A hyperspectral image super-resolution restoration method based on a 3D convolutional neural network

Abstract

The invention discloses a hyperspectral image super-resolution restoration method based on a 3D convolutional neural network. According to the technical scheme, the 3D residual dense network is characterized by comprising 3D convolution kernel to convolve the hyperspectral image spectral dimension and 3D sub-pixel recombination to enlarge the image and reconstruct the high-resolution image part, and unifying the two parts in the deep convolutional neural network framework 3D-RDN; hierarchical characteristics of the convolutional layer are fully utilized through structures such as residual dense blocks, and super-resolution restoration of the hyperspectral image is achieved. At present, when an existing method based on deep learning is applied to a hyperspectral image, the characteristics of the hyperspectral image are not fully considered, and therefore it is difficult to effectively utilize rich spectral dimension information of the hyperspectral image to reconstruct a high-resolutionimage. According to the method, all spatial spectrum information of the hyperspectral image is fully utilized, efficient super-resolution restoration is achieved, and the PSNR value is superior to that of an existing method.

Description

technical field [0001] The invention belongs to the field of image restoration in computer vision, in particular to super-resolution restoration of hyperspectral images, and further relates to a method for super-resolution restoration of hyperspectral images using convolutional neural networks, making full use of 3D convolution kernels, 3D Subpixel reorganization techniques and characteristics of residual dense networks. This method has been verified with good recognition performance on public datasets. Background technique [0002] Hyperspectral image is a kind of image that can realize the simultaneous acquisition of spatial information and spectral information of ground objects. It has very important application value in the fields of deep space exploration, geological exploration, crop remote sensing and even face recognition. However, due to the constraints of imaging principles, the spatial resolution of hyperspectral images is often lower than that of ordinary visibl...

AI Technical Summary

Problems solved by technology

However, the existing methods of this type are mainly developed for ordinary two-dimensional images, and have not fully considered the characteristics of hyperspectral images themselves, so it is difficult to effectively use the spatial and spectral dimension information of hyperspectral images to reconstruct high-resolution images. The convolutional neural network fails to make full use of the features extracted by each layer of convolution

Images