Hyperspectral imagery classification with cascaded support vector machines and multi-scale superpixel segmentation

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.

Recently, the superpixel segmentation is introduced into the hyperspectral image (HSI) classification to exploit the spatial information. However, the size of superpixel is hard to determine since small superpixels lack enough spatial information and large superpixels usually result in error segmentation. Therefore, a multiscale superpixels based sparse representation (MSSR) algorithm is proposed to utilize the spatial-spectral information of multiscale superpixels for the HSI classification. Specifically, multiscale superpixels of a HSI are generated firstly. Then, the joint sparse representation classification (JSRC) is used to obtain the class labels of superpixels of different scales. Finally, the majority voting is applied on the labels of different scales to create the final class label for each pixel. Experimental results show that the proposed MSSR algorithm outperforms several well-known classification algorithms.

In this paper, a soft classification algorithm based on composite kernels, which incorporate both spectral and spatial information, is proposed for hyperspectral image. Compared with hard classification, soft classification provides more information about the probabilities one pixel belongs to each class. To calculate these probabilities, the proposed algorithm uses Support Vector Machine (SVM), and it successfully converts SVM output values into probabilities, while at the same time integrates spatial and spectral information by composite kernels. To validate the proposed algorithm, experiments are conducted on hyperspectral images with 126 and 186 bands, and experimental results show that soft classification using SVM can yield better results compared with Maximum Likelihood Classifier (MLC),and the introduction of spectral-spatial kernels can greatly improve classification accuracies.

This paper aims to improve the accuracy and the efficiency of high resolution land cover mapping in urban area. To this end, an improved approach for classification of hyperspectral imagery is proposed and evaluated. This approach benefits from both inherent spectral and spatial information of an image. The weighted genetic (WG) algorithm is first used to obtain the subspace of hyperspectral data. The obtained features are then fed into the enhanced marker-based minimum spanning forest (EMSF) classification algorithm. In this algorithm, the markers are extracted from the classification maps obtained by both support vector machine and watershed segmentation algorithm classifiers. For this purpose, the class’s pixels with the largest population in the classification map are kept for each region of the segmentation map. Then, the most reliable classified pixels are chosen from among the exiting pixels as markers. To evaluate the efficiency of the proposed approach, three hyperspectral data sets acquired by ROSIS-03, Hymap and Hyper-Cam LWIR are used. Experimental results showed that the proposed WG–EMSF approach achieves approximately 9, 8 and 6% better overall accuracy than the original MSF-based algorithm for these data sets respectively.

High dimensionality of hyperspectral data and relatively limited training samples induce the Hughes phenomenon in hyperspectral image classification. To prevent this problem and decrease the computational cost, feature extraction often acts as pre-processing. In this paper, a subspace weighting kernel method combining clustering-based grouping is proposed for feature extraction in hyperspectral imagery classification. In the proposed method, spectral bands of hyperspectral data are firstly grouped into subspaces and a subspace-modulated kernel principal component analysis (SM-KPCA) is given for feature extraction, where the modulated kernel is determined by classification-oriented schemes. Support vector machine (SVM) classifier is performed on the extracted features to validate the performance. Experiments are conducted on real data and the results prove that the proposed SM-KPCA is effective on feature extraction for improving the accuracy of hyperspectral 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.

ABSTRACTHyperspectral image feature extraction generally does not consider the optimisation of structure element types, resulting in the loss of spatial correlation information. To solve this issue, this paper proposes a novel adaptive classification method (MPDTNC-SVM), in which spatial information is extracted by Morphological Profile Filter (MPF) and Domain Transform Normalised Convolution Filter (DTNCF). First, MPF extracts the spatial features on multiple principal component analysis (PCA) components of the hyperspectral image, and DTNCF works over all spectral bands to extract spatially correlated features. The two spatial features are then combined and fed into Support Vector Machine (SVM). Second, a two-step optimisation is implemented in the classification process. Specifically, the best structure of MPF is chosen for classification. Next, the optimal parameters of the structural elements are obtained through iterative classification optimisation, with the best classification performance produced in the process. Experimental results of actual hyperspectral images show that the proposed MPF and DTNCF with SVM (MPDTNC-SVM) method is superior to other classification methods, including Edge-Preserving Filter and Recursive Filter method, morphological feature-based methods, and the SVM methods with raw spectral features, reduced-dimensional and spatial-spectral information.

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.

Non-overlapping classification of hyperspectral imagery with superpixel segmentation

Recently, low-rank matrix recovery has been demonstrated to be an effective tool in hyperspectral images (HSIs) denoising. However, the previous low-rank matrix recovery method with a window of the fixed-shape cannot adaptively exploit spatial structure information and nonlocal similarity. In this paper, multiscale low-rank matrix recovery (MC-LRMR) is proposed to recover HSI corrupted by different kinds of noise. The proposed method contains three main steps. First, HSI is transformed by the principal component analysis (PCA) algorithm and multiscale superpixel segmentation is applied to the first principal component, to segment the HSI into non-overlapping homogeneous regions. Then, the mixed noises, including Gaussian noise, impulse noise, dead lines noise, stripes noise, are removed by the low-rank matrix recovery (LRMR) in a superpixel-by-superpixel manner. Finally, a fusion rule, i.e., average operator, is adopted to combine denoising results of various scales, to acquire a fused noise-free estimation. Experiments on simulated and real HSI data sets can demonstrate the effectiveness of the proposed method.

Recently, the sparse representation based classification methods have received particular attention in the classification of hyperspectral imagery. However, current sparse representation based classification models have not considered all the test pixels simultaneously. In this paper, we propose a hyperspectral classification method with spatial filtering and norm (SFL) that can deal with all the test pixels simultaneously. The norm regularization is used to extract relevant training samples among the whole training data set with joint sparsity. In addition, the norm loss function is adopted to make it robust for samples that deviate significantly from the rest of the samples. Moreover, to take the spatial information into consideration, a spatial filtering step is implemented where all the training and testing samples are spatially averaged with its nearest neighbors. Furthermore, the non-negative constraint is added to the sparse representation matrix motivated by hyperspectral unmixing. Finally, the alternating direction method of multipliers is used to solve SFL. Experiments on real hyperspectral images demonstrate that the proposed SFL method can obtain better classification performance than some other popular classifiers.

Superpixel segmentation aims at grouping discretizing pixels into high-level correlative units and reducing the complexity of subsequent tasks, e.g., saliency detection and object tracking. Existing superpixel segmentation algorithms mainly focus on maintaining the geometrical information, while neglecting the irregular structure of superpixels. In this paper, a superpixel segmentation method is proposed to generate approximately structural superpixels with sharp boundary adherence and comprehensive semantic information. The superpixel segmentation is formulated as a square-wise asymmetric partition problem, where the semantic perceptual superpixels are recorded in a square level to preserve abundant semantic information and save storage simultaneously. Moreover, in order to achieve regular-shape superpixel units to better adhere to image boundaries and contours, a combinatorial optimization strategy is devised to achieve an optimal combination of squares and isolated pixels. Experimental comparisons with some state-of-the-art superpixel segmentation methods on the public benchmarks demonstrate the effectiveness of the proposed method quantitatively and qualitatively. In addition, we have applied the method to brain tissue segmentation to illustrate superior performance.

The fusion of traditional HSI mainly involves the fusion of HSI with MSI or PAN. Nevertheless, the combination of the former is constrained by the inadequate spatial resolution of MSI, and therefore the enhancement of the spatial resolution of HSI is inadequate. The latter is constrained by the substantial disparity in spectral information between PAN and HSI, leading to pronounced distortion of high-resolution HSI. To address this issue, this paper proposes IFNPMH, which is an integrated fusion network for used for the fusion of HSI, MSI, and PAN. The IFNPMH algorithm is capable of efficiently extracting spatial information from MSI and PAN, as well as spectral information from HSI. It combines the spatial and spectral information to produce fused images that possess precise spectral information and distinct spatial details. Specifically, IFNPMH incorporates the CBAM attention mechanism module to enhance feature representation capabilities. Finally, this article proposes a loss function that combines spectral and spatial information loss in IFNPMH training. The effectiveness of the proposed IFNPMH is demonstrated by extensive testing in comparison to other sophisticated approaches. This model has shown promising performance in terms of qualitative visual effects and quantitative evaluation metrics.

Pansharpening is an important way of integrating spatial and spectral information in the field of remote sensing. This field uses the complementary and redundant information between multispectral (MS) images and panchromatic (PAN) images to obtain high spectral and high spatial resolution images. Various pansharpening methods have been introduced so far, each one attempting to provide a pansharpened image with the least distortion and maximum preservation of spectral and spatial information. Due to the importance of this issue, there should be methods and indices to evaluate the performance of different pansharpening algorithms and assess the quality of pansharpened images. In this paper, a segmentation-based method for assessing the quality of fused images is proposed. The advantage of this approach over pixel-based methods is that the pixel-based methods consider the fused images as a set of separate pixels while segmentation can take into account useful spatial information such as neighbourhoods, textures, etc. In the proposed method, by using k-means clustering algorithm, the reference and pansharpened images are segmented into areas with similar spectral and spatial features and the corresponding segments of the images are compared. This method is tested on three real data sets acquired by Pleiades, GeoEye-1, and QuickBird sensors. Experimental results demonstrate the effectiveness of the proposed method in evaluation of the quality of fused images.

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.