, and optical detection are combined in examples of real sensors studied at the block-diagram level. Sensor system examples include passive infrared imagers, polarization imagers, and hyperspectral imaging spectrometers, and active laser radars (lidars or ladars) for sensing distributed or hard targets. The course organization is approximately one third on the radiometric analysis of sensor systems, one third on atmospheric phenomenology and detector parameters, and one third on example calculations and examination of sensor systems at the block-diagram level.

Sofradir latest developments for infrared space detectors 

Paper 8012-1 of Conference 8012
Date: Monday, 25 April 2011

Author(s): Philippe Chorier, Patricia Pidancier, Yoanna-Reine Nowicki-Bringuier, Anne Delannoy, Bruno Fieque, SOFRADIR (France)


Sofradir is one of the leading companies that develop and produce infrared detectors. Space applications have become a significant activity. In this paper, we present a review of latest Sofradir's development for infrared space applications. A presentation of Sofradir infrared detectors answering hyperspectral needs from visible up to VLWIR waveband will be made. In addition a particular emphasis will be placed on the different programs currently running, with a presentation of the associated results as they relate to performances and qualifications for space use.

Issues in algorithm fusion 

Paper 8048-2 of Conference 8048
Date: Monday, 25 April 2011

Author(s): Alan P. Schaum, U.S. Naval Research Lab. (United States)


We require that a new theory of detection for composite hypothesis problems meet several requirements: It should: (1) be invariant to an arbitrary transformation of coordinates; (2) produce optimal algorithms for problems admitting uniformly most powerful solutions; (3) be superior to prior methods in at least some cases. The new theory of continuum fusion (CF), which was developed for hyperspectral detection applications, is examined in the light of these requirements.

Hyperspectral near-infrared imaging for detection of cuticle cracks on tomatoes 

Paper 8027-18 of Conference 8027
Date: Wednesday, 27 April 2011

Author(s): Hoon-Soo Lee, Chungnam National Univ. (Korea, Republic of); Danhee Jeong, Moon S. Kim, Agricultural Research Service, USDA (United States); Byoung-Kwan Cho, Chungnam National Univ. (Korea, Republic of); Stephen R. Delwiche, Kuanglin Chao, Agricultural Research Service, USDA (United States)


Cuticle cracks on tomatoes could be potential harbor sites of pathogenic infection which may cause deleterious consequences to consumer health in fresh cut markets. The feasibility of hyperspectral near-infrared imaging technique with the spectral range of 1000 nm to 1700 nm was investigated for detecting defects on tomatoes. Spectral information obtained from the regions of interest on both defected and whole areas were analyzed to determine optimal wavebands ratio used for further image processing to discriminate the defected areas from the whole tomato surfaces. Unsupervised multivariate analysis method, such as principal component analysis was also explored to improve the detection accuracy. Results showed that the defected tomatoes could be differentiated from the sound ones with accuracy of 94.4%.

Estimation of low-resolution visible spectra from RGB imagery II: simulation results 

Paper 8048-48 of Conference 8048
Date: Wednesday, 27 April 2011

Author(s): Harvey C. Schau, Meridian Systems LLC (United States)


In a previous paper [Schau, H.C.,"Estimation of Low Resolution Visible Spectra from RGB Imagery", Proc. Algorithms and Technology for Multispectral, Hyperspectral, and Ultraspectral Imagers X ,SPIE,Orlando (2009)] , it was demonstrated that an estimate of a low resolution visible spectra of a naturally illuminated outdoor scene can be estimated from RGB values measured by a conventional color imager. In this paper we present a refined algorithm and document results in a study to estimate visible source spectra from solar illumination scenes using reflectance spectra generated from the USGS data base.

Generalized statistics framework for lagrange constraint neural networks 

Paper 8058-22 of Conference 8058
Date: Wednesday, 27 April 2011

Author(s): Ravi C. Venkatesan, Systems Research Corp. (India); Arun Sharma, SecureALL Corp. (United States)


The theory of Lagrange Constraint Neural Networks is re-formulated within the framework of generalized statistics of Tsallis. A numerical algorithm for unmixing endmembers in hyperspectral imaging is formulated. Numerical results exemplifying the theory are presented. A self-consistent methodology to assign values to the Lagrange multipliers based on the theory of phase transitions is presented.

Graph theoretic metrics for spectral imagery with application to change detection 

Paper 8048-8 of Conference 8048
Date: Monday, 25 April 2011

Author(s): James A. Albano, David W. Messinger, Ariel Schlamm, William F. Basener, Rochester Institute of Technology (United States)


A new model for spectral data is presented that is based on graph theory. The spectral graph is constructed by joining a pixel with its m-nearest neighbors with an undirected weighted edge. The weight of each edge corresponds to the spectral Euclidean distance between the connected pixels. We then apply different graph theoretic metrics, such as the Normalized Edge Volume (NEV), to quantify important structural characteristics of the resulting graph. Finally, a graph-based spectral change detection algorithm is presented that is based on the NEV metric. Results are shown for both multispectral and hyperspectral data sets.

Trilateral filter on multispectral imagery for classification and segmentation 

Paper 8048-38 of Conference 8048
Date: Wednesday, 27 April 2011

Author(s): Weihua Sun, David W. Messinger, Rochester Institute of Technology (United States)


We present a new approach to filtering high spatial resolution multispectral (MSI) or hyperspectral imagery (HSI) for classification and segmentation. Our approach is inspired by the bilateral filtering method (Tomasi 1998) that smooths images while preserving important edges. To achieve a similar goal for MSI/HSI, we build a nonlinear tri-lateral filter that takes into account both spatial and spectral similarities. Our approach works on a pixel by pixel basis; the spectrum of each pixel in the filtered image is the combination of the spectra of its adjacent pixels in the original image weighted by the three factors: geometric closeness, spectral Euclidean distance and spectral angle separation. Our approach reduces small clutter across the image while keeping edges with strong contrast. A k-means classifier is applied to the filtered image and its results show our approach can produce a much less cluttered class map.

Infrared imaging technology for detection of bruising damages of 'Singo' pear 

Paper 8027-17 of Conference 8027
Date: Wednesday, 27 April 2011

Author(s): Byoung-Kwan Cho, Chungnam National Univ. (Korea, Republic of); Moon S. Kim, U.S.D.A. Agricultural Research Service (United States); Hoon-Soo Lee, Chungnam National Univ. (Korea, Republic of); Stephen R. Delwiche, U.S.D.A. Agricultural Research Service (United States)


Of the quality attributes of pear bruising damage is the most crucial external quality factor which should be detected in sorting processes. Development of sensitive detection methods for the defects is necessary to ensure accurate quality measurement. Infra-red imaging technique has good potentials for identifying and detecting anomalies due to defects on agricultural materials. In this study, feasibility of hyperspectral infra-red imaging technique for the detection of bruising damages underneath the pear skin was investigated. Damages exist underneath the skin are not easily discernable by using conventional imaging technique at visible wavelength ranges. Simple image combination methods as well as multivariate image analyses were explored to develop optimal image analysis algorithm to detect bruising damages of pear. Results demonstrated good potential of the infra-red imaging for detection of bruising damages underneath the pear skin.

LED induced fluorescence imaging technology for detection of cuticle cracking on cherry tomatoes 

Paper 8027-22 of Conference 8027
Date: Wednesday, 27 April 2011

Author(s): In-Suck Baek, Byoung-Kwan Cho, Chungnam National Univ. (Korea, Republic of); Moon S. Kim, U.S.D.A. Agricultural Research Service (United States); Young-Sik Kim, SangMyung Univ. (Korea, Republic of)


Nondestructve quality measurement is one of the most important postharvest processes in cherry tomato industry. Of the quality attributes of cherry tomatoes, cuticle cracking which are fine hair-like cracks on surfaces produces quality and safety problems. Cracking is the main cause of retailers' rejection and common site for pathogenic penetration and infection. Hence, the cherry tomatoes exposed on the defects should be discriminated in quality sorting processes. In this study, optimal excitation wavelength was investigated using fluorescence emission and excitation matrix of sound and defected areas on cherry tomatoes. High power LEDs of the optimal wavelength were used for hyperspectral fluorescence imaging system to explore the best combination of the emission spectral images. The LED induced fluorescence imaging technique showed excellent potential for discriminating cracked cherry tomatoes.

iCATSI: a multi-pixel imaging differential standoff chemical detection sensor 

Paper 8018-40 of Conference 8018
Date: Wednesday, 27 April 2011

Author(s): Louis M. Moreau, Florent Prel, ABB Analytical Measurement (Canada); Hugo Lavoie, Defence Research and Development Canada (Canada); Claude B. Roy, Christian A. Vallieres, ABB Analytical Measurement (Canada); Jean-Marc Theriault, Defence Research and Development Canada (Canada)


iCATSI is a combination of the CATSI instrument, a standoff differential FTIR optimised for the characterisation of chemicals and the MRi, the hyperspectral imaging spectroradiometer of ABB Bomem. The instrument is equipped with a dual-input telescope to perform optical background subtraction. With that method, the signal from the background is automatically removed from the signal of the object of interest. The instrument is capable of sensing in the VLWIR (cut-off near 14 µm) to support research related to standoff chemical detection. Overview of the capabilities of the instrument and results from tests and field trials will be presented.