Image processing method, apparatus, device, and computer-readable storage medium

Abstract

The present application provides an image processing method, device, equipment, and computer-readable storage medium; the method includes: acquiring an image to be processed, and performing feature extraction on the image to be processed to obtain image features of the image to be processed; based on the Determine the control vector and compression degree coefficient based on image features; determine the target exit stage of the trained image enhancement sub-model based on the compression degree coefficient and the model structure of the trained image enhancement sub-model; based on the image features and the control vector, obtain The image processing result of the trained image enhancement sub-model when reaching the target exit stage; output the image processing result. According to the present application, the exit stage of image processing can be determined in an adaptive manner, and the efficiency of image processing can be improved.

Description

Technical field [0001] The present application relates to image processing techniques, and more particularly to an image processing method, apparatus, device, and computer readable storage medium. Background technique [0002] At present, with the development of terminal technology, the pixels of image acquisition devices are getting higher and higher, so the size of images or video files collected by image acquisition devices also increases. In order to save transmission bandwidth, storage space, JPG compression technology is widely used. During the JPG compression processing of the user, different compression ratios are used to varying the image to varying degrees of compression based on the actual needs of the network. Different compression ratios will bring different degrees of block effects, ringing effects, and blurring, resulting in different degrees of image quality decline. Currently, an image-de-compression effect enhancement algorithm based on deep learning is a good e...

AI Technical Summary

Problems solved by technology

Different compression rates will bring different degrees of block effects, ringing effects and blurring, etc., resulting in different degrees of image quality degradation

At present, the image decompression effect enhancement algorithm based on deep learning has achieved good results. However, a stable network will inevitably bring a small amount of loss of detail while removing the compression effect.

Because simply using a wide range of JPG compression coefficient pictures to train the model, a general-purpose network that can handle different compression situations is obtained. Since the network can only implement a deterministic mapping, it often cannot achieve the optimal in each situation. The application benefit is too small; while analyzing the quality of the image separately, training multiple networks for different compression situations is inefficient and consumes a lot of resources, and for the currently processed image, the compression factor is usually unknown

Images