Hyperspectral image classification by Multi-Scale Vector Tunnel

Hyperspectral image classification by a variable interval spectral average and spectral curve matching combined algorithm

ABSTRACTHyperspectral images have acquired an increasing interest in the scientific and civil community to detect or classify materials by their spectral signature in remote sensing applications. However, spectral signatures can vary because of lighting or atmospheric conditions. This phenomenon is known as spectral variability, and ignoring this variability is an important source of error in the analysis of hyperspectral images. Therefore, this work focuses on analysing different causes and effects of the spectral variability in hyperspectral images, by studying the latest advances in this area, including approaches that do not address the phenomenon and methods that address spectral variability. Experiments are executed with five different approaches to compare their accuracy in the estimation of abundances in synthetic data generated with spectral variability. Specifically, a case study is developed on hyperspectral data of oil palm crops in Colombia, which exhibit high spectral variability. The results of identification of stressed oil palms are compared by mapping the abundances of three classes (diseased oil palm, healthy oil palm, and background), evaluating the performance of the different approaches in terms of the root mean squared error (RMSE), reconstruction error (RE), and spectral angle mapping (SAM). Results show that the coordinate descent algorithm-mismodelling effects (CDA-ME) present the best performance, outperforming in up to 70% the other approaches with and without spectral variability since it can jointly model changes of illumination, effects of variability and outlier values present in the hyperspectral images.

Applied in Djebel Meni (Northwestern of Algeria), this research highlights the results obtained from the supervised classification using the Spectral Angle Mapper (SAM) algorithm, through introducing the spectral signatures of illite, kaolinite, and montmorillonite, via Jet Propulsion Laboratory (JPL) spectral library. These results were compared to the ones of the SAM classification, which use spectral signatures obtained by the Sequential Maximum Angle Convex Cone (SMACC) endmembers extraction algorithm. This implies the ability to detect and identify any object present on the Earth’s surface, whether its nature is mineral, vegetal, or human made, from hyperspectral imaging. By extracting the spectral signatures with the SMACC algorithm and matching them to the current signatures of JPL spectral library, comparing spectral signatures with another is not an easy task. Indeed, for a better comparison and a more appropriate interpretation in the use of the SAM classification, the results obtained were very relatively convincing because, regarding very strong similarities. It appears also that the signatures extracted with SMACC occupy the same areas as those of the JPL spectral library. This method of detection and identification of any present object on the Earth’s surface is rather conclusive.

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.

A hyperspectral image can be considered as an image cube where the third dimension is the spectral domain represented by hundreds of spectral wavelengths. As a result, a hyperspectral image pixel is actually a column vector with dimension equal to the number of spectral bands and contains valuable spectral information that can be used to account for pixel variability, similarity and discrimination. We present a new hyperspectral measure, the spectral information measure (SIM), to describe spectral variability and two criteria, spectral information divergence and spectral discriminatory probability for spectral similarity and discrimination, respectively. The spectral information measure is an information-theoretic measure which treats each pixel as a random variable using its spectral signature histogram as the desired probability distribution. Spectral information divergence (SID) compares the similarity between two pixels by measuring the probabilistic discrepancy between two corresponding spectral signatures. The spectral discriminatory probability calculates spectral probabilities of a spectral database (library) relative to a pixel to be identified so as to achieve material identification. In order to compare the discriminatory power of one spectral measure relative to another, a criterion is also introduced for performance evaluation, which is based on the power of discriminating one pixel from another relative to a reference pixel. The experimental results demonstrate that the new hyperspectral measure can characterize spectral variability more effectively than the commonly used spectral angle mapper (SAM).

This paper presents a new algorithm for hyperspectral image analysis using spectral-angle based support vector clustering (SVC) and principal component analysis (PCA). In the classical approach to hyper-spectral dimensionality reduction based on principal component analysis (PCA), no meaning or behavior of the spectrum is considered and results are influenced by majority components in the scene. A spectral angle based classification before dimensionality reduction is a possible solution to this problem. Clustering based on support vectors using spectral based kernels is proposed in this work, which is found to generate good results in hyperspectral image classification. The algorithm is tested with two hyperspectral image data sets of 210 bands each, which are taken with hyper-spectral digital imagery collection experiment (HYDICE) air-borne sensors. A comparative study of the proposed algorithm and other two conventional algorithms (PCA alone and PCA with spectral angle mapping (SAM)) is also done

Fuel types are one of the key variables that drive wildfire ignition and propagation. A new method is proposed to automatically classify and map fuel types from LiDAR data. The 3-dimensional Discrete Anisotropic Radiative Transfer (DART) model generated a fuel type LiDAR signature library. These simulations provided reference endmembers and additional data to demonstrate the feasibility to classify fuel types using spectral matching algorithms, like multiple endmember spectral mixture analysis (MESMA) and spectral angle mapper (SAM). When choosing a single endmember per fuel type, MESMA outperformed SAM with 63.3% and 48.9% agreement, respectively. Multiple endmembers per fuel type improved the classification results to 85.3% in SAM and 86.5% in MESMA. Endmembers need to identify different scan angles that account for the variability in height and number of trees for better results. Contrary to empirical models, a fuel type LiDAR signature library provides a comprehensive suite of solutions to classify fuel types from LiDAR data that is less study site dependent and applicable to multiple sensors.

Hyperspectral Image (HSI) classification plays a key role remote sensing field. Recently, deep learning has demonstrated its effectiveness in HSI Classification field. This paper presents a spectral-spatial HSI classification technique established on the deep learning based deep belief network (DBN) for deep and abstract feature extraction and adaptive boundary adjustment based segmentation. Proposed approach focuses on integrating the deep learning based spectral features and segmentation based spatial features into a framework for improved performance. Specifically, first the deep DBN model is exploited as a spectral feature extraction based classifier to extract the deep spectral features. Second, spatial contextual features are obtained by utilizing effective adaptive boundary adjustment based segmentation technique. Finally, maximum voting based criteria is operated to integrate the results of extracted spectral and spatial information for improved HSI classification. In general, exploiting spectral features from DBN process and spatial features from segmentation and integration of spectral and spatial information by maximum voting based criteria, has a substantial effect on the performance of HSI classification. Experimental performance on real and widely used hyperspectral data sets with different contexts and resolutions demonstrates the accuracy of the proposed technique and performance is comparable to several recently proposed HSI classification techniques.

Convolutional neural networks (CNNs) and graph convolutional networks (GCNs) have made considerable advances in hyperspectral image (HSI) classification. However, most CNN-based methods learn features at a single-scale in HSI data, which may be insufficient for multi-scale feature extraction in complex data scenes. To learn the relations among samples in non-grid data, GCNs are employed and combined with CNNs to process HSIs. Nevertheless, most methods based on CNN-GCN may overlook the integration of pixel-wise spectral signatures. In this paper, we propose a pyramid cascaded convolutional neural network with graph convolution (PCCGC) for hyperspectral image classification. It mainly comprises CNN-based and GCN-based subnetworks. Specifically, in the CNN-based subnetwork, a pyramid residual cascaded module and a pyramid convolution cascaded module are employed to extract multiscale spectral and spatial features separately, which can enhance the robustness of the proposed model. Furthermore, an adaptive feature-weighted fusion strategy is utilized to adaptively fuse multiscale spectral and spatial features. In the GCN-based subnetwork, a band selection network (BSNet) is used to learn the spectral signatures in the HSI using nonlinear inter-band dependencies. Then, the spectral-enhanced GCN module is utilized to extract and enhance the important features in the spectral matrix. Subsequently, a mutual-cooperative attention mechanism is constructed to align the spectral signatures between BSNet-based matrix with the spectral-enhanced GCN-based matrix for spectral signature integration. Abundant experiments performed on four widely used real HSI datasets show that our model achieves higher classification accuracy than the fourteen other comparative methods, which shows the superior classification performance of PCCGC over the state-of-the-art methods.

Most hyperspectral image (HSI) classification methods assume that all classes in the test set are present during training. However, in real-world applications, acquiring labeled training samples is challenging. As a result, it is difficult for the training dataset to cover all possible land cover types, leading to the generalized zero-shot learning (GZSL) problem. Recently, vision-language models (VLMs) have provided rich semantic priors for land cover classes, offering promising potential for GZSL. However, two fundamental gaps hinder their application to HSI classification: the task paradigm gap, arising from the difference between image-level VLMs and the pixel-level HSI classification task; and the knowledge gap, due to the inconsistency between VLM features and HSI spectral–spatial representations. To bridge both gaps, a novel framework leveraging VLM semantic priors for GZSL in HSI classification is proposed, primarily using pseudo-labeling technique to provide knowledge for unseen classes. Specifically, a pseudo-label generation and enhancement module enables a paradigm transition from image-level understanding to pixel-level classification by incorporating HSI’s spatial information. A pseudo-label correction module then refines noisy labels using spectral cues to address the knowledge gap. Finally, a global learning strategy integrates pseudo-label distillation, supervised learning, and feature regularization to classify seen classes while enabling generalization to unseen ones. Experiments on benchmark HSI datasets demonstrate the proposed method’s superiority in generalized zero-shot classification. This work highlights the potential of VLMs in advancing HSI classification in practical applications.

BackgroundIn the field of earth observation, hyperspectral detector systems allow precise target detections of surface components from remote sensing platforms. This enables specific land covers to be identified without the need to physically travel to the areas examined. In the medical field, efforts are underway to develop optical technologies that detect altering tissue surfaces without the necessity to perform an excisional biopsy. With the establishment of expedient classification procedures, hyperspectral imaging may provide a non-invasive diagnostic method that allows determination of pathological tissue with high reliability. In this study, we examined the performance of a hyperspectral hybrid method classification for the automatic detection of altered mucosa of the human larynx.Materials and methodsHyperspectral Imaging was performed in vivo and 30 bands from 390 to 680 nm for 5 cases of laryngeal disorders (2x hemorrhagic polyp, 3x leukoplakia) were obtained. Image stacks were processed with unsupervised clustering (linear spectral unmixing), spectral signatures were extracted from unlabeled cluster maps and subsequently applied as end-members for supervised classification (spectral angle mapper) of further medical cases with identical diagnosis.ResultsLinear spectral unmixing clearly highlighted altered mucosa as single spectral clusters in all cases. Matching classes were identified, and extracted spectral signatures could readily be applied for supervised classifications. Automatic target detection performed well, as the considered classes showed notable correspondence with pathological tissue locations.ConclusionsUsing hyperspectral classification procedures derived from remote sensing applications for diagnostic purposes can create concrete benefits for the medical field. The approach shows that it would be rewarding to collect spectral signatures from histologically different lesions of laryngeal disorders in order to build up a spectral library and to prospectively allow non-invasive optical biopsies.

Hyperspectral image (HSI) classification has become a core task in hyperspectral remote sensing interpretation, with deep learning dominating due to its ability to learn hierarchical features without manual engineering. As the model complexity has grown, manual design limitations have prompted a shift to automated approaches such as differentiable architecture search (DARTS), where the architectures are optimized for greater accuracy and efficiency. However, applying gradient-based neural architecture search (NAS) methods directly to hyperspectral classification presents several challenges. Regarding search space design, there is a lack of lightweight operators that can mitigate the spectral variability, spatial heterogeneity, and scale differences inherent in hyperspectral imagery. In terms of search strategy, the traditional DARTS approach directly derives the topology from operation weights, which can lead to suboptimal topological structures, and thus affects the performance of the network in HSI classification. In this article, to address these issues, we propose L3M, which is a lightweight multiscale and multiattention HSI classification network based on multistage search. The proposed approach introduces a novel lightweight operator to address the spectral variability, spatial heterogeneity, and scale differences in HSIs. The operation search and topology search are also decomposed into a multistage process to prevent a suboptimal network by searching for and determining the topological order of the candidate operations in a predefined operation space. L3M was validated on four public datasets, where the proposed model demonstrated a superior classification performance, compared to other lightweight models, while maintaining a low parameter count, low model complexity, and high inference speed.

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.

In recent studies, superpixel segmentation has been integrated into hyperspectral (HS) image classification methods. However, the existing superpixel-based classification methods usually suffer from two serious problems. First, the accuracy and efficiency of current superpixel segmentation approaches cannot meet the demands of practical applications for HS images; second, conventional superpixel-based classification methods generally consider each generated superpixel as a unit for the image classification, which may help to reduce the computing time but result in a significant decrease of the classification accuracy. To solve the problems, we propose a fast region growing based superpixel segmentation (FRGSS) algorithm and a novel texture-adaptive superpixel integration strategy (TASIS) for the HS image classification. Experimental results on real Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) HS images demonstrate that the proposed FRGSS outperforms the state-of-the-art superpixel algorithm. In addition, the superiority of the TASIS is verified compared to the pixel-wise and the conventional superpixel-based classification methods.

Target detection is one of research focuses for precision chemical application. This study developed a method to identify seedling cabbages and weeds using hyperspectral imaging. In processing the image data with ENVI software, after dimension reduction, noise reduction, de-correlation for high-dimensional data, and selection of the region of interest, the SAM (Spectral Angle Mapping) model was built for automatic identification of cabbages and weeds. With the HSI (Hyper Spectral Imaging) Analyzer, the training pixels were used to calculate the average spectrum as the standard spectrum. The parameters of the SAM model, which had the best classification results with 3-point smoothing, zero-order derivative, and 6-degrees spectral angle, was determined to achieve the accurate identification of the background, weeds, and cabbages. In comparison, the SAM model can completely separate the plants from the soil background but not perfect for weeds to be separated from the cabbages. In conclusion, the SAM classification model with the HSI analyzer could completely distinguish weeds from background and cabbages. Keywords: hyperspectral imaging, weed identification, cabbage, seedlings DOI: 10.3965/j.ijabe.20150805.1492 Citation: Deng W, Huang Y B, Zhao C J, Wang X. Identification of seedling cabbages and weeds using hyperspectral imaging. Int J Agric & Biol Eng, 2015; 8(5): 65－72.