Binocular image super-resolution reconstruction method based on multi-scale feature fusion

Abstract

The invention provides a binocular image super-resolution method based on multi-scale feature fusion. Firstly, a feature extraction module adopts mixed jump type residual connection, pooling blocks of an improved void space pyramid form a loopback structure for extracting multi-scale features of an image, and then an expansion residual and the loopback structure are alternately cascaded to fuse the extracted features; then, a parallax attention module is introduced to obtain a corresponding relation in the binocular image, useful information of an integrated image pair is integrated, and a transitional expansion residual block is used to learn the network capability of stereo matching features to obtain a parallax image of the binocular image; and finally, mapping a low-dimensional space to a high-dimensional space by using four expanded residual blocks, reconstructing a super-resolution left (right) graph through sub-pixel convolution, and applying FReLU to the whole network to improve the efficiency of capturing spatial correlation. According to the method, the multi-scale features of the image are extracted by using the expansion residual loop structure, the excellent super-resolution performance is realized, and the method has wide applicability.

Description

technical field [0001] The present invention relates to image processing technology, in particular to a super-resolution binocular image reconstruction method using an expanded residual loop structure to extract and fuse multi-scale features. Background technique [0002] Visual information is the main source for human beings to obtain information about all things. The process of improving the original image resolution through software is called image super-resolution (Super Resolution, SR) reconstruction. Image super-resolution technology meets the needs of people's perception. The visual field is booming. The research on image super-resolution has experienced leaps and bounds from the initial interpolation-based method to the current learning-based method, from artificial processing to artificial intelligence, from shallow network to deep network, and from deep network to lightweight network. Image super-resolution technology has made outstanding contributions to the deve...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to solve the problem of image super-resolution reconstruction, provide two low-resolution images of the left and right viewpoints in the same scene, and directly restore a high-resolution image through a deep learning network, which can be reconstructed Super-resolution left (right) image

Images