Unsupervised Hyperspectral Band Selection Research Articles

Unsupervised Hyperspectral Band Selection via Structure-Conserved and Neighborhood-Grouped Evolutionary Algorithm

Unsupervised Hyperspectral Band Selection via Structure-Conserved and Neighborhood-Grouped Evolutionary Algorithm

Spatial and Spectral Structure Preserved Self-Representation for Unsupervised Hyperspectral Band Selection

Spatial and Spectral Structure Preserved Self-Representation for Unsupervised Hyperspectral Band Selection

Stochastic gate-based autoencoder for unsupervised hyperspectral band selection

Due to its strong feature representation ability, the deep learning (DL)-based method is preferable for the unsupervised band selection task of hyperspectral image (HSI). However, the current DL-based UBS methods have not further investigated the nonlinear relationship between spectral bands, a more robust DL model with effective loss function is desired. To solve the above problem, a novel stochastic gate-based autoencoder (SGAE) has been proposed for the UBS task. With the proposed stochastic gate layer, the desired band subset with learnable parameters can be directly obtained. For obtaining better UBS results, a nonlinear regularization term is added with the loss function to supervise the training process of SGAE. Furthermore, an early stopping criteria with a regularization term-based threshold is developed. Experimental results on four publicly available remote sensing datasets prove the effectiveness of our SGAE.

Robust Dual Graph Self-Representation for Unsupervised Hyperspectral Band Selection

Unsupervised band selection aims to select informative spectral bands to preprocess hyperspectral images (HSIs) without using labels. Traditional band selection methods only work well on Euclidean data, but ignore structural information of pixels and spectral bands. Moreover, they treat each HSI as a whole to exploit latent spatial information while ignoring the difference of spatial distribution between diverse homogeneous regions. In this paper, we propose a robust dual graph self-representation (RDGSR) method for unsupervised band selection. RDGSR uses superpixel segmentation technique to generate homogenous regions of each HSI to extract spatial information. Based on the segmentation result, the superpixel-based similarity graph and band-based similarity graph are constructed from HSIs to record spatial and structural information. With this knowledge, the dual graph convolution is developed and the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">l</i> <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2,1</sub> -norm is introduced in the loss function and regularization term to eliminate the noise in rows for robust and effective band selection. The novelty of RDGSR is the joint utilization of the geometric structure of pixels with spatial consistency and the geometric structure of spectral bands to enhance the performance of band selection in a robust <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">l</i> <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2,1</sub> -norm manner. An iterative optimization algorithm is designed to solve the proposed formulation. Substantial experiments on HSI datasets are conducted to verify the superiority of the proposed RDGSR over the state-of-the-art methods. The source code is available at https://github.com/ZhangYongshan/RDGSR.

Marginalized Graph Self-Representation for Unsupervised Hyperspectral Band Selection

Unsupervised band selection is an essential step in preprocessing hyperspectral images (HSIs) to select informative bands. Most existing methods exploit the spatial information from the entire HSI while ignoring the difference between diverse homogeneous regions. Moreover, traditional methods utilize the limited size of data for model training that may result in degraded generalization performance. In this article, we propose a marginalized graph self-representation (MGSR) method for unsupervised hyperspectral band selection. To explore the spatial information from diverse homogenous regions, MGSR generates the segmentations of an HSI by superpixel segmentation and records the relationships between adjacent pixels of the same segmentation in a structural graph. Meanwhile, to improve the generalization and robustness, infinite corrupted samples are obtained from the original pixels by introducing noises in spectral bands for model training. To solve the proposed formulation, we design an alternating optimization algorithm to marginalize out the corruption and search for the optimal solution. Experimental studies on HSI datasets demonstrate the effectiveness of the proposed MGSR and the superiority over the state-of-the-art methods. The source code is available at <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/ZhangYongshan/MGSR</uri> .

Unsupervised Hyperspectral Band Selection Based on Hypergraph Spectral Clustering

Hyperspectral images can provide spectral characteristics related to the physical properties of different materials, which arouses great interest in many fields. Band selection (BS) could effectively solve the problem of high dimensions and redundant information of HSI data. However, most BS methods utilize a single measurement criterion to evaluate band importance so that the assessment of bands is not comprehensive. To dispose of these issues, we propose the hypergraph spectral clustering band selection (HSCBS) method in this letter. First, a novel hypergraph construction method is proposed to combine bands selected by different priority criteria. Second, based on the hypergraph Laplacian matrix, an unsupervised band selection model named HSCBS is presented to cluster the bands into compact clusters with high within-class similarity and low between-class similarity. The results of comprehensive experimental on two public real datasets demonstrate the effectiveness of HSCBS.

Unsupervised hyperspectral band selection with deep autoencoder unmixing

ABSTRACT Hyperspectral imaging (HSI) is a beneficial source of information for numerous civil and military applications, but high dimensionality and strong correlation limits HSI classification performance. Band selection aims at selecting the most informative bands to minimise the computational cost and eliminate redundant information. In this paper, we propose a new unsupervised band selection approach that benefits from the current dominant stream of deep learning frameworks. The proposed approach consists of two consecutive phases: unmixing and cluster. In the unmixing phase, we utilised a nonlinear deep autoencoder to extract accurate material spectra. In the cluster phase, we calculate the variance for each obtained endmember to construct a variances vector. Then, classical K-mean was adopted to cluster the variances vectors. Finally, the optimal band subset was obtained by choosing only one spectral band for each cluster. We carried out several experiments on three hyperspectral datasets to test the feasibility and generality of the proposed approach. Experimental results indicate that the proposed approach surpasses several state-of-the-art counterparts by an average of 4% in terms of overall accuracy.

Unsupervised Hyperspectral Band Selection With Multigraph Integrated Embedding and Robust Self-Contained Regression

Band selection is an effective means to alleviate the curse of dimensionality in hyperspectral data. Many methods select a compact and low redundant band subset, which is inadequate as it may degrade the classification performance. Instead, more emphasis shall be put on selecting representative bands. In this article, we propose a robust unsupervised band selection method to address this issue. Our method reveals bandwise representativeness based on the comprehensive interband neighborhood structure. It incorporates an interband neighborhood graph into a sparse self-contained regression model in order to provide a reasonable measure for bandwise representativeness. The derived coefficient matrix not only uncovers bandwise importance values but also is coherent to the generalized interband local neighborhood structure. For constructing the interband neighboring structural graph, an integrated multigraph model is employed to achieve better generalization performance. It combines the benefit of multiple graphs but is insusceptible to the defects of a single one. To enhance the reliability of this model, a joint trace minimum and nonnegative constraint is imposed on the coefficient matrix. Accordingly, a multigraph integrated embedding and robust self-contained regression model (MGRSR) is formulated. In addition, an iterative update algorithm is developed to solve the problem. Comparative experiments on three hyperspectral data sets illustrate that MGRSR is robust to various data and has superior performance compared with several state-of-the-art methods.

Band Ranking via Extended Coefficient of Variation for Hyperspectral Band Selection

Hundreds of narrow bands over a continuous spectral range make hyperspectral imagery rich in information about objects, while at the same time causing the neighboring bands to be highly correlated. Band selection is a technique that provides clear physical-meaning results for hyperspectral dimensional reduction, alleviating the difficulty for transferring and processing hyperspectral images caused by a property of hyperspectral images: large data volumes. In this study, a simple and efficient band ranking via extended coefficient of variation (BRECV) is proposed for unsupervised hyperspectral band selection. The naive idea of the BRECV algorithm is to select bands with relatively smaller means and lager standard deviations compared to their adjacent bands. To make this simple idea into an algorithm, and inspired by coefficient of variation (CV), we constructed an extended CV matrix for every three adjacent bands to study the changes of means and standard deviations, and accordingly propose a criterion to allocate values to each band for ranking. A derived unsupervised band selection based on the same idea while using entropy is also presented. Though the underlying idea is quite simple, and both cluster and optimization methods are not used, the BRECV method acquires qualitatively the same level of classification accuracy, compared with some state-of-the-art band selection methods

MIMN-DPP: Maximum-information and minimum-noise determinantal point processes for unsupervised hyperspectral band selection

Band selection plays an important role in hyperspectral imaging for reducing the data and improving the efficiency of data acquisition and analysis whilst significantly lowering the cost of the imaging system. Without the category labels, it is challenging to select an effective and low-redundancy band subset. In this paper, a new unsupervised band selection algorithm is proposed based on a new band search criterion and an improved Determinantal Point Processes (DPP). First, to preserve the original information of hyperspectral image, a novel band search criterion is designed for searching the bands with high information entropy and low noise. Unfortunately, finding the optimal solution based on the search criteria to select a low-redundancy band subset is a NP-hard problem. To solve this problem, we consider the correlation of bands from both original hyperspectral image and its spatial information to construct a double-graph model to describe the relationship between spectral bands. Besides, an improved DPP algorithm is proposed for the approximate search of a low-redundancy band subset from the double-graph model. Experiment results on several well-known datasets show that the proposed optical band selection algorithm achieves better performance than many other state-of-the-art methods.

Variable precision rough set based unsupervised band selection technique for hyperspectral image classification

Unsupervised band selection is still a relevant research topic for mitigating certain challenges of hyperspectral image classification. In this paper, a greedy unsupervised hyperspectral band selection technique is proposed based on variable precision rough set (VPRS). The proposed technique defined a novel dependency measure by exploiting VPRS. Furthermore, the dependency measure is defined in such a way that it became less sensitive to the degree of misclassification parameter β in VPRS. Our technique first computed the similarity between every pair of bands using the proposed dependency measure and selected a band from the pair that produced maximum similarity value. After that a novel criterion is proposed to select the informative bands one-by-one by adopting first order incremental search. The effectiveness of the proposed band selection technique is assessed by comparing it with five state-of-the-art techniques using three hyperspectral data sets.

Unsupervised hyperspectral band selection using ranking based on a denoising error matching approach

ABSTRACT In this paper, we focus on utilizing the image denoising method for ranking of significant bands in hyperspectral imagery. We make use of the fact that the denoising error of bands varies with the significant information content of the bands in hyperspectral imagery. The denoising error is computed for each band individually and compared using a matching parameter with the denoising error of a reference image. The reference image is selected to be the first principal component corresponding to the maximum information. Three matching parameters including mutual information (MI), correlation coefficient (r) and the structural similarity index (SSIM) were used for ranking the bands based on the match with the denoising error of the reference image. The proposed algorithm is tested using three datasets, namely, Indian Pines, Salinas and Dhundi. The Indian Pines and Salinas datasets were acquired from the Airborne Visible Infrared Imaging Spectrometer (AVIRIS) sensor and comprised rural and agricultural area. The Dhundi dataset of Hyperion comprises mostly of features corresponding to snow-covered mountainous regions. To assess the accuracy of the proposed method, a supervised classification was carried out using a random forest classifier with 20% training pixels selected randomly from the ground reference. The proposed method yielded significantly better results determined by the kappa coefficient (κ) of 0.756, 0.910 and 0.996 for the Indian Pines, Salinas and Dhundi datasets, respectively, over several other state of the art methods. The classification results of the proposed method also yielded better results than those obtained by the state-of-the-art methods for hyperspectral band selection.

MIMR-DGSA: Unsupervised hyperspectral band selection based on information theory and a modified discrete gravitational search algorithm

Band selection plays an important role in hyperspectral data analysis as it can improve the performance of data analysis without losing information about the constitution of the underlying data. We propose a MIMR-DGSA algorithm for band selection by following the Maximum-Information-Minimum-Redundancy (MIMR) criterion that maximises the information carried by individual features of a subset and minimises redundant information between them. Subsets are generated with a modified Discrete Gravitational Search Algorithm (DGSA) where we definine a neighbourhood concept for feature subsets. A fast algorithm for pairwise mutual information calculation that incorporates variable bandwidths of hyperspectral bands called VarBWFastMI is also developed. Classification results on three hyperspectral remote sensing datasets show that the proposed MIMR-DGSA performs similar to the original MIMR with Clonal Selection Algorithm (CSA) but is computationally more efficient and easier to handle as it has fewer parameters for tuning.

An Unsupervised Hyperspectral Band Selection Method Based on Shared Nearest Neighbor and Correlation Analysis

Band selection is an important dimensionality reduction (DR) methodology for hyperspectral images (HSI). In recent years, many ranking-based clustering band selection methods have been developed. However, these methods do not consider the combination of bands in different clusters but only select the desired number of clustering centers based on band ranking to construct the reduced band subset, which may lead to obtaining a set of bands with low redundancy but little information or a set of bands with a large amount of information but high redundancy, thus falling into the local optimal solution set. To solve this problem, an unsupervised hyperspectral band selection method based on shared nearest neighbor and correlation analysis (SNNCA) is proposed in this paper. The proposed SNNCA method considers the interaction of bands in different clusters, and can obtain a set of bands with a large amount of information and low redundancy. First, this method uses the shared nearest neighbor to describe the local density of each band and takes the product of local density and distance factor as the weight to rank each band to select the required number of clustering centers, which ensures low redundancy among the clustering centers. Then, all bands are grouped into several clusters based on the Euclidean distance matrix and the clustering centers. Finally, the correlation among intra-cluster and inter-cluster bands and the information entropy are further analyzed, and the most representative band is selected from each cluster. The experimental results on two HSI datasets demonstrate that the proposed SNNCA method achieves better classification performance than that of other state-of-the-art comparison methods and possesses competitive running time.

A Local Potential-Based Clustering Algorithm for Unsupervised Hyperspectral Band Selection

Unsupervised band selection plays an increasingly important role in a hyperspectral image (HSI) classification because of inadequate labeling samples. However, how to select more representative, less redundant, and informative band is an open problem. Recently, a fast density-peak-based clustering (FDPC) algorithm has been proposed. The FDPC chooses the cluster center through the local density and the intracluster distance of each point and ranks each point through a certain rule. For HSI band selection, the FDPC has the following problems. First, when calculating the local density of bands, the difference between bands is not considered, so the local density cannot better characterize the band distribution. Second, the ranking rule in FDPC is more inclined to select bands with larger density and intracluster distance values. However, some boundary bands with abundant information and low redundancy cannot be found. This paper proposes a local potential-based clustering algorithm for unsupervised hyperspectral band selection (LPC). The LPC algorithm improves the FDPC algorithm in three aspects of band selection. First, the local potential of each band is calculated according to the similarity of between bands, and the larger similarity has a greater effect on the local potential. The local potential can better characterize the band distribution. Second, the LPC proposes a weighted ranking rule, which integrates the three factors of local potential, intracluster distance and standard deviation to guide the algorithm to select more discriminative bands. In particular, some boundary bands with abundant information and low redundancy can be found. Finally, an effective method is designed to automatically select the appropriate number of bands. The experimental results on three real hyperspectral data sets demonstrate that the proposed method outperforms the other state-of-the-art methods.

Unsupervised hyperspectral band selection by combination of unmixing and sequential clustering techniques

ABSTRACT Selecting the decisive spectral bands is a key issue in unsupervised hyperspectral band selection techniques. These methods are the most popular ways for dimensionality reduction of original data. A compact data representation without compromising the physical information and optimizing the separation between different materials are the main objectives of such selection processes. In this work, a hyperspectral band selection approach is proposed based on linear spectral unmixing and sequential clustering techniques. The use of these two specific techniques constitutes the main novelty of this investigation. The proposed approach operates in different successive steps. It starts with extracting material spectra contained in the considered data using an unmixing method. Then, the variance of extracted spectra samples is calculated at each wavelength, which results in a variances vector. This one is segmented into a fixed number of clusters using a sequential clustering strategy. Finally, only one spectral band is selected for each segment. This band corresponds to the wavelength at which a maximum variance value is obtained. Experiments on three real hyperspectral data demonstrate the superiority of the proposed approach in comparison with four methods from the literature.

Unsupervised hyperspectral band selection for apple Marssonina blotch detection

Apple Marssonina blotch (AMB) is a severe fungal disease that has been plaguing top apple producing countries in the world since it was first found in Japan in 1907. The disease causes premature defoliation and eventually leads to fruit shrinkage and reduction of starch content. AMB has a long latency period ranging from two to five weeks and at its early symptomatic stage, the disease develops symptoms similar to other apple blotch-like diseases, thus making it difficult to detect using only visible information. Hyperspectral imagery was investigated in this study for the detection of different stages of AMB. While hyperspectral images contain a wealth of information that can help distinguish between similar-looking objects, they also contain a large amount of redundancy. An unsupervised feature selection method called orthogonal subspace projection (OSP) was used to perform feature selection and redundancy reduction simultaneously. Ten optimal spectral bands were selected using the algorithm, with six out the selected bands within the same near-infrared spectral region. These bands served as input features for three classifiers—ensemble bagged, decision tree and weighted k-nearest neighbor. The selected bands and classifiers achieved overall accuracy ranging from 71.3% to 84.3%, thus indicating the feasibility of using the OSP feature selection method for reducing the size of hyperspectral data and designing a multispectral imaging system for detecting various AMB disease stages.

Unsupervised Hyperspectral Band Selection by Fuzzy Clustering With Particle Swarm Optimization

Due to the lack of label information and the intrinsic complexity of hyperspectral images (HSIs), unsupervised band selection is always one of the most challenging tasks in HSI processing. Fuzzy clustering is a promising technique for unsupervised band selection, which can partition unlabeled data into groups effectively. However, due to the limits of its optimization process, standard fuzzy clustering is sensitive to initialization and easy to be trapped in a local optimum. To address the limits, a novel unsupervised band selection method is proposed, combining fuzzy clustering with particle swarm optimization (PSO). A newly designed PSO algorithm is introduced to improve the performance of fuzzy clustering band selection. Moreover, a new strategy is designed to select representative cluster centers according to the characteristics of HSIs. The experimental results indicate that the proposed method has the ability to select high-quality band subsets with good and robust performance on HSI classification.

Unsupervised Hyperspectral Band Selection by Dominant Set Extraction

Unsupervised hyperspectral band selection has been an important topic in hyperspectral imagery. This technique aims at selecting some critical and decisive spectral bands from an original image for compact representation without compromising and distorting the raw information in the relevant spectral bands. Although many efforts have been made to this topic, the structural information has not yet been well exploited during band selection, and there are still several deficiencies in search strategies, leaving room for further improvement. This paper tackles the unsupervised hyperspectral band selection problem from a global perspective and proposes a novel method claiming the following main contributions: structure-aware measures for band informativeness and independence; and a graph formulation of band selection allowing for an efficient integrated search by means of dominant set extraction. Experiments on three real hyperspectral images demonstrate the superiority of the proposed band selector in comparison with benchmark methods.