Lightweight Transformer with Transposed-attention for Hyperspectral Unmixing

Pixels in hyperspectral images are a mixing of source signals. Hyperspectral unmixing is an important issue in image processing. In this paper we consider a linear mixing model. We address the unmixing issue when some source signals are only present in few mixed pixels. We propose a new method based on Non-negative Matrix Factorization (NMF) with known endmembers' number. This method first estimates the abundant source signals. Then it detects pixels which contain the rare signals. Finally it processes those pixels to estimate the rare signals.

Hyperspectral Image (HSI) unmixing is a critical problem in remote sensing image processing. It aims to estimate the pure spectral signatures and their fractional abundances in each HSI pixel. Generally, most spectral-spatial unmixing methods do not exploit the contextual structure of HSI data cube. Therefore, in order to fully utilize the spatial information, we propose a new approach for spectral-spatial unmixing based on deep learning and Extended Morphological Profiles (EMPs). First, a latent representation from the original HSI data (spectral features) and the morphological profiles (spatial features) is obtained through the AutoEncoder (AE) network. Then, the estimation of both endmembers and abundance maps is achieved by the Convolutional AE (CAE). Experiments are performed using two HSI benchmarks which are Jasper Ridge and Samson. Compared to a spectral-spatial unmixing method, our proposed method shows significantly higher performance.

Hyperspectral imaging captures detailed spectral data for remote sensing. However, due to the limited spatial resolution of hyperspectral sensors, each pixel of a hyperspectral image (HSI) may contain information from multiple materials. Although the hyperspectral unmixing (HU) process involves estimating endmembers, identifying pure spectral components, and estimating pixel abundances, existing algorithms mostly focus on just one or two tasks. Blind source separation (BSS) based on nonnegative matrix factorization (NMF) algorithms identify endmembers and their abundances at each pixel of HSI simultaneously. Although they perform well, the factorization results are unstable, require high computational costs, and are difficult to interpret from the original HSI. CUR matrix decomposition selects specific columns and rows from a dataset to represent it as a product of three small submatrices, resulting in interpretable low-rank factorization. In this paper, we propose a new blind HU framework based on CUR factorization called CUR-HU that performs the entire HU process by exploiting the low-rank structure of given HSIs. CUR-HU incorporates several techniques to perform the HU process with a performance comparable to state-of-the-art methods but with higher computational efficiency. We adopt a deterministic sampling method to select the most informative pixels and spectrum components in HSIs. We use an incremental QR decomposition method to reduce computation complexity and estimate the number of endmembers. Various experiments on synthetic and real HSIs are conducted to evaluate the performance of CUR-HU. CUR-HU performs comparably to state-of-the-art methods for estimating the number of endmembers and abundance maps, but it outperforms other methods for estimating the endmembers and the computational efficiency. It has a 9.4 to 249.5 times speedup over different methods for different real HSIs.

Hyperspectral unmixing is the procedure by which the end component elements are calculated and their fractional abundances are found in each pixel in hyperspectral images. Sometimes several types of sound harm a hyperspectral photo. In a general scenario that takes mixed noise into consideration, this study addresses the hyperspectral non-mixing problem. Gaussian and sparse noises are expressly taken into account in the unmixing model. The problem of unmixing was formulated to use the combined shortage of abundance maps. For the modelling of flatness diagrams, a complete variation-based regularisation has also been used. For the solution of an algorithm for the optimization problem, the split-Bregman technique was used. The advantages of the method proposed are revealed by detailed preliminary findings on both real and synthetic images.

Sparse unmixing has attracted much attention in recent years. It aims at estimating the fractional abundances of pure spectral signatures in mixed pixels in hyperspectral images. To exploit spatial-contextual information present in the scene, the total variation (TV) regularization is incorporated into the sparse unmixing formulation, promoting adjacent pixels having similar not only endmembers but also fractional abundances, and thus having similar structural sparsity. It is therefore hoped to impose joint sparsity, instead of classic single sparsity, on these adjacent pixels to further improve the unmixing performance. To this end, we include the joint-sparse-blocks regression into the TV spatial regularization framework and present a new unmixing algorithm, termed joint-sparse-blocks unmixing via variable splitting augmented Lagrangian and total variation (JSBUnSAL-TV). In particular, a reweighting strategy is utilized to enhance sparsity along lines within each block. Simulated and real-data experiments show the advantages of the proposed algorithm.

This paper investigates the use of Gaussian processes to detect non-linearly mixed pixels in hyperspectral images. The proposed technique is independent of nonlinear mixing mechanism, and therefore is not restricted to any prescribed nonlinear mixing model. The observed reflectances are estimated using both the least squares method and a Gaussian process. The fitting errors of the two approaches are combined in a test statistics for which it is possible to estimate a detection threshold given a required probability of false alarm. The proposed detector is compared to a robust nonlinearity detector recently proposed using synthetic data and is shown to provide a better detection performance. The new detector is also tested on a real hyperspectral image.

Abstract. Hyperspectral image enhancement has been a concern for the remote sensing society for detailed end member detection. Hyperspectral remote sensor collects images in hundreds of narrow, continuous spectral channels, whereas multispectral remote sensor collects images in relatively broader wavelength bands. However, the spatial resolution of the hyperspectral sensor image is comparatively lower than that of the multispectral. As a result, spectral signatures from different end members originate within a pixel, known as mixed pixels. This paper presents an approach for obtaining an image which has the spatial resolution of the multispectral image and spectral resolution of the hyperspectral image, by fusion of hyperspectral and multispectral image. The proposed methodology also addresses the band remapping problem, which arises due to different regions of spectral coverage by multispectral and hyperspectral images. Therefore we apply algorithms to restore the spatial information of the hyperspectral image by fusing hyperspectral bands with only those bands which come under each multispectral band range. The proposed methodology is applied over Henry Island, of the Sunderban eco-geographic province. The data is collected by the Hyperion hyperspectral sensor and LISS IV multispectral sensor.

The normal compositional model (NCM) is a well-known and powerful model in hyperspectral unmixing which represents endmembers as independent Gaussian vectors to capture endmember variability. However, the assumption of independent endmembers diminishes the model accuracy because the high degree of correlation between endmembers of a scene and identical sources of variability demonstrate that the endmembers are dependent. This paper proposes a new hyperspectral unmixing algorithm which represents endmembers using dependent Gaussian vectors to estimate abundance fractions. To overcome the higher complexity caused by dependence assumption, this algorithm introduces new independent Gaussian vectors named Base Vectors to represent different endmembers by a weighted linear combination. Also, the proposed unmixing algorithm uses maximum likelihood method to estimate weight coefficients of Base Vectors which are used to represent mixed pixel. Finally, abundance estimation can be done using the new representation for endmembers and mixed pixel. The proposed algorithm is evaluated and compared with other state-of-the-art unmixing algorithms using simulated and real hyperspectral images. Experimental results demonstrate that the proposed unmixing algorithm can unmix pixels composed of correlated endmembers in hyperspectral images in the presence of spectral variability more accurately than previous methods.

Hyperspectral unmixing, aiming to estimate the fractional abundances of pure spectral signatures in each mixed pixel, has attracted considerable attention in analyzing hyperspectral images. Plenty of sparse unmixing methods have been proposed in the literature that achieved promising performance. However, many of these methods overlook the latent geometrical structure of the hyperspectral data which limit their performance to some extent. To address this issue, a double reweighted sparse and graph regularized unmixing method is proposed in this paper. Specifically, a graph regularizer is employed to capture the correlation information between abundance vectors, which makes use of the property that similar pixels in a spectral neighborhood have higher probability to share similar abundances. In this way, the latent geometrical structure of the hyperspectral data can be transferred to the abundance space. In addition, a double weighted sparse regularizer is used to enhance the sparsity of endmembers and the fractional abundance maps, where one weight is introduced to promote the sparsity of endmembers as a hyperspectral image typically contains fewer endmembers compared to the overcomplete spectral library and the other weight is exploited to improve the sparsity of the abundance matrix. The weights of the double weighted sparse regularizer used for the next iteration are adaptively computed from the current abundance matrix. The experimental results on synthetic and real hyperspectral data demonstrate the superiority of our method compared with some state-of-the-art approaches.

<p>Hyperspectral remote sensors collect image data for a large number of narrow, adjacent spectral bands. Every pixel in hyperspectral image involves a continuous spectrum that is used to classify the objects with great detail and precision. This paper presents hyperspectral image classification mechanism using genetic algorithm with empirical mode decomposition and image fusion used in preprocessing stage. 2-D Empirical mode decomposition method is used to remove any noisy components in each band of the hyperspectral data. After filtering, image fusion is performed on the hyperspectral bands to selectively merge the maximum possible features from the source images to form a single image. This fused image is classified using genetic algorithm. Different indices, such as K-means (KMI), Davies-Bouldin Index (DBI), and Xie-Beni Index (XBI) are used as objective functions. This method increases classification accuracy of hyperspectral image.</p>

Color is a key indicator for evaluating the quality of tea during processing; various processing procedures can significantly affect the content of fat-soluble pigments of tea, which in turn affects the color and quality of finished tea. Therefore, there is an urgent demand for the fast, non-destructive detection of pigments of stacked tea during processing. This paper presents the use of hyperspectral imaging technology (HSI), combined with machine learning algorithms, to detect chlorophyll a, chlorophyll b, and carotenoids in stacked matcha tea during processing. Firstly, a quantitative relationship between HSI data of tea and their pigment contents was developed based on regression analysis, and the results showed that exceptional prediction performance was achieved by the partial least squares regression (PLSR) algorithm combined with the feature band algorithm of competitive adaptive reweighting (CARS), and the Rp2 values of detection models of chlorophyll a, chlorophyll b and carotenoids were 0.90465, 0.92068 and 0.62666, respectively. Then, these quantitative detection models were extended to each pixel in hyperspectral images, achieving point-by-point prediction of pigment components, so the distribution of pigments of stacked tea leaves during processing procedures was successfully visualized on the processing line in situ. By integrating a hyperspectral imaging system into the real-world environment, operators can monitor pigment levels in real time and thus dynamically adjust processing parameters based on real-time data. This study enhances pigment detection efficiency in tea processing, supports process optimization, and aids in quality control.

Nonnegative matrix factorization(NMF) has been applied to hyperspectral unmixing in recent years. Different constraints based on geometrical or statistical properties of endmember and abundance are incorporated into NMF model to improve unmixing result. In this paper, a new regularizer based on spectral cluster information is proposed to strengthen the constrained relationship between original image and abundance maps. The new algorithm makes abundances of similar pixels close and abundances of dissimilar pixels be separated completely. Additionally, L 1/2 sparsity constraint is adopted to make the solutions sparse. Comparative results on real and synthetic hyperspectral datasets prove our proposed method could improve the hyperspectral unmixing accuracy.

A rapid and nondestructive method to determine the distribution map of protein, carbohydrate and sialic acid on Edible bird's nest by hyper-spectral imaging and chemometrics.

Hyperspectral unmixing is an important image interpre-tation technique that aims to estimate the pure constituent materials (endmembers) and their corresponding fractional abundances in each mixed pixel. Nonnegative matrix factorization (NMF) has attracted a lot of attention because of its ability to solve mixed pixel scenarios. The sparse NMF method achieves better unmixing results thanks to its full use of the sparse characteristics of the data. However, most existing sparse NMF unmixing techniques lack the consid-eration of spatial information. In fact, hyperspectral images contain intrinsic geometric information as well as rich spatial information. In this paper, a spatial graph regularized nonneg-ative matrix factorization unmixing framework (SGNMF) is established. For the proposed SGNMF, on the one hand, the graph regularization is introduced to characterize the latent manifold structure of the data, and on the other hand, the spatial weighting factor is used to mine the spatial correlation between pixels. The optimization problem of the SGNMF model can be solved by a multiplicative iterative rule. Exper-imental results on synthetic data sets indicate that the newly proposed SGNMF method is able to produce better results than other advanced spectral unmixing algorithms.

Spectral imaging including machine vision and multispectral imaging can provide a rapid, nondestructive means to assess agricultural commodities for their quality and safety for human consumption. These nondestructive techniques generally rely on reflectance measurements; the most commonly used spectral regions range through the visible (Vis) to the near infrared (NIR). Another optical sensing technique is based on fluorescence that is generally regarded as more sensitive optical sensing tools compared to reflectance techniques. Recently, we developed two imaging systems for food safety research; a laboratory-based hyperspectral imaging system capable of both reflectance and fluorescence measurements and a multispectral laser induced fluorescence imaging system. Because many contiguous spectral bands are acquired with each hyperspectral image pixel, the resulting data quantity makes uses of hyperspectral imaging spectrometer for on-line food safety inspection impractical. Instead, the spectral and spatial information is used to determine several optimal wavebands that can be implemented to a multispectral imaging system. We present a systematic method for using hyperspectral image data to identify few wavebands to be used in multispectral detection systems. Also presented is a two-band ratio as a multispectral fusion method along with automated detection for animal feces contamination on apples. The fluorescence imaging techniques demonstrated that even minute animal feces, not clearly visible to human eye, could be detected. Practical considerations for development of multispectral fluorescence imaging systems for on-line uses are also discussed.