Super-resolution restoration network model generation method, image super-resolution restoration method and image super-resolution restoration device

Abstract

The embodiment of the invention discloses a super-resolution restoration network model generation method. The method specifically comprises the steps: obtaining a to-be-trained image, and inputting the to-be-trained image into a student network model and a teacher network model, so as to obtain a loss function corresponding to the student network model; and updating parameters of the student network model according to the loss function, so that the loss function of the student network model meets a preset condition, and generating and obtaining a super-resolution restoration network. The student network model is an ESRGAN network, and the ESRGAN network comprises a basic module, an up-sampling module and a convolution module. The basic module comprises one or more RRDB modules, each RRDB module comprises a plurality of processing modules, the input of each processing module is used as the input of a subsequent processing module, the transmission of features is enhanced, the subsequent processing module is enabled to use more image features for training, and the restoration effect is improved. A first convolution layer with a smaller convolution kernel is added in the processing modules, so that the dimension of image features is reduced, the calculation amount is reduced, and the processing speed is improved.

Description

technical field [0001] The present application relates to the technical field of image processing, and in particular to a method for generating a super-resolution inpainting network model, an image super-resolution inpainting method and a device. Background technique [0002] With the continuous development of Internet technology, short, flat, fast and high-traffic communication content has gradually gained the favor of major platforms and users, especially the transmission of short videos. However, due to noise, compression and other losses, video and image images are blurred and noisy, resulting in poor display images. [0003] In order to improve the quality of the image, a large number of image restoration methods have appeared, such as using the super-resolution restoration network to repair the image. However, in order to achieve a better repair effect, many super-resolution repair methods are time-consuming, resulting in an inability to balance the repair effect and ...

AI Technical Summary

Problems solved by technology

However, in order to achieve a better repair effect, many super-resolution repair methods

Images