Improved filter algorithm using inequality fano to select bands for HSI classification

Recently, deep learning-based methods have drawn increasing attention in hyperspectral imagery (HSI) classification, due to their strong nonlinear mapping capability. However, these methods suffer from a time-consuming training process because of many network parameters. In this paper, the concept of broad learning is introduced into HSI classification. Firstly, to make full use of abundant spectral and spatial information of hyperspectral imagery, hierarchical guidance filtering is performing on the original HSI to get its spectral-spatial representation. Then, the class-probability structure is incorporated into the broad learning model to obtain a semi-supervised broad learning version, so that limited labeled samples and many unlabeled samples can be utilized simultaneously. Finally, the connecting weights of broad structure can be easily computed through the ridge regression approximation. Experimental results on three popular hyperspectral imagery datasets demonstrate that the proposed method can achieve better performance than deep learning-based methods and conventional classifiers.

Convolutional neural networks (CNNs) have been widely applied in hyperspectral imagery (HSI) classification. However, their classification performance might be limited by the scarcity of labeled data to be used for training and validation. In this paper, we propose a novel lightweight shuffled group convolutional neural network (abbreviated as SG-CNN) to achieve efficient training with a limited training dataset in HSI classification. SG-CNN consists of SG conv units that employ conventional and atrous convolution in different groups, followed by channel shuffle operation and shortcut connection. In this way, SG-CNNs have less trainable parameters, whilst they can still be accurately and efficiently trained with fewer labeled samples. Transfer learning between different HSI datasets is also applied on the SG-CNN to further improve the classification accuracy. To evaluate the effectiveness of SG-CNNs for HSI classification, experiments have been conducted on three public HSI datasets pretrained on HSIs from different sensors. SG-CNNs with different levels of complexity were tested, and their classification results were compared with fine-tuned ShuffleNet2, ResNeXt, and their original counterparts. The experimental results demonstrate that SG-CNNs can achieve competitive classification performance when the amount of labeled data for training is poor, as well as efficiently providing satisfying classification results.

The paradigm shift brought by deep learning in land cover object classification in hyperspectral images (HSIs) is undeniable, particularly in addressing the intricate 3D cube structure inherent in HSI data. Leveraging convolutional neural networks (CNNs), despite their architectural constraints, offers a promising solution for precise spectral data classification. However, challenges persist in object classification in hyperspectral imagery or hyperspectral image classification, including the curse of dimensionality, data redundancy, overfitting, and computational costs. To tackle these hurdles, we introduce the spectrally segmented-enhanced neural network (SENN), a novel model integrating segmentation-based, multi-layer CNNs, SVM classification, and spectrally segmented dimensionality reduction. SENN adeptly integrates spectral–spatial data and is particularly crucial for agricultural land classification. By strategically fusing CNNs and support vector machines (SVMs), SENN enhances class differentiation while mitigating overfitting through dropout and early stopping techniques. Our contributions extend to effective dimensionality reduction, precise CNN-based classification, and enhanced performance via CNN-SVM fusion. SENN harnesses spectral information to surmount challenges in “hyperspectral image classification in hyperspectral imagery”, marking a significant advancement in accuracy and efficiency within this domain.

Hyperspectral imagery (HSI) classification is essential for remote sensing analysis, utilizing various image bands. Convolutional Neural Networks (CNNs) are prevalent in deep learning for visual data processing, with recent applications in HSI classification primarily employing 2D and 3D CNNs. However, 3D CNNs demand significant computational resources due to their complexity. This paper introduces a two-branch spatial-spectral joint convolutional neural network (SSDB) leveraging an attention mechanism for HSI classification. SSDB effectively extracts spectral and spatial information while reducing model complexity, resulting in a lightweight alternative to 3D CNNs. In this comprehensive hyperspectral image (HSI) classification experiments utilizing the Indian Pines, Pavia University, and Salinas Scene datasets, we benchmarked our findings against leading-edge handcrafted and end-to-end deep learning methodologies, demonstrating exceptionally commendable performance with the SSDB approach.

In this manuscript, we propose a sparse self-representation (SSR) method to select a band subset in hyperspectral imagery (HSI) classification. The SSR method improves from multiple measurement vectors (MMV) with the measurement matrix equals to the observation matrix. The SSR regards that each band could be represented as a linear combination of the representatives of all bands and accordingly all band vectors are reconstructed by themselves accompanied with a sparse coefficient matrix. The SSR finds the indices of nonzero rows of the coefficient matrix by convex optimization to help to estimate the band subset that are constituted with representatives. Preliminary classification results on two open HSI datasets show that the SSR offers comparable results to the linear constrained minimum variance-based band correlation constraint (LCMV-BCC) method, and achieves better results than the maximum-variance principal component analysis (MVPCA) and sparse based band selection (SPaBS) methods.

Feature extraction is known to be an effective way in both reducing computational complexity and increasing accuracy in hyperspectral imagery (HSI) classification. In this article, we propose a novel HSI feature extraction based on joint adaptive structure density (JASD), which can make full use of the texture feature of feature level and effectively utilize the spatial structure feature of pixel level. Specifically, the proposed JASD method considers two weak assumptions when exploiting the structure information contained in the HSI: 1) close pixels in spectral space have the same label and 2) all the pixels in a subregion belong to the same class. The JASD method consists of the following steps. First, a principal component analysis method is applied to the original high-dimensional HSI to obtain low-dimensional features in order to promote the separability of pixels between different classes and improve the execution efficiency of the proposed method. Then, the local density peak-assisted $k$ nearest neighbor idea is introduced into oversegmentation-based regions to define the structural spatial information of the HSI and overcome the aforementioned weak assumptions. Meanwhile, recursive filtering in the transform domain is applied to the above low-dimensional HSI to preserve global edge details. Next, the global edge details and the local spatial structure feature are fused based on the stacking method. Finally, the composite hyperspectral feature is fed into the supervised classifier to obtain the classification results. Experimental results on three widely used real HSI data sets demonstrated that the proposed JASD method is superior to several comparative classification methods in terms of classification accuracy and computational burden.

Hyperspectral imagery has high spectral resolution, and spectrum of it has always been non-linear. The traditional classification methods cannot get better result when the number of samples is small. Combined with the theory of deep learning, a new semi-supervised method based on stacked autoencoders (SAE) is proposed for hyperspectral imagery classification. Firstly, with stacked autoencoders, a deep network model is constructed. Then, unsupervised pre-training is carried combined SOFTMAX classifier with unlabeled samples. Finally, fine-tuning the network model with small labeled samples, the SAE-based classifier can be got to learn implicit feature of spectrum of hyperspectral imagery and achieve classification of hyperspectral imagery. According to comparative experiments, the results indicate that the proposed method is effective to improve the hyperspectral imagery classification accuracy in case of small samples.

Hyperspectral image is a substitution of more than a hundred images, called bands, of the same region. They are taken at juxtaposed frequencies. The reference image of the region is called Ground Truth map (GT). the problematic is how to find the good bands to classify the pixels of regions; because the bands can be not only redundant, but a source of confusion, and decreasing so the accuracy of classification. Some methods use Mutual Information (MI) and threshold, to select relevant bands. Recently theres an algorithm selection based on mutual information, using bandwidth rejection and a threshold to control and eliminate redundancy. The band top ranking the MI is selected, and if its neighbors have sensibly the same MI with the GT, they will be considered redundant and so discarded. This is the most inconvenient of this method, because this avoids the advantage of hyperspectral images:: some precious information can be discarded. In this paper well make difference between useful and useless redundancy. A band contains useful redundancy if it contributes to decreasing error probability. According to this scheme, we introduce new algorithm using also mutual information, but it retains only the bands minimizing the error probability of classification. To control redundancy, we introduce a complementary threshold. So the good band candidate must contribute to decrease the last error probability augmented by the threshold. This process is a wrapper strategy; it gets high performance of classification accuracy but it is expensive than filter strategy.

Least squares unmixing methods are widely used to solve linear mixture problems for endmember abundance estimation in hyperspectral imagery. In this paper, a weighted least squares method is introduced as a generalization. When different weight matrix is used, a certain detector or classifier will be resulted. For accurate abundance fraction estimation, a constrained weighted least squares approach is developed by combining sum-to-one and nonnegativity constraints. The experimental results show that when a meaningful weight matrix is applied as a data pre-processing operator, the weighted least squares method will outperform ordinary least squares solution and the constrained methods will outperform unconstrained ones.

This paper proposes to combine locality-preserving projections (LPP) and sparse representation (SR) for hyperspectral image classification. The LPP is first used to reduce the dimensionality of all the training and testing data by finding the optimal linear approximations to the eigenfunctions of the Laplace Beltrami operator on the manifold, where the high-dimensional data lies. Then, SR codes the projected testing pixels as sparse linear combinations of all the training samples to classify the testing pixels by evaluating which class leads to the minimum approximation error. The integration of LPP and SR represents an innovative contribution to the literature. The proposed approach, called locality-preserving SR-based classification, addresses the imbalance between high dimensionality of hyperspectral data and the limited number of training samples. Experimental results on three real hyperspectral data sets demonstrate that the proposed approach outperforms the original counterpart, i.e., SR-based classification.

Recently, multiple kernel learning (MKL) methods have been developed to improve the flexibility of kernel-based learning machine. The MKL methods generally focus on determining key kernels to be preserved and their significance in optimal kernel combination. Unfortunately, computational demand of finding the optimal combination is prohibitive when the number of training samples and kernels increase rapidly, particularly for hyperspectral remote sensing data. In this paper, we address the MKL for classification in hyperspectral images by extracting the most variation from the space spanned by multiple kernels and propose a representative MKL (RMKL) algorithm. The core idea embedded in the algorithm is to determine the kernels to be preserved and their weights according to statistical significance instead of time-consuming search for optimal kernel combination. The noticeable merits of RMKL consist that it greatly reduces the computational load for searching optimal combination of basis kernels and has no limitation from strict selection of basis kernels like most MKL algorithms do; meanwhile, RMKL keeps excellent properties of MKL in terms of both good classification accuracy and interpretability. Experiments are conducted on different real hyperspectral data, and the corresponding experimental results show that RMKL algorithm provides the best performances to date among several the state-of-the-art algorithms while demonstrating satisfactory computational efficiency.

Band selection is an important step towards effective and efficient object classification in hyperspectral imagery. In this paper, we propose a semi-supervised learning method for band selection based on a sparse linear regression model. This model uses a least absolute shrinkage and selection operator to compute the regression coefficients from both labeled and unlabeled samples. These coefficients are then used to compute a contribution score for each band, which allows bands with high scores being selected for the testing step. During this process, unlabeled samples also contribute to the coefficients calculation. In order to propagate the labels to these samples, a hypergraph is first built to describe the relationship between labeled and unlabeled samples. This leads to an adjacency matrix whose entries are the sum of corresponding weights of hyperedges. Then matrix subspace learning method is used to estimate the labels of unlabeled samples. The proposed method is evaluated on the APHI dataset. Comparison with several baseline methods has shown the advantages of the proposed method on the pixel-level classification.

This paper proposes a novel multiscale graph cut based analysis framework for the supervised classification of hyperspectral imagery. This framework is aimed at obtaining accurate and reliable maps by properly considering the spatial-context information. It is made up of two main blocks: 1) a feature-extraction block exploits an object-oriented analysis and representation of hyperspectral imagery that is obtained by multiscale graph cut (MGC) based segmentation; 2) a classifier, based on support vector machines (SVMs), capable of analyzing hyperdimensional feature spaces. Experimental results confirm the effectiveness of the proposed system for the analysis of hyperspectral imagery.

Hyperspectral image (HSI) classification has become one of the most significant tasks in the field of hyperspectral analysis. However, classifying each pixel in HSI accurately is challenging due to the curse of dimensionality and limited training samples. In this paper, we present an HSI classification architecture called camera spectral response network (CSR-Net), which can learn the optimal camera spectral response (CSR) function for HSI classification problems and effectively reduce the spectral dimensions of HSI. Specifically, we design a convolutional layer to simulate the capturing process of cameras, which learns the optimal CSR function for HSI classification. Then, spectral and spatial features are further extracted by spectral and spatial attention modules. On one hand, the learned CSR can be implemented physically and directly used to capture scenes, which makes the image acquisition process more convenient. On the other hand, compared with ordinary HSIs, we only need images with far fewer bands, without sacrificing the classification precision and avoiding the curse of dimensionality. The experimental results of four popular public hyperspectral datasets show that our method, with only a few image bands, outperforms state-of-the-art HSI classification methods which utilize the full spectral bands of images.

Significant advances in compressive sensing and sparse signal encoding have provided a rich set of mathematical tools for signal analysis and representation. In addition to novel formulations for enabling sparse solutions to underdetermined systems, exciting progress has taken place in efficiently solving these problems from an optimization theoretic viewpoint. The focus of the wide body of literature in compressive sensing/sparse signal representations has however been on the problem of signal recovery from a small number of measurements (equivalently a sparse coefficient vector). This tutorial will discuss the design of sparse signal representations explicitly for the purposes of signal classification. The tutorial will focus on and build upon two significant recent advances. First, the work by Wright et al. which advocates the use of a dictionary (or basis) matrix comprising of class-specific training sub-dictionaries. In this framework, a test signal is modeled as a sparse linear combination of training vectors in the dictionary, sparsity being enforced by the assertion that only coefficients corresponding to one class (from which the test signal is drawn) ought to be active. The second set of ideas we leverage are recent key contributions in model-based compressive sensing where prior information or constraints on sparse coefficients are used to enhance signal recovery. These ideas will be combined towards the exposition of current trends: namely the development of class-specific priors or constraints to capture structure on sparse coefficients that helps explicitly distinguish between signal classes. In the second part of the tutorial, applications will be discussed including: 1.) structured sparsity for classification of medical imagery for diagnostics, 2.) low-rank approximation and sparse recovery for visual data reconstruction, and 3.) sparse representations for target detection and classification in hyperspectral imagery (guest speaker from the US Army Research Lab).