Hyperspectral image super-resolution combining with deep learning and spectral unmixing

• Published in: Signal processing. Image communication Vol. 84; p. 115833
• Main Authors: Zou, Changzhong, Huang, Xusheng
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier B.V 01.05.2020; Elsevier BV
• Subjects: Artificial neural networks; Computer vision; Deep learning; Deep residual convolutional neural network (DRCNN); High resolution; Hyperspectral image (HSI); Hyperspectral imaging; Image enhancement; Image reconstruction; Image resolution; Remote sensing; Spatial data; Spatial resolution; Spectra; Spectral unmixing; Super-resolution; Total variation (TV) regularity; Visual effects; Deep residual convolutional neural network (DRCNN); Super-resolution; Hyperspectral image (HSI); Total variation (TV) regularity; Spectral unmixing
• ISSN: 0923-5965; 1879-2677
• DOI: 10.1016/j.image.2020.115833

In recent years, hyperspectral image super-resolution has attracted the attention of many researchers and has become a hot topic in the field of computer vision. However, it is difficult to obtain high-resolution images due to imaging hardware devices. At present, many existing hyperspectral image super-resolution methods have not achieved good results. In this paper, we propose a hyperspectral image super-resolution method combining with deep residual convolutional neural network (DRCNN) and spectral unmixing. Firstly, the spatial resolution of the image is enhanced by learning a priori knowledge of natural images. The DRCNN reconstructs high spatial resolution hyperspectral images by concatenating multiple residual blocks, each containing two convolutional layers. Secondly, the spectral features of low-resolution and high-resolution hyperspectral images are linked by spectral unmixing. This approach aims to obtain the endmember matrix and the abundance matrix. The final reconstruction result is obtained by multiplying the endmember matrix and the abundance matrix. In addition, in order to improve the visual effect of the reconstructed image, the total variation regularity is used to impose constraints on the abundance matrix to enhance the relationship between the pixels. The experimental results of remote sensing data based on ground facts show that the proposed method has good performance and preserves spatial information and spectral information without the need for auxiliary images. •We propose a HSI super-resolution method without the auxiliary images.•Our framework combines with deep residual convolutional network and spectral unmixing.•The TV is used to impose constraints on the abundance matrix.•The experimental shows that our method has good performance than the state-of-the art methods.