Image super-resolution method and system based on depth feature relevance

Abstract

The invention relates to an image super-resolution method and system based on depth feature relevance. The method comprises the following steps: acquiring a low-resolution image to be reconstructed; inputting the low-resolution image to be reconstructed into a trained student model for reconstruction to obtain a high-resolution image, wherein the training process of the student model comprises the following steps: taking a low-resolution training image as the input of the student model; taking the high-resolution training image as the output of the student model, taking the overall loss function as a loss function, and utilizing a loss function attenuation mechanism to supervise and train the student model through the high-resolution image and the real image output by the trained teacher model to obtain the trained student model. According to the invention, the high-resolution image is estimated through the image super-resolution model which is small in parameter quantity and high in performance.

Description

technical field [0001] The invention relates to the field of image super-resolution, in particular to an image super-resolution method and system based on depth feature correlation. Background technique [0002] In daily life, more and more low-power devices such as mobile phones and embedded terminals are widely used. People need to process low-resolution images in order to obtain high-resolution images with better visual effects on mobile devices. resolution image. Therefore, applying image super-resolution algorithms in low-power devices has received extensive attention. [0003] The goal of image super-resolution techniques is to estimate high-resolution images from low-resolution images. The degradation process of the image super-resolution problem is usually defined as: [0004] L=SM+n, (1) [0005] Among them, L, I, n denote the low-resolution image, high-resolution image and noise, respectively, and S and K denote the matrix form of downsampling matrix and blur k...

AI Technical Summary

Problems solved by technology

Traditional interpolation-based methods are simple and fast, but the recovered high-resolution images are of poor quality

In recent years, with the rapid development of deep learning technology, the method based on deep convolutional neural network has greatly outperformed the traditional interpolation-based method in the reconstruction process of high-resolution images, and related experiments have shown that the larger the number of parameters, the deeper The more the network can significantly improve the performance of the image super-resolution algorithm, but it also brings a significant increase in computing time and memory consumption

In real scenarios, these large models with huge amount of calculations and parameters cannot be deployed to low-power devices such as mobile phones.

Images