Search

Thanks to its richness in extractable features, Hyperspectral images (HSI) find an accelerated use in medical, industrial, agricultural, and environmental fields. In this paper, we present a wavelet-based reduction technique that creates a Hypercube containing the most significant features extracted from the original HSI and representing a multi-dimensional array that is utilized for training a Convolutional Neural Network (CNN), which is designed here to classify different types of surfaces or materials. The performance of this approach is tested and proved using two distinct datasets. Then, we compare the same approach with the PCA, a widely used reduction method. The most important contribution of this paper is the implementation of an FPGA-based parallel accelerator to train the same suggested CNN in only 10% of the computational time compared to the classical CPU-based techniques. The Microblaze will be explained and exploited here to play the role of an embedded microprocessor. • A Wavelet-based technique gives a reduced Hypercube rich in pixels spectrum features. • A CNN is designed to classify different types of surfaces or materials. • An FPGA-based parallel design is created to reduce 90% of the CNN training time. • The Microblaze is used to play the role of an embedded microprocessor. [ABSTRACT FROM AUTHOR]

Newly constructed areas (NCA) are continuously emerging with the development of urbanization, which, however, triggered a series of ecological and environmental issues. Therefore, monitoring NCA is of great significance for sustainable urbanization. Extraction of NCA from remote sensing images involves semantic segmentation of constructed areas and change detection. However, the acquisition of samples for training change detection models is time-consuming. In this study, we proposed a multi-task learning framework including unsupervised change detection and supervised semantic segmentation for NCA extraction. The main contributions of this study are: (1) A deep multivariate alternation detection (DMAD) algorithm was proposed for unsupervised change detection of hyperspectral images. By introducing the optimization objective of Multivariate Alternation Detection (MAD) in the loss functions, the features of changed areas can be extracted more effectively; (2) A multi-task learning framework was proposed for the joint training of DMAD change detection and supervised semantic segmentation, where the two modules were mutually optimized via semantic masks and consistency loss. Based on Orbita Hyperspectral Satellites (OHS) images, we conducted experiments in two cities (Qinzhou and Wuzhou) of Guangxi province. The experimental results indicated that the proposed method can achieve the F1-score better than existing unsupervised change detection methods (e.g., DCCA and DSFA) by an average of 15%, and the OA better than existing semantic segmentation methods (e.g., U-Net and FreeNet) by an average of 1%. The F1-score and Kappa for NCA extraction reached above 0.80 for both study areas. The implementation of this paper will be made available at https://github.com/tulilin/Multitask_NCA. [ABSTRACT FROM AUTHOR]

• This paper proposes a cofactorization method to perform jointly classification and representation learning. • The novel coupling approach produces a coherent and fully-interpretable hierarchical model. • The proposed model offers a rich interpretation of the data. • Performance is assessed on synthetic and real data in the specific context of hyperspectral image interpretation. Supervised classification and representation learning are two widely used classes of methods to analyze multivariate images. Although complementary, these methods have been scarcely considered jointly in a hierarchical modeling. In this paper, a method coupling these two approaches is designed using a matrix cofactorization formulation. Each task is modeled as a factorization matrix problem and a term relating both coding matrices is then introduced to drive an appropriate coupling. The link can be interpreted as a clustering operation over the low-dimensional representation vectors. The attribution vectors of the clustering are then used as features vectors for the classification task, i.e., the coding vectors of the corresponding factorization problem. A proximal gradient descent algorithm, ensuring convergence to a critical point of the objective function, is then derived to solve the resulting non-convex non-smooth optimization problem. An evaluation of the proposed method is finally conducted both on synthetic and real data in the specific context of hyperspectral image interpretation, unifying two standard analysis techniques, namely unmixing and classification. [ABSTRACT FROM AUTHOR]

Hyperspectral images (HSI) contain a great number of bands, which enable better characterization of features. However, the huge dimension and information volume brought by the abundant bands may give rise to a negative influence on the efficiency of subsequent processing on hyperspectral images. Band selection (BS) is a commonly adopted to reduce the dimension of HSIs. Different from the previous work, in this paper, hyperspectral band selection problem is formulated as a multi-objective optimization problem, where the band distribution uniformity among the selected bands and inter-class separation distance from a few labeled samples are optimized simultaneously. To fully exploit the relation between the band subsets with different sizes, we construct a multi-objective evolutionary multi-tasking algorithm for hyperspectral band selection (namely MEMT-HBS) to achieve the selected band subsets for all the selected band sizes in one run. To implement MEMT-HBS, the intra-task pairwise learning based solution generation strategy is suggested to evolve the population for each task to achieve high-quality offspring whose selected band size is restricted to a fixed scope. The inter-task band coverage based knowledge transferring strategy is utilized to choose useful individuals from adjacent tasks to further enhance the performance of current task. Compared with the state-of-the-art semi-supervised and unsupervised BS algorithms, empirical results on different standard hyperspectral datasets show that our proposed MEMT-HBS can determine the superior band subset which has a higher image classification accuracy over the comparison algorithms. [ABSTRACT FROM AUTHOR]

Haze contamination is a common issue in optical remote sensing images, including hyperspectral images (HSIs), which can distort the spectral features of land cover objects. Over the last decades, although many haze removal solutions have been developed, very few studies have focused on haze removal of HSIs. Moreover, most of these methods cannot fully explore the abundant spectral information of HSIs in haze removal. To cope with the issues, a data driven method is proposed for haze removal of HSIs in this paper. Specifically, we design a spectral grouping network (SG-Net) to fully utilize the useful information in each spectral band during the reconstruction. To facilitate the relationship construction between hazy image and the corresponding haze-clear image, the proposed SG-Net first groups each HSI into several spectral subsets based on the intra-spectral correlations. Then, these subsets are convoluted in parallel with multiple branches for feature extraction. Furthermore, a novel attention block is designed to connect the adjacent branches for feature transmission, which can distill the useful information (e.g., uncontaminated information in longer wavelength bands) of each subset and assist the reconstruction of HSIs. Comprehensive experiments on both simulated and real haze HSIs showed that SG-Net is more accurate than seven state-of-the-art haze removal methods and is also more robust to different haze levels and shapes. [ABSTRACT FROM AUTHOR]

Spectral super-resolution is a very important technique to obtain hyperspectral images from only multispectral images, which can effectively solve the high acquisition cost and low spatial resolution of hyperspectral images. However, in practice, multispectral channels or images captured by the same sensor are often with different spatial resolutions, which brings a severe challenge to spectral super-resolution. This paper proposed a universal spectral super-resolution network based on physical optimization unfolding for arbitrary multispectral images, including single-resolution and cross-scale multispectral images. Furthermore, two new strategies are proposed to make full use of the spectral information, namely, cross-dimensional channel attention and cross-depth feature fusion. Experimental results on five data sets show superiority and stability of PoNet addressing any spectral super-resolution situations. • Generalized spectral super-resolution is presented for all multispectral imaging. • Considering physical degradation into CNN modeling improves model interpretability. • The proposed attention learns channel-to-channel parameters and speeds up model. • Utilizing both deep and shallow features can further improve model performance. [ABSTRACT FROM AUTHOR]

It has been proved that hyperspectral image (HSI) classification task benefits from introducing additional spatial information. However, how to classify high-dimensional hyperspectral images under the condition of limited training samples is still a challenge. In this paper, we propose a hyperspectral image classification framework based on discriminant manifold broad learning system (DMBLS). We introduce the manifold structure and discriminant information of the data samples as prior knowledge into broad learning system (BLS), which effectively solves the problem of insufficient learning caused by BLS with limited training samples. Firstly we construct two different types of manifold structures for class modelling: the intra-class manifold structures and the inter-class ones. Secondly, we integrate these local discriminant manifold structures to build a manifold regularization framework. Finally, we add this framework to the DMBLS via minimizing the intra-class manifold structures, while maximizing the inter-class ones. And we think this operation can optimize the projection direction of output weights in the DMBLS, enhancing the discrimination ability of these weights. Experimental results on three benchmark HSI datasets show that DMBLS can effectively improve the classification accuracy of hyperspectral images under the condition of limited training samples. [ABSTRACT FROM AUTHOR]

Hyperspectral images (HSIs) have attracted worldwide attention as they make it possible to remotely distinguish between spectrally similar materials. The data of hyperspectral image with high dimensionality, redundant features, unbalanced distribution and so on, however, impose great difficulties for HSI classification. An ensemble of classifiers are usually considered to provide more accurate predictions than a single classifier. In this paper, a heterogeneous ensemble consisting of a support vector machine, a kernel extreme learning machine and a multinomial logistic regression classifier is constructed to increase classification accuracies of HSIs. Furthermore, the feature extraction method is employed to transform the spectral features to promote the diversity of the ensemble, and a voting mechanism is introduced to predict the label on unseen data. Our experimental results with two widely used real hyperspectral datasets demonstrate that the proposed heterogeneous ensemble classifier obtains satisfactory performance with respect to spectral feature sets, even though the number of labelled samples is small. [ABSTRACT FROM AUTHOR]

• Solved the lack of spatial features of HSI in predicting soil carbon content. • Proposed multi-scale spectral feature network to extract spectral feature of HSI. • Proposed multi-scale spatial and spectral feature network to extract feature of HSI. • Proposed methods improve the accuracy and stability of soil carbon content prediction. Soil carbon content prediction based on hyperspectral images can achieve large-scale spatial measurement, which has the advantages of wide coverage and fast information collection, is more suitable for field data collection. However, the research on soil carbon content prediction based on hyperspectral images mainly focuses on feature extraction of spectral information, ignoring the spatial information, and cannot well reveal the intrinsic structural characteristics of data. Aiming at the lack of spatial features consideration in hyperspectral images, soil carbon content prediction methods based on multi-scale feature fusion are proposed by hyperspectral image. At the same time of extracting spectral features from hyperspectral images, the spatial information is used for the first time and a multi-scale spectral and spatial feature network (SpeSpaMN) is designed. In the SpeSpaMN, the multi-scale spectral feature network (SpeMN) is constructed to extract spectral features, the multi-scale spatial feature network (SpaMN) is constructed to extract spatial features. The two networks are fused by using the complementary relationship between different scale features to achieve soil carbon content prediction based on multi-scale feature fusion. The results showed that SpeSpaMN had the best results compared to other methods, followed by the method of SpeMN. The RPD of Inland, Aoshan Bay and Jiaozhou Bay samples based on SpeSpaMN were increased by 47.36%, 37.96% and 4.30% respectively. This paper can effectively solve the problem of the deep fusion of spatial and spectral features in the soil carbon content prediction by hyperspectral image, so as to improve the accuracy and stability of soil carbon content prediction. [ABSTRACT FROM AUTHOR]

Powdery mildew is considered to be one of the important diseases that threaten wheat production. It can cause serious economic losses and damage food security. Therefore, early detection of powdery mildew infection is essential to limit its amplification in reproduction. The purpose of this study is to use the hyperspectral image (HSI) to detect wheat powdery mildew (WPM) in an objective and accurate way. A method is proposed to classify the infection degree of WPM by combining cascade interval selection sensitive band with support vector machine (SVM). Firstly, HSI of leaf and canopy scale are obtained, and the image was segmented based on HSV (hue, saturation, value) color space and OTSU algorithm (HSV + OTSU). The spectral reflectance data were extracted after removing the background region of interest (ROI), and then the multiple scattering correction (MSC) was performed. The spectral data after MSC was called RAW, and then the first derivative (D1) and second derivative (D2) operations were performed respectively. Subsequently, the method of backward interval partial least squares (BiPLS) was used to screen out the characteristic bands of the sensitive interval. Finally, the classification models of WPM disease degree at the leaf scale and the canopy scale (RAW-BiPLS-SVM, D1-BiPLS-SVM and D2-BiPLS-SVM) were established respectively. The results showed that the spectral data after derivative operation was enhanced, and the classification effect of D1-BiPLS-SVM model had advantages at both leaf and canopy scales, in which the overall accuracy (OA) at the leaf scale was 95.2 %, and the OA at canopy scale was 74.4 %. In order to increase the difference between diseases, the canopy-scale disease severity levels were integrated. The correct rates of health and disease were 96.9 % and 99.2 %, respectively, and the OA of disease grade recombination was 87.9 %. Therefore, the method proposed in this paper can not only accurately identified healthy and infected wheat, but also realized the classification of different degrees of WPM, and provide technical support for the prevention and control of WPM. • Wheat powdery mildew (WPM) was detected based on the hyperspectral data. • A method for WPM infection classification based on interval selection sensitivity bands combined with SVM is proposed. • The proposed method can effectively classify WPM from leaf to canopy levels. • The results of the proposed method suggested its potential in field applications. [ABSTRACT FROM AUTHOR]

Hyperspectral images (HSIs) are unavoidably degraded by mixed noise, including Gaussian noise and sparse noise. In this paper, we consider a constrained tubal rank and sparsity model (CTSD) to tackle the HSIs mixed noise removal, which characterizes the clean HSIs via the low-tubal-rank constraint and the sparse noise via the l 0 and l ∞ norm constraints, respectively. Due to the strong non-convexity, the CTSD model is challenging to solve. To tackle the CTSD, we develop the proximal alternating minimization (PAM) algorithm via the exact tensor singular value decomposition (t-SVD) and establish the global convergence under mild assumptions. Since the t-SVD is computationally expensive, especially for large scale images, we further design an efficient inexact PAM algorithm via an inexact t-SVD. The inexact PAM enjoys two advantages: (1) The computational complexity for SVDs of the inexact PAM (O (rn 1 n 2 n 3)) is about twofold faster than that of the exact PAM (O (min (n 1 , n 2) n 1 n 2 n 3)) for r ≪ min (n 1 , n 2) ; (2) The accuracy of the inexact PAM is theoretically guaranteed. Extensive experiments on HSIs denoising demonstrate that the exact and inexact methods both outperform comparing methods in quantitative evaluation metrics and visual effects, and the inexact PAM can compromise between the accuracy and efficiency for large scale HSIs. [ABSTRACT FROM AUTHOR]

This paper explores the performance of Ensembles of Extreme Learning Machine classifiers for hyperspectral image classification and segmentation in a semisupervised and spatially regularized process. The approach assumes that we have available only a small training set of labeled samples, which we enrich with a set of guessed labelings on selected samples from the vast pool of unlabeled image pixels. Selection and label guessing is conditioned to an unsupervised classification of the image pixel spectra, and to the spatial proximity to the labeled samples in the image domain. Unlabeled pixels falling in the spatial neighborhood of a labeled training sample, and belonging to the same unsupervised class, acquire its label. Unsupervised classification can be performed by any clustering technique, in this paper we have resorted to the classical K-means. The classifier built from the enriched training dataset is applied to the entire hyperspectral image. Finally, we perform a spatial regularization of the classification label image, maximizing a rather general prior smoothness criterion, by the selection of the most frequent class in each pixel neighborhood. This paper reports experiments with homogeneous ensembles of ELM, rELM, and OP-ELM classifiers, including a sensitivity analysis over the ensemble size and the number of hidden nodes. Computational experiments on four well known benchmarking hyperspectral images give state-of-the-art results. [ABSTRACT FROM AUTHOR]

• Observing the fact that the 2-D snapshot measurement can preserve certain spatial structure the HSI, we use the snapshot measurement to group nonlocal similar cubes of the unknown HSI, which is both effective and efficient. • A tensor-based reconstruction algorithm is developed using nonlocal low-rank tensor train and 3-D weighted total variation that can effectively characterize the nonlocal self-similarity and the local smoothness properties of the HSI. • A series of experiments on both simulated and real CASSI datasets are carried out to verify the outperformance of NLTT-TV over several other state-of-the-art methods. In this paper, we consider the reconstruction problem of hyperspectral images (HSIs) based on a new type of imaging system termed as coded aperture snapshot spectral imaging (CASSI). CASSI compresses multiple spectral bands of HSI into a single snapshot measurement which enjoys low cost, low bandwidth, and high-speed sensing rate. Despite the promising applications of CASSI systems, the bottleneck lies in the reconstruction algorithms. The existing CASSI reconstruction methods could not fully exploit the underlying structures of HSIs, and thereby demonstrate unsatisfactory reconstruction quality. To remedy this issue, we propose in this paper a novel tensor-based method by modeling two intrinsic HSI priors, namely nonlocal low-rank tensor train (NLTT) and 3-D weighted total variation (3DWTV). Taking into consideration that the snapshot measurement and targeted HSI should share similar spatial structures, we use the snapshot measurement to help group nonlocal spatially similar cubes within the unknown HSI to form 4-D tensors. And the NLTT prior employed on these 4-D tensors can effectively learn the correlations among the spatial, spectral and nonlocal modes thanks to the well-balanced matricization scheme of TT rank. Meanwhile, we employ the 3DWTV prior to characterize the local smoothness of the spatial and spectral modes. The resulting model can be solved efficiently by using the alternative direction method of multipliers (ADMM). Extensive experiments on both simulated and real CASSI datasets validate the improved reconstruction performance of our method over several other competing ones. [ABSTRACT FROM AUTHOR]

Abstract: This paper focuses on the application of Extreme Learning Machines (ELM) to the classification of remote sensing hyperspectral data. The specific aim of the work is to obtain accurate thematic maps of soybean crops, which have proven to be difficult to identify by automated procedures. The classification process carried out is as follows: First, spectral data is transformed into a hyper-spherical representation. Second, a robust image gradient is computed over the hyper-spherical representation allowing an image segmentation that identifies major crop plots. Third, feature selection is achieved by a greedy wrapper approach. Finally, a classifier is trained and tested on the selected image pixel features. The classifiers used for feature selection and final classification are Single Layer Feedforward Networks (SLFN) trained with either the ELM or the incremental OP-ELM. Original image pixel features are computed following a Functional Data Analysis (FDA) characterization of the spectral data. Conventional ELM training of the SLFN improves over the classification performance of state of the art algorithms reported in the literature dealing with the data treated in this paper. Moreover, SLFN-ELM uses less features than the referred algorithms. OP-ELM is able to find competitive results using the FDA features from a single spectral band. [Copyright &y& Elsevier]

Multiple kernel learning (MKL) is a popular and effective method for hyperspectral image classification. However, the communication and interaction among multiple basic kernels are insufficient among multiple basic kernels during the whole training process for traditional MKL methods. In this paper, a multiple kernel learning framework based on self-learning and mutual learning (MKML) is proposed. First, each basic kernel starts pretraining with its own training samples and then uses a trained model to make predictions. Second, the basic kernel selects some informative unlabeled samples with high entropy and queries other basic kernels for labeling. All basic kernels except the one that raised the problem negotiate together to determine the class label of the unlabeled samples. Third, the new pseudo-labeled samples are added to the initial training sample sets to train the model again. Finally, all basic kernels are combined to obtain excellent classification performance by the voting mechanism. The methodology is validated on three real hyperspectral images. The experimental results show that the proposed method exhibits better classification performance than well-known MKL methods. [ABSTRACT FROM AUTHOR]

• An adaptive method is proposed for target detection in hyperspectral images. • The robust NLMS adaptive algorithm is used to design the proposed method. • The method tries to use spectral signature library rather than statistical moments. • The method has high performance for high quality hyperspectral images. In this paper, a novel adaptive algorithm for target detection in hyperspectral images (HSIs) is proposed. In a general classification, the proposed method belongs to the category of those methods which are not based on the statistical moments of the observed HSI (e.g. correlation or covariance matrix). The main processing burden of the proposed method is over a known set of spectral signatures. Assuming a linear spectral mixing model, the proposed method takes a set of spectral signatures which one of them relates to the target material and the others relate to the background materials. Based on an adaptive approach, the normalized least mean square (NLMS) adaptive algorithm is engaged to estimate a weight vector which is almost orthogonal to the background materials spectral signature whereas it makes an absolutely non-orthogonal pair with the target material spectral signature. The estimated weight vector is multiplied by the observed HSI to make the final decision. One synthetic and two real hyperspectral images are considered to evaluate the performance of the proposed method. The evaluation results show that the proposed method outperforms its counterparts. [ABSTRACT FROM AUTHOR]

Abstract Hyperspectral remote sensing has drawn great research interests in earth observation, since massive and contiguous spectrum can provide rich information of ground objects. However, such numerous bands also pose great challenges to efficient processing of hyperspectral images (HSI). Manifold learning has been widely used in feature extraction of HSI data to find intrinsic and compact representations of original spectrum. Nevertheless, lack of explicit and nonlinear mapping is still a critical limitation. In this paper, we propose a supervised learning framework for manifold-learning based HSI feature extraction and classification, which provides a nonlinear and explicit mapping for fast and efficient feature learning. We also introduce two concrete learning methods that are induced from this framework, which can extract different intrinsic topology of HSI data to achieve nonlinear dimensionality reduction. Experiments conducted on benchmark data sets demonstrate that high classification accuracy can be obtained with the proposed framework, which consistently outperforms linear mappings for manifold learning methods. [ABSTRACT FROM AUTHOR]

A candle flame is experimentally studied using a hyperspectral imaging device in this paper. The distributions of temperature and wavelength-dependent emissivity are calculated using a Newton-type iterative method. The results show that the spectral radiation intensity increases from the outside to the inside in the candle flame, and at the same spatial point, the intensity increases as the wavelength increases. The highest temperature of the candle flame is about 1980 K, occurs close to the flame border, indicating the most vigorous reaction. Additionally, the radiation characteristics of the candle flame differ from a gray-body, and the emissivity of the flame decreases as the wavelength increases. [ABSTRACT FROM AUTHOR]

Abstract One major challenge of designing deep learning systems for hyperspectral data classification is the lack of labeled training samples. Inspired by recent manifold learning researches, this paper presents a novel Locality Preserving Convolutional Network to address this challenge. The proposed method invents a semi-supervised locality-preserving regularization operation, and inserts a new layer in the three-dimensional convolutional neural network for end-to-end spatial-spectral classification. The benefits are three-fold. First, by using unlabeled training samples which are more easily available, the proposed method reduces the number of labeled samples required for training a deep learning model; Second, the proposed method incorporates the intrinsic geographical correlation among nearby samples into the extracted features, which prevents it from losing accuracy when only limited labeled samples are available; Third, with a three-dimensional architecture, the proposed method can extract the spatial and spectral features simultaneously from the hyperspectral data for classification. A gradient-decent based approach is deployed to train the whole network in a unified way. Experiments over different benchmarks show that, the proposed method relieves the Hughes phenomenon for deep learning, and achieves competitively high classification accuracy compared to other state-of-the-art approaches. [ABSTRACT FROM AUTHOR]

Abstract Hyperspectral images (HSIs) are often corrupted by noise during the acquisition process, thus degrading the HSI's discriminative capability significantly. Therefore, HSI denoising becomes an essential preprocess step before application. This paper proposes a new HSI denoising approach connecting Partial Sum of Singular Values (PSSV) and superpixels segmentation named as SS-PSSV, which can remove the noise effectively. Based on the fact that there is a high correlation between different bands of the same signal, it is easy to know the property of low rank between distinct bands. To this end, PSSV is utilized, and in order to better tap the low-rank attribute of pixels, we introduce the superpixels segmentation method, which allows pixels in HSI with high similarity to be grouped in the same sub-block as much as possible. Extensive experiments display that the proposed algorithm outperforms the state-of-the-art. [ABSTRACT FROM AUTHOR]

Abstract Nowadays, the hyperspectral remote sensing imagery HSI becomes an important tool to observe the Earth's surface, detect the climatic changes and many other applications. The classification of HSI is one of the most challenging tasks due to the large amount of spectral information and the presence of redundant and irrelevant bands. Although great progresses have been made on classification techniques, few studies have been done to provide practical guidelines to determine the appropriate classifier for HSI. In this paper, we investigate the performance of four supervised learning algorithms, namely, Support Vector Machines SVM, Random Forest RF, K-Nearest Neighbors KNN and Linear Discriminant Analysis LDA with different kernels in terms of classification accuracies. The experiments have been performed on three real hyperspectral datasets taken from the NASA's Airborne Visible/Infrared Imaging Spectrometer Sensor AVIRIS and the Reflective Optics System Imaging Spectrometer ROSIS sensors. The mutual information had been used to reduce the dimensionality of the used datasets for better classification efficiency. The extensive experiments demonstrate that the SVM classifier with RBF kernel and RF produced statistically better results and seems to be respectively the more suitable as supervised classifiers for the hyperspectral remote sensing images. [ABSTRACT FROM AUTHOR]

Abstract Recently, the hyperspectral sensors has improved our ability to monitor the earth surface with high spectral resolution. However, the high dimensionality of spectral data brings challenges for the image processing. Consequently, the dimensionality reduction is a necessary step in order to reduce the computational complexity and increase the classification accuracy. In this paper, we propose a new filter approach based on information gain for dimensionality reduction and classification of hyperspectral images. A special strategy based on hyperspectral bands selection is adopted to pick the most informative bands and discard the irrelevant and noisy ones. The algorithm evaluates the relevancy of the bands based on the information gain function with the support vector machine classifier. The proposed method is compared using two benchmark hyperspectral datasets (Indiana, Pavia) with three competing methods. The comparison results showed that the information gain filter approach outperforms the other methods on the tested datasets and could significantly reduce the computation cost while improving the classification accuracy. [ABSTRACT FROM AUTHOR]

Abstract: An object-oriented mapping approach based on subspace analysis of airborne hyperspectral images was investigated in this paper. Hyperspectral features were extracted based on subspace learning approaches, in order to reduce the redundancy of spectral space and extract the characteristic images for the further object-oriented classification. In this paper, three kinds of spectral feature extraction (FE) methods were utilized to obtain the subspace of airborne hyperspectral data: (1) unsupervised FE, such as PCA (principal component analysis), ICA (independent component analysis) and MNF (maximum noise fraction); (2) supervised FE, e.g. DBFE (decision boundary feature extraction), DAFE (discriminant analysis feature extraction) and NWFE (nonparametric weighted feature extraction); and (3) linear mixture analysis. Afterwards, the extracted subspace features were fed into the object-based classification system. The FNEA (fractal net evolution approach) was utilized to extract objects from the subspace images and SVM (support vector machines) was then used to classify the object-based features. Experiments were conducted on two airborne hyperspectral datasets: (1) the AVIRIS dataset over the northwest Indiana''s Pine with 220 spectral bands (agricultural region), and (2) the ROSIS dataset over Pavia University, northern of Italy with 102 spectral bands (urban region). Results revealed that the proposed object-based approach could give significantly higher accuracies than the traditional pixel-based subspace classification. [Copyright &y& Elsevier]

• Unlike previous studies that used optical technology to irradiate the skin surface, our findings can help pathologists complete rapid pathological diagnoses, improve diagnostic accuracy, and provide new research ideas for distinguishing melanoma and nevus. • In this paper, we innovatively demonstrate the potential to extract a large amount of fluorescence fingerprint information using a small number of hyperspectral pathological images to identify the pigmented nevus and melanoma. • We discovered substantial variances in spectral features in the 500-675nm band range and extracted spectral information in this range of bands for the classification procedure, which achieves 98% accuracy. Cutaneous melanoma, an exceedingly aggressive form of skin cancer, holds the top rank in both malignancy and mortality among skin cancers. In early stages, distinguishing malignant melanomas from benign pigmented nevi pathologically becomes a significant challenge due to their indistinguishable traits. Traditional skin histological examination techniques, largely reliant on light microscopic imagery, offer constrained information and yield low-contrast results, underscoring the necessity for swift and effective early diagnostic methodologies. As a non-contact, non-ionizing, and label-free imaging tool, hyperspectral imaging offers potential in assisting pathologists with identification procedures sans contrast agents. This investigation leverages hyperspectral cameras to ascertain the optical properties and to capture the spectral features of malignant melanoma and pigmented nevus tissues, intending to facilitate early pathological diagnostic applications. We further enhance the diagnostic process by integrating transfer learning with deep convolutional networks to classify melanomas and pigmented nevi in hyperspectral pathology images. The study encompasses pathological sections from 50 melanoma and 50 pigmented nevus patients. To accurately represent the spectral variances between different tissues, we employed reflectance calibration, highlighting that the most distinctive spectral differences emerged within the 500–675 nm band range. The classification accuracy of pigmented tumors and pigmented nevi was 89% for one-dimensional sample data and 98% for two-dimensional sample data. Our findings have the potential to expedite pathological diagnoses, enhance diagnostic precision, and offer novel research perspectives in differentiating melanoma and nevus. [ABSTRACT FROM AUTHOR]

Quality assessment indexes play a fundamental role in the analysis of hyperspectral image (HSI) cubes. To assess the quality of an HSI cube, the structural similarity (SSIM) index has been widely applied in a band-by-band manner, as SSIM was originally designed for 2D images, and then the mean SSIM (MeanSSIM) index over all bands is adopted. MeanSSIM fails to accommodate the spectral structure which is a unique characteristic of HSI. Hence in this paper, we propose a new and simple multivariate SSIM (MvSSIM) index for HSI, by treating the pixel spectrum as a multivariate random vector. MvSSIM maintains SSIM’s ability to assess the spatial structural similarity via correlation between two images of the same band; and adds an ability to assess the spectral structural similarity via covariance among different bands. MvSSIM is well founded on multivariate statistics and can be easily implemented through simple sample statistics involving mean vectors, covariance matrices and cross-covariance matrices. Experiments show that MvSSIM is a proper quality assessment index for distorted HSIs with different kinds of degradations. [ABSTRACT FROM AUTHOR]

• We propose a method of interaction subspace construction. • We design a new subspace model for accounting for the spectral variability. • Based on this subspace model, we propose two detectors, one of which exhibits the most robustness. In this paper, we examine the problem of detecting subpixel targets in hyperspectral images. The so-called subpixel target refers to a target that only occupies a part of a pixel due to the low spatial resolution of hyperspectral sensors. Considering that the subpixel target spectrum is not always reliable (e.g., due to spectral variability), an interaction subspace model is designed to deal with this problem. In this subspace model, the second-order interaction terms are introduced to better describe the spectral variability, thereby improving the robustness. Specifically, the subspace model uses a hyperplane in a high-dimensional space to model spectral variability, while traditional models (e.g., the additive model and the replacement model) use a line in the high-dimensional space to model spectral variability. Obviously, the stronger description ability of the hyperplane makes the subspace model more tolerant to the mismatch of the target spectrum. Based on this interaction subspace model, we derive adaptive detectors according to the one-step generalized likelihood ratio test and its two-step variant. Experiments conducted on hyperspectral data demonstrate that the proposed two-step detector exhibits the strongest robustness in cases where the target spectrum is not very reliable. [ABSTRACT FROM AUTHOR]

In hyperspectral images, spectral mixing occurs when objects lying beside each cannot be distinguished as different entities due to its low spatial resolution. Other hurdles in hyperspectral imaging are its huge dimension and noisy bands. In this paper, a new approach for spectral unmixing is presented where, the data is reduced dimensionally and, the bands eliminated during this are denoised using the existing denoising methods. Then, dataset with these bands is dimensionally reduced and their presence after reduction is validated using spectral unmixing methods. The effectiveness of this method is evaluated using parametric measures such as RMSE and classification accuracy. [ABSTRACT FROM AUTHOR]

The circum-galactic medium consists in gas orbiting around galaxies, whose faintness prevents any complete and easy detection. A powerful tool to detect such pattern can be found in using hyperspectral imaging. Nevertheless, detection in hyperspectral datacubes faces various problems, including well-fitted signal and noise descriptions to ensure further discrimination. A specificity of astronomical images resides in dealing with faint and very noisy signals. In this paper, we introduce a new constrained generalized likelihood ratio test adapted to the problem and a compound test to exploit most of the available information. We also investigate the use of both spatial information and multiple observations on a single scene, to enhance robustness. Numerical experiments on synthetic data are performed to quantify the gain of the different approaches. Finally, results on real hyperspectral astronomical data are presented, which may map for the first time observation of the circum-galactic medium around faint and distant galaxies. [ABSTRACT FROM AUTHOR]

Support Vector Data Description (SVDD) is a nonparametric and powerful method for target detection and classification. The SVDD constructs a minimum hypersphere enclosing the target objects as much as possible. It has advantages of sparsity, good generalization and using kernel machines. In many studies, different methods have been offered in order to improve the performance of the SVDD. In this paper, we have presented ensemble methods to improve classification performance of the SVDD in remotely sensed hyperspectral imagery (HSI) data. Among various ensemble approaches we have selected bagging technique for training data set with different combinations. As a novel technique for weighting we have proposed a correlation based weight coefficients assignment. In this technique, correlation between each bagged classifier is calculated to give coefficients to weighted combinators. To verify the improvement performance, two hyperspectral images are processed for classification purpose. The obtained results show that the ensemble SVDD has been found to be significantly better than conventional SVDD in terms of classification accuracy. [ABSTRACT FROM AUTHOR]

Spectral unmixing plays an important role in hyperspectral image applications and an amount of work has been devoted to hyperspectral unmixing (HU) in recent years. Hyperspectral images (HSIs) are typically corrupted by mixed types of noise and the noise level usually varies among different bands. The unmixing results could be misleading without individualized noise estimations for each band. In addition, there exist conspicuous sparsity of endmembers and spatial correlation in hyperspectral scenes, which can be utilized to improve the hyperspectral unmixing performance. In this paper, we propose a novel model that integrates weighted sparse unmixing with bandwise noise characterization. To deal with various noise present in HSIs, we introduce a weighting matrix derived from the Gaussian noise estimation in each band and a l 1 regularizer for sparse noise. Then we incorporate a weighted sparse regularizer, which includes both spatial and spectral signatures weighting factors to take advantage of spatial correlation among fractional abundances and enhance the sparsity of endmembers simultaneously. An efficient algorithm based on alternating direction method of multipliers (ADMM) is utilized to solve the proposed model. The experimental results on synthetic and real hyperspectral data suggest that the consolidation of noise estimation and weighted regularizer could achieve superior unmixing results compared with other advanced unmixing methods. [ABSTRACT FROM AUTHOR]

Recently, broad learning system (BLS) has been widely used for its simple, fast and excellent generalization ability in hyperspectral image (HSI) classification. However, how to implement a broad learning system for fine-grained classification of hyperspectral images with a few-shot setting is still a challenging problem. In this paper, we proposed a new method based on the discriminative locality preserving broad learning system (DPBLS) for hyperspectral image classification by exploiting the manifold structure between neighbouring pixels of hyperspectral image. To make full use of the spectral and spatial information of hyperspectral images.we firstly leverage edge-preserving filters to fuse both spectral and spatial features of hyperspectral image samples. Secondly, we introduce discriminative information and local manifold structure of samples into the broad learning system to enhance the discriminative ability of output weights and improve its performance on hyperspectral image classification task. In order to verify the performance of the framework proposed in this paper, we conducted experiments on four hyperspectral image datasets. experiment results show that the method we proposed is well-performed on hyperspectral image classification tasks. [ABSTRACT FROM AUTHOR]

Recently, convolutional neural networks have demonstrated excellent prediction performance in hyperspectral image (HSI) classification. However, in traditional methods, the specific design of classification networks requires extensive professional knowledge, and the fixed network architecture lacks adaptability to different datasets. In this paper, a spectral feature perception evolving network (SFPEN), which is a dataset-oriented network method, is proposed. First, to overcome the drawbacks of traditional methods and improve the classification accuracy, an SFPEN driven by an evolutionary algorithm is proposed. The SFPEN automatically designs the network architecture based on a given HSI. Second, spectral feature perception modules are designed to extract the spectral features of HSIs and eliminate redundant information in the HSI narrow bands. Finally, a two-stage network fitness evaluation strategy is designed to reduce the number of training epochs of numerous networks and improve the efficiency of the network evaluation. The experimental results for the available datasets indicate that the proposed method achieves high classification accuracy and demonstrates great adaptability to different datasets. [ABSTRACT FROM AUTHOR]

Feature extraction plays a crucial role in improvement of hyperspectral images classification. Nonparametric feature extraction methods show better performance compared to parametric ones when distribution of classes is non normal-like. Moreover, they can extract more features than parametric methods do. In this paper, a new nonparametric linear feature extraction method is introduced for classification of hyperspectral images. The proposed method has no free parameter and its novelty can be discussed in two parts. First, neighbor samples are specified by using Parzen window idea for determining local mean. Second, two new weighting functions are used. Samples close to class boundaries will have more weight in the between-class scatter matrix formation and samples close to class mean will have more weight in the within-class scatter matrix formation. The experimental results on three real hyperspectral data sets, Indian Pines, Salinas and Pavia University, demonstrate that the proposed method has better performance in comparison with some other nonparametric and parametric feature extraction methods. [ABSTRACT FROM AUTHOR]

Unmixing is the estimation of hyperspectral image pixels composition, specified as the fractional abundances of the composing materials, achieving image segmentation at sub-pixel resolution. Linear unmixing assumes that pixels are convex combinations of endmember spectra, hence endmember identification is required prior to unmixing processes. In our approach to non-linear unmixing by Extreme Learning Machine (ELM) regression ensembles, we do not need to perform endmember identification, which is implicit in the non-linear transformation. Instead we provide estimates of the fractional abundances of predefined material classes, which have been characterized by pure pixels extracted from the image according to available ground truth. In this paper, we introduce a formal discussion of the convergence properties of ELM regression ensembles that endorses the empirical results. The analysis shows them to converge to the exact regression value when the number of components of the ensemble grows, provided that the output is the average of the individual outputs. Besides, the proposed approach allows for a general validation procedure based on the reconstruction error over the entire hyperspectral image. Reconstruction error can be estimated using the mapping from fractional abundances to reconstructed spectra, also achieved by ELM regression ensembles. Therefore, validation can be carried out independently of training data, which can be used completely for model construction. Experimental results on well known benchmark images show that the approach has big advantage over state-of-the-art unmixing approaches. [ABSTRACT FROM AUTHOR]

The problem of limited training samples is always a major concern in hyperspectral remote sensing image classification. In this paper, a sample-screening multiple kernel learning (S2MKL) method is proposed for hyperspectral image classification with limited training samples. The core idea of the proposed method is to employ boosting strategy for screening the limited training samples under MKL framework. Different from existing methods, the proposed MKL method exploits the boosting trick to try different combinations of the limited training samples and adaptively determine the optimal weights of base kernels in the linear combination. Morphological profiles are firstly extracted as the both spatial and spectral features for classification instead of the original spectra. With the morphological profiles, AdaBoost strategy is then introduced to guide the construction of multiple kernel learning machine. By means of boosting strategy, the limited samples are effectively screened and used for classification. Meanwhile, the weights of base kernels in the linear combination are automatically determined in the process of screening samples. Three real hyperspectral data sets are used to evaluate the proposed method. The experimental results show that the proposed boosting-based multiple kernel learning method is superior to state-of-the-art methods in terms of classification performance while limited samples are used. [ABSTRACT FROM AUTHOR]

The digital files are becoming larger and larger with the development of computer hardware and computing power, and the fast processing for large files, e.g., hyperspectral images, is becoming feasible for not only companies but also individuals. However, the acquirement of hyperspectral images still costs a lot. Therefore, security issues like copyright ownership of hyperspectral images need to be taken seriously. Reversible data hiding (RDH) is a technology that can embed watermarking information into multimedia cover to protect copyright. However, the natural-images-based RDH methods cannot exploit the large amount of inter-band redundancy contained by hyperspectral images, which leads to a low efficiency for copyright protection. In this paper, a novel RDH method specially designed for hyperspectral images is proposed. We use the value information from the pixel of an adjacent band to classify each pixel into one of the five types when predicting it, and an adaptive predictor is matched for the pixels of each type to achieve a high prediction accuracy. Finally, a complexity calculation method containing three components is put forward to further improve the embedding performance. Experiments show that the proposed method outperforms the existing RDH method for hyperspectral images and other state-of-the-art RDH methods for natural images. [ABSTRACT FROM AUTHOR]

• At present, there are few studies on Gongcheng moon persimmon. A method based on hyperspectral imaging (400–1000 nm) to identify early damage of Gongcheng persimmon is proposed. Hyperspectral imaging technology has potential application prospects for early persimmon bruise detection. • Develop an early bruise detection model based on the hyperspectral data of Gongcheng persimmon. To provide theoretical support for the early detection of bruises in persimmons. • To explore the possibility of image segmentation of the bruised area by single-band grayscale images at characteristic wavelengths. Using persimmon images to identify bruised areas has potential theoretical implications. This paper explores early bruise detection and bruise region extraction on the Gongcheng persimmon by hyperspectral imaging techniques in the visible and near-infrared (400–1000 nm) regions. In picking and storage, the surface of the Gongcheng persimmon will inevitably be slightly scratched due to impact, extrusion, or abrasion, which is not easy to detect by the naked eye. To find a suitable method to identify crisp persimmons in three stages (sound stage, 12 h, and 24 h after bruising), hyperspectral images of Gongcheng persimmon with wavelengths between 400 and 1000 nm were acquired using a hyperspectral imaging system. The effects of multivariate scatter correction (MSC), standard normal variate transform (SNV), normalization, and Savitzky–Golay (SG) smoothing to remove noise on the accuracy of the support vector machine (SVM) classification model were compared. A genetic algorithm (GA) was used to optimize the SVM parameters, and SPA was used to select the best wavelength for identifying the bruised part of the Gongcheng persimmon. The results show that using a successive projections algorithm (SPA) for characteristic wavelength selection can reduce redundant information and improve computing efficiency. The SVM model has the best classification effect after MSC preprocessing, and the classification accuracy can reach 100 %. The minimal noise fraction (MNF) method was used to successfully segment the bruised parts of Gongcheng persimmons according to the selected optimal wavelength, which indicated that MNF is an effective method to identify the bruised areas. [ABSTRACT FROM AUTHOR]

• Individual models between spectral indices and emissivity of urban surfaces are proposed. • The best wavelengths of low-altitude hyperspectral data for calculating spectral indices are determined. • DVI is the best for higher-emissivity surfaces, and OSAVI is the fittest for lower-emissivity surfaces. • The largest temperature error caused by the emissivity deviation reaches up to 3.7 K. The estimation of land surface energy budgets and land surface temperature (LST) require accurate information on land surface emissivity (LSE). Despite the use of remote sensing or low-altitude sensing data in numerous models for the computation of LSE, it is still highly challenging to precisely predict the LSE of high-heterogeneous surfaces on a micro-scale. This paper thus proposes several individual models to retrieve each surface's emissivity on the pixel level. Various emissivity and reflectance values were obtained upon spectral resampling to found individual models between emissivity and spectral indices (NDVI, DVI, MSAVI, OSAVI, RRI) for classified surfaces. Based on the smallest root mean square error (RMSE), DVI models accurately indicated the emissivity of concrete floors (0.0012), asphalt roads (0.0012), grasses (0.0062), parking lots (0.0035), pavements (0.0057), and red manhole covers (0.0089). RRI models better predicted the emissivity of tiles (0.0039), grey manhole covers (0.0092), and shrubs (0.0005). OSAVI was also the ideal index for retrieving the emissivity of rooftops (0.0009). Validation results showed their effectiveness, with accuracy within 0.001. Finally, temperature errors caused by the deviation between traditional or default emissivity and measured emissivity fluctuated from 0.0 K to 2.9 K or 0.2 K to 3.7 K, respectively. [ABSTRACT FROM AUTHOR]

In the process of semi-supervised hyperspectral image classification, spatial neighborhood information of training samples is widely applied to solve the small sample size problem. However, the neighborhood information of unlabeled samples is usually ignored. In this paper, we propose a new algorithm for hyperspectral image semi-supervised classification in which the spatial neighborhood information is combined with classifier to enhance the classification ability in determining the class label of the selected unlabeled samples. There are two key points in this algorithm: (1) it is considered that the correct label should appear in the spatial neighborhood of unlabeled samples; (2) the combination of classifier can obtains better results. Two classifiers multinomial logistic regression (MLR) and k-nearest neighbor (KNN) are combined together in the above way to further improve the performance. The performance of the proposed approach was assessed with two real hyperspectral data sets, and the obtained results indicate that the proposed approach is effective for hyperspectral classification. [ABSTRACT FROM AUTHOR]

Hyperspectral sensors capture images in hundreds of narrow spectral channels. The spectral signatures for each spatial location provide rich information about an image scene, leading to better “separation” between physical materials and objects. Hyperspectral data are spectrally overestimated and the useful signals usually occupy lower dimensional subspace which needs to be inferred. The signal information is usually concentrated in lower dimensional subspaces. For estimating the number of spectrally distinct signatures present in a hyperspectral data the concept of Virtual Dimensionality (VD), give us the minimum number of spectrally distinct signal sources that characterize this data from a perspective view of target detection and classification. Considering these facts it is important to reduce the volume of data with minimum loss of information and this is the main idea of our algorithm. In this paper, a new VD modified method for estimating the number of spectral distinct signatures has been proposed. To demonstrate the applicability of the developed software tools, we used a few well-known hyperspectral image-data sets. [ABSTRACT FROM AUTHOR]

In this paper, a method for semi-supervised multiview feature extraction based on the multiset regularized kernel canonical correlation analysis (kCCA) is proposed for the classification of hyperspectral images. The covariance matrix of this type of data is naturally composed of distinct blocks of spectral channels, which in turn compose the hypercube. To reduce the dimensionality of the data and extract discriminant features taking advantage of this particular structure, a multiview feature extraction method is applied prior to the classification. The proposed scheme exploits both the labels (as a distinct view on the data) and unlabeled pixels into the computation of cross-correlations and regularizations terms. First, we propose a technique to automatically obtain the segmentation of the spectral profile, based on the correlation between channels. Then, the multiset kernel canonical correlation analysis is applied to find a latent space which represents mutually correlated projected views and labels. Experiments on three real hyperspectral images with two linear classifiers and comparisons to state-of-the-art feature extraction methods show the benefits of this approach, which provides classification accuracies equal or superior to those obtained by training classifiers on the original input space but with only a fraction of the original data dimensionality. [ABSTRACT FROM AUTHOR]

Abstract: Approximate matrix factorization techniques with both nonnegativity and orthogonality constraints, referred to as orthogonal nonnegative matrix factorization (ONMF), have been recently introduced and shown to work remarkably well for clustering tasks such as document classification. In this paper, we introduce two new methods to solve ONMF. First, we show mathematical equivalence between ONMF and a weighted variant of spherical k-means, from which we derive our first method, a simple EM-like algorithm. This also allows us to determine when ONMF should be preferred to k-means and spherical k-means. Our second method is based on an augmented Lagrangian approach. Standard ONMF algorithms typically enforce nonnegativity for their iterates while trying to achieve orthogonality at the limit (e.g., using a proper penalization term or a suitably chosen search direction). Our method works the opposite way: orthogonality is strictly imposed at each step while nonnegativity is asymptotically obtained, using a quadratic penalty. Finally, we show that the two proposed approaches compare favorably with standard ONMF algorithms on synthetic, text and image data sets. [Copyright &y& Elsevier]

Abstract: This paper proposes a semi-supervised approach based on probabilistic relaxation theory. The algorithm performs a consistent multi-class assignment of labels according to the contextual information constraints. We start from a fully connected graph where each initial sample of the input data is a node of the graph and where only a few nodes have been labelled. A local propagation process is then performed by means of a support function where a new compatibility measure has been proposed. Contributions also include a comparative study of a wide variety of data sets with recent and well-known state-of-the-art algorithms for semi-supervised learning. The results have been provided by an analysis of their statistical significance. Our methodology has demonstrated a noticeably better performance in multi-class classification tasks. Experiments will also show that the proposed technique could be especially useful for applications such as hyperspectral image classification. [Copyright &y& Elsevier]

Abstract: For hyperspectral target detection, it is usually the case that only part of the targets pixels can be used as target signatures, so can we use them to construct the most proper background subspace for detecting all the probable targets? In this paper, a dynamic subspace detection (DSD) method which establishes a multiple detection framework is proposed. In each detection procedure, blocks of pixels are calculated by the random selection and the succeeding detection performance distribution analysis. Manifold analysis is further used to eliminate the probable anomalous pixels and purify the subspace datasets, and the remaining pixels construct the subspace for each detection procedure. The final detection results are then enhanced by the fusion of target occurrence frequencies in all the detection procedures. Experiments with both synthetic and real hyperspectral images (HSI) evaluate the validation of our proposed DSD method by using several different state-of-the-art methods as the basic detectors. With several other single detectors and multiple detection methods as comparable methods, improved receiver operating characteristic curves and better separability between targets and backgrounds by the DSD methods are illustrated. The DSD methods also perform well with the covariance-based detectors, showing their efficiency in selecting covariance information for detection. [Copyright &y& Elsevier]

Abstract: This paper proposes an object-based classification scheme for handling remotely sensed images. The method combines the results of a supervised pixel-based classifier with spatial information extracted from image segmentation. First, pixel-wise classification is implemented by a fuzzy output SVM classifier using spectral and textural features of pixels. This classification results to a set of fuzzy membership maps. Operating on this transformed space, a Genetic Sequential Image Segmentation (GeneSIS) algorithm is next developed to partition the image into homogeneous regions. GeneSIS follows a sequential object extraction approach, whereby at each iteration a single object is extracted by invoking a GA-based object extraction algorithm. This module evaluates the fuzzy content of candidate regions, and through an effective fitness function design provides objects with optimal balance between three fuzzy components: coverage, consistency and smoothness. The final classification map is obtained automatically via segmentation, since each segment is extracted with its own class label. The validity of the proposed method is shown on the land cover classification of three different remote sensing images, with varying number of spectral bands (multispectral/hyperspectral), different spatial resolutions and ground truth cover types. The accuracy results of our approach are favorably compared with the ones obtained by other segmentation-based classification techniques. [Copyright &y& Elsevier]

Band selection is a direct and effective method to reduce the spectral dimension, which is one of popular topics in hyperspectral remote sensing. Compared with unsupervised band selection methods, semi-supervised methods seek not only informative but also discriminative band subset by using both labeled and unlabeled samples. However, most currently semi-supervised selection methods simply use a unified criterion on both labeled and unlabeled samples for searching optimal bands, which lacks sample pertinence and adds calculation burden. Since different samples possess different numerical characteristics, optimal criterion on these two kinds of samples may be different. Therefore, a method is required, which can concentrate on the characteristics of labeled and unlabeled samples providing different measure criteria to utilize samples more purposefully. In this paper, a multicriteria semi-supervised model is designed for hyperspectral images band selection. The model is established into two specific tasks: One task measures the amount of information and the redundancy contained in the selected bands from unlabeled samples, the other task utilizes the labeled samples to measure the discrimination of the selected bands. To optimize this model, a multitask optimization strategy is designed to merge the bands information and accelerate the speed of searching the promising bands. In addition, the de-duplication genetic operators are designed to fit the characteristics of hyperspectral images. In this way, the proposed multitask band selection method can select bands with high information, high discrimination, and low redundancy from hyperspectral data in an efficient way according to fully exploiting the numerical characteristics of both labeled and unlabeled samples. Experimental results show the superiority of the proposed method, and demonstrate that the proposed model works more efficiently than the comparison band selection methods. [ABSTRACT FROM AUTHOR]

Abstract: Dimensionality reduction techniques such as principal component analysis (PCA) are powerful tools for the analysis of high-dimensional data. In hyperspectral image analysis, nonnegativity of the data can be taken into account, leading to an additive linear model called nonnegative matrix factorization (NMF), which improves interpretability of the decomposition. Recently, another technique based on underapproximations (NMU) has been introduced, which allows the extraction of features in a recursive way, such as PCA, but preserving nonnegativity, such as NMF. Moreover, in some situations, NMU is able to detect automatically the materials present in the scene being imaged. However, for difficult hyperspectral datasets, NMU can mix some materials together, and is therefore not able to separate all of them properly. In this paper we introduce sparse NMU by adding a sparsity constraint on the abundance matrix and use it to extract materials individually in a more efficient way than NMU. This is experimentally demonstrated on the HYDICE images of the San Diego airport and the Urban dataset. [Copyright &y& Elsevier]

Abstract: We propose a specific content-based image retrieval (CBIR) system for hyperspectral images exploiting its rich spectral information. The CBIR image features are the endmember signatures obtained from the image data by endmember induction algorithms (EIAs). Endmembers correspond to the elementary materials in the scene, so that the pixel spectra can be decomposed into a linear combination of endmember signatures. EIA search for points in the high dimensional space of pixel spectra defining a convex polytope, often a simplex, covering the image data. This paper introduces a dissimilarity measure between hyperspectral images computed over the image induced endmembers, proving that it complies with the axioms of a distance. We provide a comparative discussion of dissimilarity functions, and quantitative evaluation of their relative performances on a large collection of synthetic hyperspectral images, and on a dataset extracted from a real hyperspectral image. Alternative dissimilarity functions considered are the Hausdorff distance and robust variations of it. We assess the CBIR performance sensitivity to changes in the distance between endmembers, the EIA employed, and some other conditions. The proposed hyperspectral image distance improves over the alternative dissimilarities in all quantitative performance measures. The visual results of the CBIR on the real image data demonstrate its usefulness for practical applications. [Copyright &y& Elsevier]

Abstract: This paper presents a three step methodology based on the use of chemical oriented models (MCR and CLS) for extracting out the chemical distribution maps (CDMs) from hyperspectral images, afterwards performing multivariate image analysis (MIA) on the CDMs, and finally extracting “channel” and textural features from the score images related to quality characteristics These features show complementary properties to those directly obtained from the CDMs, since they take advantage of their internal correlation structure. The approach has been successfully applied to the evaluation of homogeneity and cluster presence of API in a novel formulation developed to improve the dissolution of poorly soluble drugs. [Copyright &y& Elsevier]

Abstract: MCR-ALS is a resolution method that has been applied in many different fields, such as process analysis, environmental data and, recently, hyperspectral image analysis. In this context, the algorithm provides the distribution maps and the pure spectra of the image constituents from the sole information in the raw image measurement. Based on the distribution maps and spectra obtained, additional information can be easily derived, such as identification of constituents when libraries are available or quantitation within the image, expressed as constituent signal contribution. This work summarizes first the protocol followed for the resolution on two examples of kidney calculi, taken as representations of images with major and minor compounds, respectively. Image segmentation allows separating regions of images according to their pixel similarity and is also relevant in the biomedical field to differentiate healthy from non-healthy regions in tissues or to identify sample regions with distinct properties. Information on pixel similarity is enclosed not only in pixel spectra, but also in other smaller pixel representations, such as PCA scores. In this paper, we propose the use of MCR scores (concentration profiles) for segmentation purposes. K-means results obtained from different pixel representations of the data set are compared. The main advantages of the use of MCR scores are the interpretability of the class centroids and the compound-wise selection and preprocessing of the input information in the segmentation scheme. [Copyright &y& Elsevier]