An Improved Hyperspectral Pansharpening Algorithm Based on Optimized Injection Model

Abstract

The hyperspectral pansharpening is a significant preprocessing technology in hyperspectral images application. A new optimized injection model-based hyperspectral pansharpening algorithm is proposed in this paper. Compared with the traditional pansharpening methods, the algorithm achieves two major improvements: 1) the total injected spatial information is obtained by integrating the spatial components of hyperspectral (HS) and panchromatic (PAN) images by PCA transformation; and 2) the gain matrix proposed in this paper is composed of two factors which constraint spectral and spatial distortions respectively. Specifically, the morphological open-closing operation and Laplacian of Gaussian enhancement scheme are used for denoising the interpolated HS and PAN images, respectively. Then, the spatial components of the denoised HS and PAN images are respectively extracted by the morphological gradient operation and homomorphic filtering. The PCA transform is applied to the results to obtain the first principal component served as total spatial details. The total spatial information weighted by the gain matrix is finally combined with the interpolated HS images to generate the pan-sharpened images, in which a new gain matrix is constructed to minimize the spectral and spatial distortions. The extensive experiments have demonstrated the potential of the proposed method in balancing spectral preservation and spatial sharpness. INDEX TERMS Hyperspectral pansharpening, hyperspectral image, panchromatic image, spatial information, gain matrix.

Hyperspectral Image Super-Resolution with 1D–2D Attentional Convolutional Neural Network

“…Therefore, hybrid methods take advantages of algorithms in different algorithms. In addition, several variants and several variants with PCA [18,19] are exploited widely in hyperspectral pansharpening.…”

“…For example, the popular single HSI super-resolution methods are bilinear [2] and bicubic interpolation [3] based on interpolation, and [4,5] based on regularization. Pansharpening methods can be roughly divided into five categories: component substitution (CS) [6][7][8], which may cause spectral distortion; multiresolution analysis (MRA) [9][10][11][12], which can keep spectral consistency at the cost of much computation and great complexity of parameter setting; bayesian methods [13][14][15] and matrix factorization [16], which can achieve prior spatial and spectral performance at a very high computational cost; and hybrid methods [17][18][19], which are combinations of different algorithms.In situations without prior high resolution images, hyperspectral single image super-resolution (HSISR) is a challenging task. Although several deep learning-based HSISR algorithms have been proposed, they cannot effectively utilize sufficient spatial-spectral features while ignoring the influence from non-local regions.…”

Pansharpening Based on Low-Rank Fuzzy Fusion and Detail Supplement

“…1) The structural features of HF information are ignored. In recent years, with the development of image processing, multiscale analysis [32] has achieved great success. Remote sensing images cover a large area; thus, there are many multiscale and geometric singularities in the images.…”

Two-Stage Pansharpening Based on Multi-Level Detail Injection Network

“…Subsequently, the component with spatial structure of transformed MS image is replaced by the PAN image. The classic CS-based methods include intensity-hue-saturation (IHS) fusion method [4], principal component analysis (PCA) fusion method [5], and Gram-Schmidt (GS) fusion method [6]. CS-based methods can obtain rich detail, but the spectral distortion is usually serious.…”