An image super-resolution reconstruction method based on a dense feature fusion network

Abstract

The invention discloses an image super-resolution reconstruction method based on a dense feature fusion network. The method comprises the following steps: 1) preprocessing data; 2) establishing an image super-resolution reconstruction model; And 3) inputting the to-be-processed image into the model to obtain a high-resolution image. The image super-resolution reconstruction model comprises a coarse feature extraction network, a dense feature fusion network and an image reconstruction network; the coarse feature extraction network is used for extracting coarse image features of a low-resolutioncolor image; the dense feature fusion network is used for extracting high-order image features from the coarse image features; And the image reconstruction network is used for adding and fusing the coarse image features and the high-order image features to obtain dense image features, and then reconstructing the dense image features to obtain a color high-resolution image. According to the method, noise caused by a traditional interpolation amplification super-resolution algorithm can be effectively reduced, more high-frequency information is obtained to realize high-resolution image detail restoration, and the precision of super-resolution reconstruction is improved.

Description

technical field [0001] The invention is an image super-resolution reconstruction method based on a dense feature fusion network, which belongs to the field of image processing. Background technique [0002] The image super-resolution problem, especially the single image super-resolution (SISR, Single Image SuperResolution) problem is a classic problem in computer vision. It aims to obtain a visually pleasing high resolution (HR, High Resolution) image from a single low resolution (LR, Low Resolution) image produced by a low-cost imaging system and limited environmental conditions. Since super-resolution reconstruction does not need to consider too much cost, it is widely used in surveillance video, medical imaging, face recognition and other scenarios. [0003] Methods for SISR problems can be roughly divided into three categories: interpolation-based methods, reconstruction-based methods, and learning-based methods. Among these methods, the super-resolution reconstruction...

AI Technical Summary

Problems solved by technology

[0004] Although these super-resolution methods based on deep learning have improved the image reconstruction quality and algorithm efficiency to a certain extent, there are still deficiencies. In order to achieve the target image spatial resolution, the existing methods choose to directly process the input low-resolution image. Preprocessing for interpolation and amplification

For the very deep neural network, this approach not only increases the computational load, but also causes some details of the original low-resolution image to be lost.

Secondly, the reconstructed image is too blurry or smooth, the details are not restored enough, and even artifacts are generated.

These last methods only improve performance by increasing depth, which not only has little effect, but also makes network training difficult

Images