Image super-resolution reconstruction method based on residual learning

Abstract

The invention discloses an image super-resolution reconstruction method based on residual learning, and the method comprises the steps: obtaining a public image data set for a training system, and processing the public image data set to obtain a training set and a test set; obtaining a to-be-processed low-resolution image; constructing and fusing all modules of a system core neural network to forma deep convolutional network based on residual learning; calculating network prediction and label loss, and adjusting network parameters according to the loss; training a deep residual network and inputting a to-be-processed low-resolution image to the deep residual network; outputting the reconstructed high-resolution image, evaluating the image reconstruction precision, and completing the imagesuper-resolution reconstruction. According to the invention, the extraction capability of image features is improved, the generalization and high efficiency of an image reconstruction system are improved, and the image reconstruction precision is more accurate.

Description

technical field [0001] The invention relates to an image processing method, in particular to an image super-resolution reconstruction method based on residual learning, which belongs to the field of deep learning. Background technique [0002] Image super-resolution reconstruction is a classic problem in the field of computer vision, which aims to convert one or more low-resolution images into high-resolution images through algorithms. As a standard for evaluating image quality, the higher the resolution, the higher the pixel density of the image, the richer the detailed information contained, and the higher the image quality. In reality, due to the influence of many objective factors such as acquisition equipment, noise, and shooting environment, the acquired pictures often cannot meet the needs. Therefore, improving the resolution and quality of images has always been a hot issue that people pay attention to. The most direct way to improve the image resolution is to impro...

AI Technical Summary

Problems solved by technology

[0008] Purpose of the invention: In view of the problems of weak network learning ability, long training time and the need to improve the quality of reconstructed images in most networks, the present invention provides an image super-resolution method based on residual learning, which uses the original low-resolution image As the input of the model, a small convolution kernel is used to cascade a deep convolutional network to expand the receptive field and make full use of the context information; by introducing the residual block, the high-frequency information of the image is learned, which reduces the computational complexity and improves the network training. Convergence speed to achieve high-precision real-time reconstruction of images

Images