Hyperspectral Imagery Classification Research Articles

Spatial Peak-Aware Collaborative Representation for Hyperspectral Imagery Classification

In this letter, a novel spatial peak-aware collaborative representation (SPaCR) method is proposed for hyperspectral imagery (HSI) classification, which introduces spectral–spatial information among superpixel clusters into regularization terms to construct a new collaborative representation (CR)-based closed-form solution. The proposed method is composed of the following key steps. First, the raw HSI is clustered into many superpixels according to an oversegmentation strategy. Then, cluster pixels are determined based on spectral–spatial correlation between pixels within each superpixel. Next, spectral distance and spatial coherence of superpixel clusters corresponding to training samples and testing pixels are fused to define differences between pixels. Finally, the difference information between clusters as a spectral–spatial feature-induced regularization term is incorporated into the objective function. Experimental results on the Indian Pines and the University of Pavia HSIs indicated that the proposed SPaCR method, without any preprocessing and postprocessing, outperforms well-known and state-of-the-art classifiers on the limited labeled samples.

Hyperspectral Imagery Classification via Random Multigraphs Ensemble Learning

Hyperspectral imagery (HSI) classification, which attempts to assign hyperspectral pixels with proper labels, has drawn significant attention in various applications. Recently, the graph-based semisupervised learning (SSL) approaches have shown the outstanding ability to handle the situation of limited labeled data for classification task. However, it is hard to construct the pairwise adjacent graph due to the high dimensionality of hyperspectral data. Besides, the graph-based SSL models are usually decided by a single classifier, which fail to effectively learn the complex structures and intrinsic properties of HSI. To address these problems, we propose a novel graph-based SSL classification model for HSI, which is based on random multigraphs construction and ensemble strategy (RMGE). Specifically, the anchor graph (AG) is constructed with spatial–spectral features, which integrates the spatial characteristics extracted by local binary pattern on each selected spectrum, preserving fine structures of local region. In order to enhance the discriminative capability of the classifier and avoid the trivial solution, the maximum entropy regularization is added into adjacent AG model. In addition, to capture the diversity of HSI data effectively, we design the ensemble framework by employing multiple AGs to learn HSI features. Experiments conducted on real hyperspectral datasets indicate that the proposed RMGE shows better performance than that of state-of-the-art approaches.

Extended Collaborative Representation-Based Hyperspectral Imagery Classification

Collaborative representation (CR) has been demonstrated to be very effective for hyperspectral image classification. However, insufficient diversity of training samples often results in limited classification accuracy under small-training-sample conditions, especially when diverse spectral variation is presented in testing samples. In order to alleviate such a problem, a spectral variation augmented-based linear mixed model (SV-LMM) is proposed, in which the spectral variation is extracted by conducting singular value decomposition (SVD) over training samples. Such spectral variation is further utilized to extend the CR for hyperspectral classification. Experiments over two benchmark datasets, i.e., the Pavia Center dataset and the University of Houston dataset, demonstrate that the proposed extended CR-based classifier (ECRC) clearly improves the performance of conventional CRC for hyperspectral classification and outperforms several state-of-the-art algorithms.

SPNet: Spectral Patching End-to-End Classification Network for UAV-Borne Hyperspectral Imagery With High Spatial and Spectral Resolutions

In deep learning (DL)-based hyperspectral imagery classification, “spatial patching” is primarily used as a preprocessing for incorporating local spatial information. This operation can help to promote classification accuracy but it is facing new challenges in the unmanned aerial vehicle (UAV)-borne hyperspectral imagery with high spatial and spectral resolutions (H² imagery). The ground objects’ various spatial scales result in it being challenging to determine the optimal size for the spatial patches. In addition, due to the severe spectral variability and spatial heterogeneity of the H² imagery, “spatial patching” only exploits the local spatial information and results in serious salt-and-pepper (SP) noise and isolated areas in the classification maps. In this article, to address these issues, a novel spectral patching network (SPNet) with an end-to-end DL architecture is proposed for UAV-borne H² imagery classification. The “spectral patching” approach is proposed to preserve the global spatial information and almost all the spectral information of the original hyperspectral imagery. An end-to-end deep encoder–decoder network is then constructed based on the spectral patching mechanism, which introduces the deep residual network (ResNet) and atrous spatial pyramid pooling (ASPP) modules to extract multiscale high-level semantic information for the H² imagery classification. The experimental results obtained with the Wuhan UAV-borne H² imagery (WHU-Hi) UAV-borne hyperspectral data set demonstrate that SPNet can achieve state-of-the-art accuracy and visualization performance in the classification of H² imagery.

Spectral Variation Augmented Representation for Hyperspectral Imagery Classification With Few Labeled Samples

Due to variation of imaging conditions, spectra of the same type of ground objects usually exhibit certain discrepancy, leading to intra-class spectral distance increase and inter-class distance decrease. As a result, classification accuracy is greatly affected, especially in cases with few labeled samples. For representation based classifiers, the spectral variability within limited training samples is far from sufficient to represent diverse variations within testing ones. To handle this problem, a spectral variation augmented representation for hyperspectral imagery classification (SVARC) with few labeled samples is proposed in this article. Firstly, a novel class-independent and class-dependent components based linear representation model (CICD-LRM) is proposed to emphasize the representation of spectral variation. Secondly, depending on spatial and spectral correlation, the CICD-LRM guided global and local spectral variation extraction schemes are designed, and a fused spectral variation dictionary is constructed by concatenation. Finally, a classifier for hyperspectral images based on the CICD-LRM and spectral variation dictionary is proposed, and specifically three different spectral variation reconstruction strategies are designed. Similar to most of the representation based classifiers, residual-driven decision is also employed in the proposed classifier. Comparative experiments are conducted with eight classical and state-of-the-art methods using two benchmark datasets. The experimental results demonstrate that the proposed SVARC method significantly outperforms the compared ones in cases with few labeled samples.

Attention-Aware Pseudo-3-D Convolutional Neural Network for Hyperspectral Image Classification

Convolutional neural networks (CNNs) have been applied for hyperspectral image classification recently. Among this class of deep models, 3-D CNN has been shown to be more effective by learning discriminative features from abundant spectral signatures and spatial contexts in hyperspectral imagery (HSI). However, by simply imposing 3-D CNN to HSI, a large amount of initial information might be lost in this CNN pipeline. The proposed attention-aware pseudo-3-D (AP3D) convolutional network for HSI classification is motivated by two observations. First, each dimension of the 3-D HSI is not equally important, different attention should be paid to different dimensions of the initial HSI image, especially in the first convolution operation. Second, intermediate representations of the 3-D input image at different stages in the 3-D CNN pipeline represent different levels of features and should not be neglected and abandoned. Instead, a 2-D matrix of scores for each feature map should be fed to the final softmax layer. Quantitative and qualitative results demonstrate that the proposed AP3D model outperforms the state-of-the-art HSI classification methods in agricultural and rural/urban data sets: Indian Pines, Pavia University, and Salinas Scene.

A spectral-spatial attention aggregation network for hyperspectral imagery classification

ABSTRACT For the classification of hyperspectral imagery (HSI), the convolutional neural network (CNN) can learn the discriminative spatial-spectral information of the image better than the traditional classification methods. However, when CNN uses the local receptive field to extract the features of HSI, it may cause the feature expression of the same pixel on the feature map to be inconsistent, and eventually cause noise in the classification results. To overcome this, we introduce the attention mechanism in the CNN model to improve the feature expressiveness. A spectral-spatial attention aggregation network (SSAAN) for HSI classification is designed, and there are two attention branches in our method. The spectral attention module with the squeeze-and-excitation (SESAM) automatically obtains the importance of each feature channel of HSI, and then enhances the useful band features and suppresses the less-useful band features according to this importance. In the spatial attention module with selective kernel (SKSAM), first, different convolution kernels of 2D-CNN are used to extract the shallow-middle-deep layer features from the principal components after dimension reduction, and the pixel spatial information from the three paths is combined and aggregated. Then, the feature maps of kernels of different sizes are aggregated according to the selection weights. Finally, the feature vectors obtained from the two branches of the spatial attention module and the spectral attention module are connected to further improve feature representation, and the classification result is obtained by the softmax function. Experimental results through three real HSI data sets show that our proposed method SSAAN achieves better performance compared to the state-of-the-art methods.

Self-training algorithm for hyperspectral imagery classification based on mixed measurement k-nearest neighbor and support vector machine

Self-training is commonly used for hyperspectral imagery semi-supervised learning, which selects confident unlabeled samples and puts them into the labeled set iteratively. However, self-training suffers from the influence of noises and outliers that are wrongly selected as confident samples and guide the accuracy of the classifier going down in iteration. To solve this problem, a self-training algorithm for hyperspectral imagery classification based on a mixed measurement k-nearest neighbor (k-NN) and support vector machine (SVM) framework is proposed. Different from the traditional self-training based on k-NN and SVM, the proposed method utilizes a mixed measurement k-NN to select confident unlabeled samples, and the SVM is applied to help the k-NN method to label the unlabeled samples. The mixed measurement is combined with the spectral distance, spatial distance, and local outlier factor (LOF) distance to explore the feature differences of spectra. Meanwhile, the factors of LOF and k-NN are obtained adaptively, which effectively reduces the computational complexity of the proposed method to select optimized parameters. Experimental results on five hyperspectral images show that the proposed method outperforms the competitors.

Lithology Classification Using TASI Thermal Infrared Hyperspectral Data with Convolutional Neural Networks

In recent decades, lithological mapping techniques using hyperspectral remotely sensed imagery have developed rapidly. The processing chains using visible-near infrared (VNIR) and shortwave infrared (SWIR) hyperspectral data are proven to be available in practice. The thermal infrared (TIR) portion of the electromagnetic spectrum has considerable potential for mineral and lithology mapping. In particular, the abovementioned rocks at wavelengths of 8–12 μm were found to be discriminative, which can be seen as a characteristic to apply to lithology classification. Moreover, it was found that most of the lithology mapping and classification for hyperspectral thermal infrared data are still carried out by traditional spectral matching methods, which are not very reliable due to the complex diversity of geological lithology. In recent years, deep learning has made great achievements in hyperspectral imagery classification feature extraction. It usually captures abstract features through a multilayer network, especially convolutional neural networks (CNNs), which have received more attention due to their unique advantages. Hence, in this paper, lithology classification with CNNs was tested on thermal infrared hyperspectral data using a Thermal Airborne Spectrographic Imager (TASI) at three small sites in Liuyuan, Gansu Province, China. Three different CNN algorithms, including one-dimensional CNN (1-D CNN), two-dimensional CNN (2-D CNN) and three-dimensional CNN (3-D CNN), were implemented and compared to the six relevant state-of-the-art methods. At the three sites, the maximum overall accuracy (OA) based on CNNs was 94.70%, 96.47% and 98.56%, representing improvements of 22.58%, 25.93% and 16.88% over the worst OA. Meanwhile, the average accuracy of all classes (AA) and kappa coefficient (kappa) value were consistent with the OA, which confirmed that the focal method effectively improved accuracy and outperformed other methods.

Flexible 3-D Gabor features fusion for hyperspectral imagery classification

In recent years, Gabor filtering has been successfully applied in spectral–spatial hyperspectral image (HSI) classification tasks due to its strong power to characterize surface materials. Generally, the standard Gabor filter involves the real and imaginary parts. However, the traditional usage to integrate both the two parts, i.e., the Gabor magnitude feature, might weaken some unique characteristics of each part, therefore leading to possible information loss. To solve this problem, we proposed a flexible 3-D Gabor features fusion (F3DGF) approach to make better use of two parts of Gabor features, based on the phase-induced 3-D Gabor feature, which is rarely exploited before. As the term suggests, the phase-induced feature is guided by a phase parameter P, which can flexibly combine the information from two parts of Gabor components. The proposed F3DGF scheme explores all the phase-induced 3-D Gabor features by means of a decision-level fusion strategy, where the obtained probability outputs are directly gathered to generate the decision map. Experimental results on three real HSIs demonstrate that our approach exhibits good improvements, as compared to the traditional real part and the magnitude Gabor features-based classification methods. The results show great potential to introduce phase-induced 3-D Gabor features for classification tasks.

ContrastNet: Unsupervised feature learning by autoencoder and prototypical contrastive learning for hyperspectral imagery classification

Hyperspectral classification is a fundamental problem for applying hyperspectral technology, and unsupervised learning is a promising direction to address the issue. Unsupervised feature learning algorithm extracts representative features, which help efficiently classify the hyperspectral pixels. This paper combines the popular contrastive learning method (prototypical contrastive learning) and the classic representation learning method (autoencoder) to design an unsupervised feature learning network for hyperspectral classification. First, two different encoders are used to extract different features as augmentation function in a contrastive network (ContrastNet). Second, the prototypical contrastive learning method is adopted to train a contrastive network. Third, features extracted by the contrastive network are used for classification. Experiments have proved that our two proposed autoencoder networks show good feature learning capabilities by themselves and the designed contrastive learning network can further learn better representative features from the two modules. The three modules compose an efficient framework for unsupervised feature learning. Our method also performs reasonably fast in the testing phase, which implies that it is applicable in practice under specific conditions.

Training redefinition with entropy-based structure set density for supervised hyperspectral imagery classification

ABSTRACT Reliable labelled samples have always played a vital role in the supervised paradigm of hyperspectral imagery (HSI) classification due to the fact that the inclusion of incorrect label information in the training set can seriously degrade the performance of classification methods. Recently, although some inter-class difference-based detection algorithms have been developed to remove mislabelled samples (i.e. noisy labels) in training set, the benefit of contextual information for each sample has not been fully explored yet. In this paper, a training redefinition with entropy-based structure set density (ESSD) method is designed, which consists of following main steps. First, the proposed ESSD method employs an over-segmentation technique to cluster the HSI into many shape-adaptive regions that correspond to sample sets. Then, each sample set is represented with an affine hull (AH) model, which exploits both the similarity and variance of samples within each sample set to adaptively characterize the set. Specifically, considering spectral and spatial weak assumptions among samples in each sample set, the idea of entropy trick-based -nearest neighbour is introduced into each sample set to redefine its structure by removing different class from the sample set. Next, the distance among AH corresponding to each training sample is calculated based on the AH model. Meanwhile, the set-to-set distance is fed to the density peak algorithm to obtain the density of training samples. Finally, a decision-making value is applied to the density of each training sample to cleanse mislabelled samples within noisy training set. Experimental results on real HSI date sets demonstrate the superiority of the proposed method over several well-known training redefinition methods in terms of detection accuracy.

Hyperspectral image classification based on composite kernel relevance vector machine

This paper presents a composite kernel Relevance Vector Machine(RVM) algorithm, for enhanced classification accuracy of hyperspectral images. This paper constructs three forms of composite kernels based on properties of kernels. The spatial feature is extracted using multi-scale morphological method from the image after principal components transform. The final classification is achieved by composite kernel RVM classifier. The proposed approach is tested in experiments on AVIRIS data. Compared with spectral kernel RVM, the OA and Kappa coefficient of composite kernel RVM increased obviously. However, the training time dose not increased. Meanwhile, composite kernel RVM has ability to get high accuracy with relative small training set. The proposed method has practical use in hyperspectral imagery classification.

A semi-supervised learning method for hyperspectral imagery based on self-training and local-based affinity propagation

ABSTRACT In hyperspectral remote sensing, the classification of hyperspectral imagery is an important issue of concern. However, obtaining sufficient labelled samples for the classification is hard work and consumes enormous labour and material resources. To solve this problem, many scholars make their efforts to the semi-supervised learning method. The self-training model is a popular semi-supervised learning method, which trains a classifier using a few labelled samples and many unlabelled samples in an iterative manner. Nevertheless, the self-training model may provide misclassification as adding incorrect unlabelled samples. This paper proposes a novel semi-supervised learning method, which attempts to introduce the affinity propagation and spatial constraint to self-training model. In the proposed algorithm, a self-training model is employed to expand the labelled sample set. The affinity propagation is applied to select a group of reliable unlabelled samples. To further protect the reliability of unlabelled samples, the selection process of self-training model is constrained in the spatial local area of the labelled samples. In the iteration, the reliable unlabelled samples are selected and added to the labelled sample set. After that, the classifier is trained by the updated labelled samples. The algorithm finishes as the iterations exceed the predetermined value or the labelled sample set is unchanged. Different from the traditional self-training-based semi-supervised learning method, there are mainly three improvements. First, the selection of reliable unlabelled samples is conducted by the affinity propagation, which can effectively improve the reliability of the unlabelled samples. Second, the spatial constraint is introduced to restrict the selection process of self-training model, which can protect the reliability of unlabelled samples and alleviate the computational complexity. Finally, the spectral correlation angle is applied to construct the similarity matrix of affinity propagation, so that the algorithm is more suitable for hyperspectral imagery. A comparative analysis was conducted with classical self-training semi-supervised algorithms to test the effectiveness of the method. It was found that the proposed method provided higher overall accuracy, average accuracy and kappa coefficient (κ) than the competitors.

An Advanced Spectral–Spatial Classification Framework for Hyperspectral Imagery Based on DeepLab v3+

DeepLab v3+ neural network shows excellent performance in semantic segmentation. In this paper, we proposed a segmentation framework based on DeepLab v3+ neural network and applied it to the problem of hyperspectral imagery classification (HSIC). The dimensionality reduction of the hyperspectral image is performed using principal component analysis (PCA). DeepLab v3+ is used to extract spatial features, and those are fused with spectral features. A support vector machine (SVM) classifier is used for fitting and classification. Experimental results show that the framework proposed in this paper outperforms most traditional machine learning algorithms and deep-learning algorithms in hyperspectral imagery classification tasks.

Small sample classification for hyperspectral imagery using temporal convolution and attention mechanism

ABSTRACT Recently, deep learning models have achieved remarkable results in hyperspectral imagery (HSI) classification. However, most deep learning models tend to struggle due to the serious overfitting problem under the condition of small sample. For this purpose, we design an end-to-end classification framework and optimize it with the idea of meta-learning, to improve the accuracy and robustness of HSI small sample classification. Specifically, the embedding network based on 3D convolution is used to extract spatial-spectral features in HSI, the temporal convolution is adopted to analyse the abstract relationship between feature vectors by aggregating contextual information, and the attention mechanism is introduced to pinpoint the critical information conducive to classification, so as to further improve the classification accuracy. We adopt the typical mechanism of ‘meta-training + fine-tuning’ to optimize the designed model, enabling the model to acquire stronger generalization ability through a large number of different tasks. Experiments based on three public HSI data sets demonstrate that our method outperforms the regular deep learning models under the condition of small sample.

Classification of hyperspectral imagery using a fully complex-valued wavelet neural network with deep convolutional features

The number of spectral bands obtained by hyperspectral sensors improves the ability to distinguish physical objects and materials. But it also brings new challenges to image classification and analysis. In this study, a novel deep learning-based hybrid model called CNN-CVWNN is presented for the hyperspectral images classification (HSIs). The model uses a convolutional neural network (CNN) to extract multilayer image representation and uses the complex valued wavelet neural network (CVWNN) to classify the image using extracted features. The process steps of the proposed method are briefly as follows. First of all, the CNN algorithm has been applied to hyperspectral images. After this stage, efficient features have been obtained. These extracted features were then converted into a complex-valued number format using a novel random based transformation method. Thus, a novel complex-valued attribute set has been obtained for the HSI classification. The obtained features have been presented as input to the CVWNN algorithm. The hybrid method replaces real valued neural network inside CNN with CVWNN to enhance robustness and generalization of CNN. The experiments have been carried out on three data sets consisted of three popular hyperspectral airborne images. The developed method increases classification accuracy compared to other classification approaches.

Dual-Weighted Kernel Extreme Learning Machine for Hyperspectral Imagery Classification

Due to its excellent performance in high-dimensional space, the kernel extreme learning machine has been widely used in pattern recognition and machine learning fields. In this paper, we propose a dual-weighted kernel extreme learning machine for hyperspectral imagery classification. First, diverse spatial features are extracted by guided filtering. Then, the spatial features and spectral features are composited by a weighted kernel summation form. Finally, the weighted extreme learning machine is employed for the hyperspectral imagery classification task. This dual-weighted framework guarantees that the subtle spatial features are extracted, while the importance of minority samples is emphasized. Experiments carried on three public data sets demonstrate that the proposed dual-weighted kernel extreme learning machine (DW-KELM) performs better than other kernel methods, in terms of accuracy of classification, and can achieve satisfactory results.

Deep Mutual-Teaching for Hyperspectral Imagery Classification

Hyperspectral imagery (HSI) classification is a widely used method in remote sensing, which can provide accurate label information for each pixel. Though the classification accuracy is very high for many publicly available data sets in many research articles, they often exhibit much worse performance in practical applications. Because most of the articles adopt a random sampling strategy to select training and test samples from the same image, the high correlation between training and test samples will bring optimistic results. However, this strategy is not suitable for practical application. Because the training and test samples are collected from different locations in most situations, in this letter, the nonoverlapped sampling is adopted to reduce the correlation between training and test samples. Four key factors are presented to analyze the HSI classification; then, a new deep mutual-teaching method is proposed to classify the HSI. The experimental results show that the performance of the proposed method outperforms comparison methods.

Spectral Shift Mitigation for Cross-Scene Hyperspectral Imagery Classification

In cross-scene hyperspectral imagery (HSI) classification, labeled samples are only available in source scene, and how to properly reduce the spectral shift between source and target scenes is a matter of concern. In this article, we investigate this issue by considering the causes of the spectral shift and propose spectral shift mitigation (SSM) that includes amplitude shift mitigation (ASM) and adjacency effect mitigation (AEM). First, in ASM, the amplitude shift between source and target scenes is reduced by employing amplitude normalization on pixels of both source and target scenes. Then, in AEM, the spectral variation of target scene caused by adjacency effect is reduced by taking the weighted average spectral vector of surrounding pixels of a query pixel as the new spectral vector of the query pixel. Finally, a classifier trained by labeled samples from source scene is used for target scene. Superior classification performance on several cross-scene HSI data pairs demonstrates the effectiveness of the proposed SSM.