Showing papers in "IEEE Geoscience and Remote Sensing Letters in 2007"
TL;DR: Experimental results reveal that, not only does the proposed PCA- based coder yield rate-distortion and information-preservation performance superior to that of the wavelet-based coder, the best PCA performance occurs when a reduced number of PCs are retained and coded.
Abstract: Principal component analysis (PCA) is deployed in JPEG2000 to provide spectral decorrelation as well as spectral dimensionality reduction. The proposed scheme is evaluated in terms of rate-distortion performance as well as in terms of information preservation in an anomaly-detection task. Additionally, the proposed scheme is compared to the common approach of JPEG2000 coupled with a wavelet transform for spectral decorrelation. Experimental results reveal that, not only does the proposed PCA-based coder yield rate-distortion and information-preservation performance superior to that of the wavelet-based coder, the best PCA performance occurs when a reduced number of PCs are retained and coded. A linear model to estimate the optimal number of PCs to use in such dimensionality reduction is proposed
407 citations
TL;DR: The method described makes use of series reversion, the method of stationary phase, and Fourier transform pairs to derive the two-dimensional point target spectrum for an arbitrary bistatic synthetic aperture radar configuration.
Abstract: This letter derives the two-dimensional point target spectrum for an arbitrary bistatic synthetic aperture radar configuration. The method described makes use of series reversion, the method of stationary phase, and Fourier transform pairs to derive the point target spectrum. The accuracy of the spectrum is controlled by keeping enough terms in the two series expansions, and is verified with a point target simulation
347 citations
TL;DR: In this letter, a realization of the keystone transform avoiding interpolation is presented and the moving target is coarsely focused according to the SAR geometry and the platform velocity while exploiting the scaling principle.
Abstract: Synthetic aperture radar (SAR) image formation for a ground moving target necessitates the compensation of the unknown target trajectory. The keystone transform has been employed to remove the linear component of the range migration for the moving target, where interpolation is required. In this letter, a realization of the keystone transform avoiding interpolation is presented. The kernel of this transform, i.e., the range-frequency-dependent azimuth time rescaling, is implemented using only complex multiplications and fast Fourier transforms based on the scaling principle, which has been successfully applied in the equalization of the space-variant range cell migration in SAR processing. In addition, the moving target is coarsely focused according to the SAR geometry and the platform velocity while exploiting the scaling principle. This preliminary focusing is helpful in the isolation of the moving target from ground clutter, so as to facilitate a more refined processing with respect to each mover. SAR raw data combined with simulated echoes of moving targets are utilized to validate the presented approach
285 citations
TL;DR: Effective April 2, 2007, the radiometric calibration of Landsat-5 (L5) Thematic Mapper data that are processed and distributed by the U.S. Geological Survey Center for Earth Resources Observation and Science (EROS) will be updated.
Abstract: Effective April 2, 2007, the radiometric calibration of Landsat-5 (L5) Thematic Mapper (TM) data that are processed and distributed by the U.S. Geological Survey (USGS) Center for Earth Resources Observation and Science (EROS) will be updated. The lifetime gain model that was implemented on May 5, 2003, for the reflective bands (1-5, 7) will be replaced by a new lifetime radiometric-calibration curve that is derived from the instrument's response to pseudoinvariant desert sites and from cross calibration with the Landsat-7 (L7) Enhanced TM Plus (ETM+). Although this calibration update applies to all archived and future L5 TM data, the principal improvements in the calibration are for the data acquired during the first eight years of the mission (1984-1991), where the changes in the instrument-gain values are as much as 15%. The radiometric scaling coefficients for bands 1 and 2 for approximately the first eight years of the mission have also been changed. Users will need to apply these new coefficients to convert the calibrated data product digital numbers to radiance. The scaling coefficients for the other bands have not changed.
261 citations
TL;DR: This letter introduces an embedded-feature-selection (EFS) algorithm that is tailored to operate with support vector machines (SVMs) to perform band selection and classification simultaneously.
Abstract: Hyperspectral images consist of large number of bands which require sophisticated analysis to extract. One approach to reduce computational cost, information representation, and accelerate knowledge discovery is to eliminate bands that do not add value to the classification and analysis method which is being applied. In particular, algorithms that perform band elimination should be designed to take advantage of the structure of the classification method used. This letter introduces an embedded-feature-selection (EFS) algorithm that is tailored to operate with support vector machines (SVMs) to perform band selection and classification simultaneously. We have successfully applied this algorithm to determine a reasonable subset of bands without any user-defined stopping criteria on some sample AVIRIS images; a problem occurs in benchmarking recursive-feature-elimination methods for the SVMs.
248 citations
TL;DR: The results showed that the measured directional gap fraction distributions were similar for both hemispherical photography and TLS data with a high degree of precision in the area of overlap of orthogonal laser scans.
Abstract: A terrestrial laser scanner (TLS) was used to measure canopy directional gap fraction distribution in forest stands in the Swiss National Park, eastern Switzerland. A scanner model was derived to determine the expected number of laser shots in all directions, and these data were compared with the measured number of laser hits to determine directional gap fraction at eight sampling points. Directional gap fraction distributions were determined from digital hemispherical photographs recorded at the same sampling locations in the forest, and these data were compared with distributions computed from the laser scanner data. The results showed that the measured directional gap fraction distributions were similar for both hemispherical photography and TLS data with a high degree of precision in the area of overlap of orthogonal laser scans. Analysis of hemispherical photography to determine canopy gap fraction normally requires some manual data processing; laser scanners offer semiautomatic measurement of directional gap fraction distribution plus additional three-dimensional information about tree height, gap size, and foliage distributions
229 citations
TL;DR: It is shown that approximately the same classification accuracy is obtained using RVM- based classification, with a significantly smaller relevance vector rate and, therefore, much faster testing time, compared with SVM-based classification.
Abstract: This letter presents a hyperspectral image classification method based on relevance vector machines (RVMs). Support vector machine (SVM)-based approaches have been recently proposed for hyperspectral image classification and have raised important interest. In this letter, it is genuinely proposed to use an RVM-based approach for the classification of hyperspectral images. It is shown that approximately the same classification accuracy is obtained using RVM-based classification, with a significantly smaller relevance vector rate and, therefore, much faster testing time, compared with SVM-based classification. This feature makes the RVM-based hyperspectral classification approach more suitable for applications that require low complexity and, possibly, real-time classification.
182 citations
TL;DR: The results show that the new set of reduced spatial features has better performance than the existing length-width extraction algorithm and PSI and is evaluated on two QuickBird datasets.
Abstract: Classification and extraction of spatial features are investigated in urban areas from high spatial resolution multispectral imagery. The proposed approach consists of three steps. First, as an extension of our previous work [pixel shape index (PSI)], a structural feature set (SFS) is proposed to extract the statistical features of the direction-lines histogram. Second, some methods of dimension reduction, including independent component analysis, decision boundary feature extraction, and the similarity-index feature selection, are implemented for the proposed SFS to reduce information redundancy. Third, four classifiers, the maximum-likelihood classifier, backpropagation neural network, probability neural network based on expectation-maximization training, and support vector machine, are compared to assess SFS and other spatial feature sets. We evaluate the proposed approach on two QuickBird datasets, and the results show that the new set of reduced spatial features has better performance than the existing length-width extraction algorithm and PSI
176 citations
TL;DR: An extension of the iterated constrained endmember (ICE) algorithm that incorporates sparsity-promoting priors to find the correct number of endmembers is presented.
Abstract: An extension of the iterated constrained endmember (ICE) algorithm that incorporates sparsity-promoting priors to find the correct number of endmembers is presented. In addition to solving for endmembers and endmember fractional maps, this algorithm attempts to autonomously determine the number of endmembers that are required for a particular scene. The number of endmembers is found by adding a sparsity-promoting term to ICE's objective function.
174 citations
TL;DR: A through-wall imaging problem for a 2-D scalar geometry is addressed and the performances that are achievable by such an inversion scheme are assessed by exploiting synthetic data.
Abstract: A through-wall imaging problem for a 2-D scalar geometry is addressed. It is cast as an inverse scattering problem and tackled under the linear model of the electromagnetic scattering that is provided by the Born approximation. A truncated singular value decomposition inversion scheme is exploited, and the performances that are achievable by such an inversion scheme are assessed by exploiting synthetic data. The cases of weakly and strongly scattering objects are both considered. Finally, an example of reconstruction that is obtained by exploiting experimental data is presented.
173 citations
TL;DR: This letter proposes a two-step method for tree detection consisting of segmentation followed by classification using weighted features from aerial image and lidar, such as height, texture map, height variation, and normal vector estimates.
Abstract: In this letter, we present an approach to detecting trees in registered aerial image and range data obtained via lidar. The motivation for this problem comes from automated 3-D city modeling, in which such data are used to generate the models. Representing the trees in these models is problematic because the data are usually too sparsely sampled in tree regions to create an accurate 3-D model of the trees. Furthermore, including the tree data points interferes with the polygonization step of the building roof top models. Therefore, it is advantageous to detect and remove points that represent trees in both lidar and aerial imagery. In this letter, we propose a two-step method for tree detection consisting of segmentation followed by classification. The segmentation is done using a simple region-growing algorithm using weighted features from aerial image and lidar, such as height, texture map, height variation, and normal vector estimates. The weights for the features are determined using a learning method on random walks. The classification is done using the weighted support vector machines, allowing us to control the misclassification rate. The overall problem is formulated as a binary detection problem, and the results presented as receiver operating characteristic curves are shown to validate our approach
TL;DR: An algorithm to automatically extract the power line from aerial images acquired by an aerial digital camera onboard a helicopter is presented and has successfully been applied in China National 863 project for power line surveillance, 3-D reconstruction, and modeling.
Abstract: There has been little investigation for the automatic extraction of power lines from aerial images due to the low resolution of aerial images in the past decades. With increasing aerial photogrammetric technology and sensor technology, it is possible for photogrammetrists to monitor the status of power lines. This letter analyzes the property of imaged power lines and presents an algorithm to automatically extract the power line from aerial images acquired by an aerial digital camera onboard a helicopter. This algorithm first uses a Radon transform to extract line segments of the power line, then uses the grouping method to link each segment, and finally applies the Kalman filter technology to connect the segments into an entire line. We compared our algorithm with the line mask detector method and the ratio line detector, and evaluated their performances. The experimental results demonstrated that our algorithm can successfully extract the power lines from aerial images regardless of background complexity. This presented method has successfully been applied in China National 863 project for power line surveillance, 3-D reconstruction, and modeling.
TL;DR: In this letter, three data analysis techniques in linear unmixing, detection, and classification are applied to evaluate spectral information within a spatial scene context and it is demonstrated that those old but simplest approaches, i.e., Brovey and multiplicative methods, can generally yield acceptable data analysis results.
Abstract: The limitations of the currently existing pan-sharpening quality indices are analyzed: the absolute difference between pixel values, mean shifting, and dynamic range change is frequently used as spatial fidelity measurement, but they may not correlate well with the actual change of image content; and spectral angle is a widely used metric for spectral fidelity, but the spectral angle remains the same if two vectors are multiplied by two individual constants, which means the average spectral angle between two multispectal images is zero even if pixel vectors are multiplied by different constants. Therefore, it is important to evaluate the quality of a pan-sharpened image under a task of its practical use and to assess spectral fidelity in the context of an image. In this letter, three data analysis techniques in linear unmixing, detection, and classification are applied to evaluate spectral information within a spatial scene context. It is demonstrated that those old but simplest approaches, i.e., Brovey and multiplicative (or after straightforward adjustment) methods, can generally yield acceptable data analysis results. Thus, it is necessary to consider the tradeoff between computational complexity, actual improvement on application, and hardware implementation when developing a pan-sharpening method.
TL;DR: A modified Fisher's linear discriminant analysis (MFLDA) for dimension reduction in hyperspectral remote sensing imagery is presented and the classification result using the MFLDA-transformed data shows that the desired class information is well preserved and they can be easily separated in the low-dimensional space.
Abstract: In this letter, we present a modified Fisher's linear discriminant analysis (MFLDA) for dimension reduction in hyperspectral remote sensing imagery. The basic idea of the Fisher's linear discriminant analysis (FLDA) is to design an optimal transform, which can maximize the ratio of between-class to within-class scatter matrices so that the classes can be well separated in the low-dimensional space. The practical difficulty of applying FLDA to hyperspectral images includes the unavailability of enough training samples and unknown information for all the classes present. So the original FLDA is modified to avoid the requirements of training samples and complete class knowledge. The MFLDA requires the desired class signatures only. The classification result using the MFLDA-transformed data shows that the desired class information is well preserved and they can be easily separated in the low-dimensional space.
TL;DR: An eigensubspace-based filtering approach is proposed for NBI suppression in SAR without using passive-sniff data as the reference signal, and can deal with smart or interrupted NBI.
Abstract: Synthetic aperture radar (SAR) has found wide applications in many areas, e.g., battlefield awareness. However, SAR is vulnerable to various kinds of interference, among which narrow-band interference (NBI) is commonly used. In this letter, an eigensubspace-based filtering approach is proposed for NBI suppression in SAR without using passive-sniff data as the reference signal. Moreover, the proposed method can deal with smart or interrupted NBI. Both simulation and experimental results are provided to illustrate the performance of the proposed approach
TL;DR: This letter presents a comparison between three Fourier-based motion compensation algorithms for airborne synthetic aperture radar (SAR) systems that circumvent the limitations of conventional MoCo, namely the assumption of a reference height and the beam-center approximation.
Abstract: This letter presents a comparison between three Fourier-based motion compensation (MoCo) algorithms for airborne synthetic aperture radar (SAR) systems. These algorithms circumvent the limitations of conventional MoCo, namely the assumption of a reference height and the beam-center approximation. All these approaches rely on the inherent time-frequency relation in SAR systems but exploit it differently, with the consequent differences in accuracy and computational burden. After a brief overview of the three approaches, the performance of each algorithm is analyzed with respect to azimuthal topography accommodation, angle accommodation, and maximum frequency of track deviations with which the algorithm can cope. Also, an analysis on the computational complexity is presented. Quantitative results are shown using real data acquired by the Experimental SAR system of the German Aerospace Center (DLR).
TL;DR: The nonlinear dimensionality reduction and its effects on vector classification and segmentation of hyperspectral images are investigated and the underlying concepts of the proposed framework are presented and experimental results showing the significant classification improvements are presented.
Abstract: The nonlinear dimensionality reduction and its effects on vector classification and segmentation of hyperspectral images are investigated in this letter. In particular, the way dimensionality reduction influences and helps classification and segmentation is studied. The proposed framework takes into account the nonlinear nature of high-dimensional hyperspectral images and projects onto a lower dimensional space via a novel spatially coherent locally linear embedding technique. The spatial coherence is introduced by comparing pixels based on their local surrounding structure in the image domain and not just on their individual values as classically done. This spatial coherence in the image domain across the multiple bands defines the high-dimensional local neighborhoods used for the dimensionality reduction. This spatial coherence concept is also extended to the segmentation and classification stages that follow the dimensionality reduction, introducing a modified vector angle distance. We present the underlying concepts of the proposed framework and experimental results showing the significant classification improvements
TL;DR: A new data assimilation-based approach for the continental-scale evaluation of remotely sensed surface soil moisture retrievals is applied to four separate soil moisture products over the contiguous U.S.
Abstract: A new data assimilation-based approach for the continental-scale evaluation of remotely sensed surface soil moisture retrievals is applied to four separate soil moisture products over the contiguous U.S. The approach is based on quantifying the ability of a given soil moisture product to correct for known rainfall errors when sequentially assimilated into a simple water balance model. Analysis results provide new insight into the continental-scale performance of surface soil moisture retrieval algorithms based on satellite passive microwave, scatterometer, and thermal remote sensing observations.
TL;DR: Preliminary results of mapping rice crop growth using ENVISAT advanced synthetic aperture radar (ASAR) alternating polarization HH/HV data confirm that C-band SAR data have great potential in the development of an operational system for monitoring rice crop Growth in Southern China.
Abstract: This research letter presents preliminary results of mapping rice crop growth using ENVISAT advanced synthetic aperture radar (ASAR) alternating polarization HH/HV data. Four ASAR HH/HV images were collected in the early rice-growth cycle in the test site in 2006, and the temporal response of ASAR data to the rice field was analyzed. The height and biomass of rice were measured during acquisition of ASAR data, and empirical relationships were established between the backscattering coefficient and these two parameters. Based on the temporal variation of the radar response, a method for mapping a rice growth area was developed using the combination of ASAR HH and HV polarization data between two acquisition dates. The results confirm that C-band SAR data have great potential in the development of an operational system for monitoring rice crop growth in Southern China.
TL;DR: The ground resolution of a fixed-receiver bistatic system is studied, showing that it is comparable to that of a monostatic system, and first focused images obtained with the SABRINA-ENVISAT combination are discussed.
Abstract: This letter discusses the implementation of SABRINA, Synthetic Aperture radar Bistatic Receiver for Interferometric Applications. The ground resolution of a fixed-receiver bistatic system is studied, showing that it is comparable to that of a monostatic system. Due to the short distance from target to receiver, large sensitivity is obtained. The noncooperative nature of the bistatic system forces a conservative data-acquisition strategy based on continuously sampling the scattered signal during a temporal window around the predicted satellite overpass time. Also, to be able to synchronize the system in time and in frequency, sampling of a direct signal obtained through an antenna pointed at the satellite is required. Besides the signal processing required to phase-lock the received signal, the bistatic synthetic aperture radar processing needs to take into account the azimuth-dependent phase history. First focused images obtained with the SABRINA-ENVISAT combination are discussed
TL;DR: Results from the FinSAR project, where the E-SAR and Helsinki University of Technology Scatterometer instruments were operated together in order to validate tree-height retrieval algorithms for boreal forest show that the forest height values are in good agreement.
Abstract: In this letter, we present results from the FinSAR project, where the E-SAR and Helsinki University of Technology Scatterometer (HUTSCAT) instruments were operated together in order to validate tree-height retrieval algorithms for boreal forest. The campaign was carried out in Finland in fall 2003. The main instruments of the campaign were the E-SAR airborne radar (operating at L- and X-band) and the HUTSCAT helicopter-borne profiling scatterometer (operating at X- and C-band). We compare and discuss forest height obtained from the inversion quad-pol polarimetric interferometric synthetic aperture radar (SAR) data sets at L-band and forest height obtained from the inversion of single-pol X-band in SAR data with forest height estimates from HUTSCAT scatterometer data. Our results show that the forest height values, which are estimated by means of two different radar instruments, are in good agreement. The correlation between HUTSCAT and E-SAR height estimates ( at L-band and at X-band) underlines the good agreement between the results obtained by the two approaches.
TL;DR: This letter considers a dual-baseline formulation of coherence tomography and shows how practical application of the method is limited by numerical stability, and proposes a regularization technique based on a matrix singular value decomposition to stabilize the inversion.
Abstract: In this letter, we consider a dual-baseline formulation of coherence tomography and show how practical application of the method is limited by numerical stability. To help reduce this, we propose a regularization technique based on a matrix singular value decomposition to stabilize the inversion. We then apply the new dual-baseline algorithm to ground-based radar data from the European Microwave Signature Laboratory. We consider a sample of maize plants and employ dual-baseline interferometric data to reconstruct vertical tomograms through the vegetation as a function of frequency. We use these reconstructions to interpret the primary scattering mechanisms and their polarization dependence
TL;DR: This letter explicitly formulate multichannel and single-channel blind image deconvolution as a PCA problem and shows that the PCA-based blind image decomvolution runs faster and is more robust to noise.
Abstract: Our earlier work revealed a connection between blind image deconvolution and principal components analysis (PCA). In this letter, we explicitly formulate multichannel and single-channel blind image deconvolution as a PCA problem. Although PCA is derived from blur models that do not contain additive noise, it can be justified on both theoretical and experimental grounds that the PCA-based restoration algorithm is actually robust to the presence of white noise. The algorithm is applied to the restoration of atmospheric turbulence-degraded imagery and compared to an adaptive Lucy-Richardson maximum-likelihood algorithm on both real and simulated atmospheric turbulence blurred images. It is shown that the PCA-based blind image deconvolution runs faster and is more robust to noise.
TL;DR: It was found that at least 20 transect measurements, 3 m in length, for study sites ranging from 3.5 to 1225 m2 in size were necessary to get a consistent hrms measurement, and error associated with measurement of hrms generally exceeded plusmn20% of soil-moisture prediction.
Abstract: Surface roughness is a crucial input for radar backscatter models. Roughness measurements of root mean-squared height (hrms) of the same surface can vary depending on the measuring instrument and how the data are processed. This letter addresses the error in hrms associated with instrument bias and instrument deployment issues such as number and length of measurement transects. It was found that at least 20 transect measurements, 3 m in length, for study sites ranging from 3.5 to 1225 m2 in size were necessary to get a consistent hrms measurement. Also, roughness heights of longer transect lengths were highly dependent on the method of detrending the transects. Finally, soil moisture was predicted by inverting the integral equation model using roughness heights taking into account instrument bias, number of measurements, and the detrending method. For common configurations of the Radarsat sensor and reasonable hrms values, error associated with measurement of hrms generally exceeded plusmn20% of soil-moisture prediction
TL;DR: A new multicomponent image segmentation method is developed using a nonparametric unsupervised artificial neural network called Kohonen's self-organizing map (SOM) and hybrid genetic algorithm (HGA) that is used to detect the main features that are present in the image.
Abstract: Image segmentation is an essential process for image analysis. Several methods were developed to segment multicomponent images, and the success of these methods depends on several factors including (1) the characteristics of the acquired image and (2) the percentage of imperfections in the process of image acquisition. The majority of these methods require a priori knowledge, which is difficult to obtain. Furthermore, they assume the existence of models that can estimate its parameters and fit to the given data. However, such a parametric approach is not robust, and its performance is severely affected by the correctness of the utilized parametric model. In this letter, a new multicomponent image segmentation method is developed using a nonparametric unsupervised artificial neural network called Kohonen's self-organizing map (SOM) and hybrid genetic algorithm (HGA). SOM is used to detect the main features that are present in the image; then, HGA is used to cluster the image into homogeneous regions without any a priori knowledge. Experiments that are performed on different satellite images confirm the efficiency and robustness of the SOM-HGA method compared to the Iterative Self-Organizing DATA analysis technique (ISODATA).
TL;DR: This experiment demonstrated the possibility of remotely monitoring surface displacements of the monitored glacier up to a distance of about 3 km even if, due to the lack of ground truths on the observed area, the data interpretation must be carefully worked out.
Abstract: Spaceborne differential synthetic aperture radar (SAR) interferometry has been proven to be a powerful tool in monitoring environmental phenomena and, in particular, in observing glaciers and retrieving information about their surface topography and dynamics. In the last decade, the use of this technique has been successfully extended from space to ground-based observations as a tool for monitoring, on a smaller scale, single landslides, unstable slopes, and more recently, areas covered by snow but not yet glaciers. In this letter, the results of an experimental activity carried out to evaluate the potential of ground-based microwave interferometry to estimate the velocity of an unstable area belonging to a glacier is reported. This experiment demonstrated the possibility of remotely monitoring surface displacements of the monitored glacier up to a distance of about 3 km even if, due to the lack of ground truths on the observed area, the data interpretation must be carefully worked out.
TL;DR: The calibration and first results with the prototype instrument are presented and the prospect of using a supercontinuum laser source in a broadband (hyperspectral) lidar is explored.
Abstract: We have tested the use of a supercontinuum laser source in laser-based spectral backscatter measurement. The calibration and first results with the prototype instrument are presented with a discussion of improvements and applications in laser-based hyperspectral remote sensing and laboratory measurements. This technique enables the spectral study of the backscatter effects and the calibration and test measurements for the purpose of airborne laser measurement. We also explore the prospect of using a supercontinuum laser source in a broadband (hyperspectral) lidar
TL;DR: The theoretical analysis and experiments prove that the proposed filter can greatly remove decorrelation noise while preserving the fringe phase well, even for those fringes with strong curvatures for InSAR processing.
Abstract: An adaptive contoured window filter is proposed to filter off the noise from phase images of interferometric synthetic aperture radar (InSAR) in this letter. The contoured windows can best satisfy the requirement that constrains the phase signal constant inside windows on which low-pass filtering can remove the noise well while the fringe phases are well preserved. The contoured windows are determined by tracing along the local fringe orientation. An algorithm for determining window sizes adaptive to the fringe density is also proposed. The theoretical analysis and experiments prove that the proposed filter can greatly remove decorrelation noise while preserving the fringe phase well, even for those fringes with strong curvatures for InSAR processing
TL;DR: This work investigated the performance of genetic algorithms with partial least squares (GA-PLS) modeling to retrieve live FMC and its components, equivalent water thickness (EWT) and dry matter content (DM), from fresh leaf reflectance in the leaf optical properties experiment dataset.
Abstract: Fuel moisture content (FMC) is an important parameter in forest fire modeling. We investigated the performance of genetic algorithms with partial least squares (GA-PLS) modeling to retrieve live FMC and its components, equivalent water thickness (EWT) and dry matter content (DM), from fresh leaf reflectance in the leaf optical properties experiment dataset. The results show that GA-PLS achieved a good estimation of FMC directly (R2=0.878-0.893) or indirectly (R 2=0.815-0.862) through the joint retrieval of EWT and DM; future work is required to assess the effectiveness of GA-PLS when applied to datasets that consist of low FMC values
TL;DR: This letter proposes a GPU-based implementation of the automated morphological end member extraction algorithm, which is used in this letter as a representative case study of joint spatial/spectral techniques for hyperspectral image processing.
Abstract: Spatial/spectral algorithms have been shown in previous work to be a promising approach to the problem of extracting image end members from remotely sensed hyperspectral data. Such algorithms map nicely on high-performance systems such as massively parallel clusters and networks of computers. Unfortunately, these systems are generally expensive and difficult to adapt to onboard data processing scenarios, in which low-weight and low-power integrated components are highly desirable to reduce mission payload. An exciting new development in this context is the emergence of graphics processing units (GPUs), which can now satisfy extremely high computational requirements at low cost. In this letter, we propose a GPU-based implementation of the automated morphological end member extraction algorithm, which is used in this letter as a representative case study of joint spatial/spectral techniques for hyperspectral image processing. The proposed implementation is quantitatively assessed in terms of both end member extraction accuracy and parallel efficiency, using two generations of commercial GPUs from NVidia. Combined, these parts offer a thoughtful perspective on the potential and emerging challenges of implementing hyperspectral imaging algorithms on commodity graphics hardware.