Image super-resolution model training and reconstruction method and device, equipment and storage medium

Abstract

The embodiment of the invention provides an image super-resolution model training and reconstruction method and device, equipment and a storage medium, and the method comprises the steps: obtaining a plurality of sample images from a training data set; inputting the sample images into a first super-resolution reconstruction model to obtain sample reconstruction images respectively corresponding to the sample images; obtaining an image effect of the sample reconstruction image, and determining a sample reconstruction image with the image effect not meeting a preset reconstruction requirement; obtaining the same kind of images of the sample reconstruction image with the image effect not meeting the preset reconstruction requirement; updating the training data set based on the same kind of images; and using the updated training data set to train to obtain a new image super-resolution reconstruction model. Therefore, the loss of image details in the super-resolution reconstruction process can be reduced.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to an image super-resolution model training and reconstruction method, device, equipment and storage medium. Background technique [0002] Image Super Resolution (Image Super Resolution), that is, image super-resolution technology, refers to reconstructing a corresponding high-resolution image from a low-resolution image. Monitoring equipment, satellite images, medical imaging and other fields have important application value. In related technologies, a model for super-resolution reconstruction is obtained through training based on a large number of pictures, and super-resolution reconstruction is realized through the pre-trained model. [0003] However, in related technologies, the pictures used in the process of training the model are generally obtained from one or several independent open source data sets. However, most of the pictures in the existing open sour...

AI Technical Summary

Problems solved by technology

However, most of the pictures in the existing open source datasets are used for image classification, image segmentation, etc., and there is no open source dataset specifically for image super-resolution reconstruction, so that the model for super-resolution reconstruction based on image training in open source datasets can be used In super-resolution reconstruction, the loss of image details is serious

Images