Image super-resolution reconstruction method based on multi-scale attention cascade network

Abstract

The invention provides an image super-resolution reconstruction method based on a multi-scale attention cascade network. The image super-resolution reconstruction method comprises the following steps:firstly, extracting shallow features of a low-resolution image by using convolution operation; secondly, inputting the shallow features into a feature extraction subnet to obtain cascade features; thirdly, the cascade features pass through a convolution layer with a convolution kernel of 1, and obtaining optimized features; fourthly, adopting a bicubic linear interpolation algorithm for a low-resolution image ILR to obtain a reconstructed image while inputting the optimized features into an image deep learning up-sampling module to obtain a reconstructed image; and finally, fusing the reconstructed image to obtain a final high-resolution reconstructed image ISR. The method is suitable for super-resolution reconstruction of the image, and the obtained reconstructed image is high in definition, more real in texture and good in perception effect.

Description

technical field [0001] The invention belongs to the field of image restoration, and relates to an image super-resolution reconstruction method, in particular to an image super-resolution reconstruction method based on a multi-scale attention cascade network. Background technique [0002] Single Image Super-Resolution Reconstruction (SISR) has recently received a lot of attention. In general, the purpose of SISR is to produce a visually high-resolution (HR) output from a low-resolution (LR) input. However, the whole process is completely irreversible because there are multiple solutions for the mapping between LR and HR. Therefore, a large number of image super-resolution reconstruction (SR) methods have been proposed, ranging from early interpolation-based methods and model-based methods to recent deep learning-based methods. [0003] The method based on interpolation is simple and fast, but cannot be widely used because of poor image quality. For more flexible SR methods...

AI Technical Summary

Problems solved by technology

[0006] The problem to be solved by the present invention is: in the existing super-resolution reconstruction methods based on deep learning, most methods blindly increase the depth of the network to improve the performance of the network, but ignore the characteristics of fully utilizing the LR image; Moreover, as the depth of the network increases, the feature information will gradually disappear during the transmission process; using the LR image enlarged by interpolation as the input of the network will increase the computational complexity and is not conducive to the network's learning of image details

Images