Search

Hyperspectral unmixing decomposes a hyperspectral imagery (HSI) into a number of constituent materials and associated proportions. Recently, nonnegative tensor factorization (NTF)-based methods have been proposed for hyperspectral unmixing thanks to their capability in representing an HSI without any information loss. However, tensor factorization-based HSI processing approaches often suffer from low-signal-to-noise ratio condition of HSI and nonuniqueness of the solution. This problem can be effectively alleviated by introducing various spatial constraints into tensor factorization to suppress the noise and decrease the number of extreme, stationary, and saddle points. On the other hand, total variation (TV) adaptively promotes piecewise smoothness while preserving edges. In this paper, we propose a TV regularized matrix–vector NTF method. It takes advantage of tensor factorization in preserving global spectral–spatial information and the merits of TV in exploiting local spatial information, thus generating smooth abundance maps with preserved edges. Experimental results on synthetic and real-world data show that the proposed method outperforms the state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Many spectral unmixing approaches ranging from geometry, algebra to statistics have been proposed, in which nonnegative matrix factorization (NMF)-based ones form an important family. The original NMF-based unmixing algorithm loses the spectral and spatial information between mixed pixels when stacking the spectral responses of the pixels into an observed matrix. Therefore, various constrained NMF methods are developed to impose spectral structure, spatial structure, and spectral-spatial joint structure into NMF to enforce the estimated endmembers and abundances preserve these structures. Compared with matrix format, the third-order tensor is more natural to represent a hyperspectral data cube as a whole, by which the intrinsic structure of hyperspectral imagery can be losslessly retained. Extended from NMF-based methods, a matrix-vector nonnegative tensor factorization (NTF) model is proposed in this paper for spectral unmixing. Different from widely used tensor factorization models, such as canonical polyadic decomposition CPD) and Tucker decomposition, the proposed method is derived from block term decomposition, which is a combination of CPD and Tucker decomposition. This leads to a more flexible frame to model various application-dependent problems. The matrix-vector NTF decomposes a third-order tensor into the sum of several component tensors, with each component tensor being the outer product of a vector (endmember) and a matrix (corresponding abundances). From a formal perspective, this tensor decomposition is consistent with linear spectral mixture model. From an informative perspective, the structures within spatial domain, within spectral domain, and cross spectral-spatial domain are retreated interdependently. Experiments demonstrate that the proposed method has outperformed several state-of-the-art NMF-based unmixing methods. [ABSTRACT FROM AUTHOR]

As single-pixel imaging technology is highly sensitive and low-cost, a large number of hyperspectral imaging methods based on single-pixel have been proposed. Although the image of each band can be reconstructed independently, the parameters that need to be computed for reconstructing hyperspectral images will increase in parallel with the number of bands. Therefore, in this paper, we propose a redundant compressed single-pixel hyperspectral imaging system (RCSHI), which takes full advantage of the spectral redundancy of hyperspectral images by jointly reconstructing images from different bands and compressing the spectral dimension of the data. RCSHI reduces the amount of redundant hyperspectral information stacking and the number of parameters that need to be computed to reconstruct hyperspectral images. In addition, we use progressive fusion enhancement to recover the compressed spectral dimension and enhance the spatial dimension of the images. Experimental results show that RCSHI is able to acquire hyperspectral data with high spatial and spectral resolution and performs well on various datasets and actual scenarios, especially at lower compressed sampling rates. [ABSTRACT FROM AUTHOR]

• We design an undersampling rapid coded aperture spectrometer system. This system refers to the spatial information of RGB to select the sampling order and uses the DCT matrix as the coding patterns. The designed imaging system can better address the contradiction between image resolution and imaging speed to a certain extent, so as to achieve high speed and high resolution hyperspectral imaging. • A spectral image enhancement system is designed to improve the image spatial detail recognition capability that consists of a CMMI information fusion function and a PMS-net-based feature extraction module. By improving accuracy at a lower cost, the network structure is made simpler, which improves the generalization and robustness of the network to some extent. • Based on the fast conversion of the micromirror unit in DMD, we built a DMD-based hyperspectral imaging system. We obtained real hyperspectral datasets in reality by this system, and verified that our proposed algorithm has good performance on real spectral datasets. In recent years, Coded Aperture Snapshot Spectroscopic Imaging System (CASSI) has attracted more and more attention. However, in the case of undersampling, the effect of the reconstructed image is often unsatisfactory. For spectral imaging system, the slower the imaging speed, the higher the image resolution of hyperspectral images, and vice versa. To solve this problem, in this paper, we design an undersampling rapid coded aperture spectrometer system. Based on the analysis of the internal relationship between coded spectral imaging and single pixel imaging, the order of the discrete cosine transform (DCT) coding patterns is selected by using the energy concentration property of the DCT and the rich spatial information contained in the RGB images. In addition, we built a spectral image enhancement network to improve the quality of the reconstructed images. At the sampling rate of 0.0625, the PSNR, SSIM and SAM of the spectral images reconstructed based on our system are 43.8341, 0.9869 and 2.0862, respectively, the reconstruction effect of this method is significantly superior to other methods, and this method provides an idea to address the contradiction between image resolution and imaging speed to a certain extent, so as to achieve high speed and high resolution hyperspectral imaging. [ABSTRACT FROM AUTHOR]

• In existing multiplexed spectral imaging methods based on matrix operations (especially HTS), the reconstructed results are similar to miscalculations in the case of under-sampling. In contrast, our FCTS based on frequency domain calculation discards only the high-frequency components and has higher coding efficiency and physical interpretability. • Compared with existing snapshot systems (especially CASSI), FCTS has no limitation on the number of dispersion channels and can use high dispersion elements such as gratings according to practical needs. The experimental system in this paper uses an Amici prism with a spectral resolution of about 4 nm, which is tested to be lower than 1 nm if a 300-line grating is used, and has an unparalleled advantage in spectral resolution for snapshot systems. • Real systems and simulations show that the FCTS frequency domain calculation method is less affected by optical misalignment, diffraction and distortion in optical systems. It is more robust to optical distortions. • Because of the Fourier transform process of FCTS under-sampling we only take the low frequency part of the Fourier spectrum. So, there are some high frequency details missing in the reconstruction. We can also use the common super resolution reconstruction or image enhancement methods to complete the image details to improve the image quality. We consider using Fourier coded aperture transform to solve the acquisition and reconstruction problem of three-dimensional (3-D) hyperspectral imaging. In recent years, Coded aperture snapshot spectral imaging (CASSI) system has received increasing attention. However, reconstruction errors are unavoidable due to its ill-posed problems and the number of dispersion channels is limited. The Hadamard transform spectral (HTS) imaging method has no limitation on the number of dispersion channels, but it usually requires a long data acquisition time. By exploring the process of coded spectral imaging, in this paper, we propose Fourier coded aperture transform spectral imaging (FCTS). We compared the performance of HTS and FCTS by theoretical analysis and experiments. The results show that FCTS has better concentrated energy than HTS on natural hyperspectral data. Compared with the Hadamard transform method, the Fourier basis mode has better physical correlation with the natural image, and the reconstruction quality of FCTS is better than that of HTS in the case of under-sampling. [ABSTRACT FROM AUTHOR]