Image recovery quality evaluation method based on multilevel differential learning

Abstract

The invention belongs to the field of computer vision, and particularly relates to an image recovery quality evaluation method based on multilevel differential learning, which comprises the following steps: acquiring a to-be-evaluated recovery image, and inputting the recovery image into a trained image recovery quality evaluation model based on multilevel differential learning to obtain a quality evaluation result; classifying and storing the image according to a quality evaluation result; wherein the image restoration quality evaluation model based on multi-level difference learning comprises a multi-level difference generation sub-network and a perception difference regression sub-network; according to the method provided by the invention, the quality evaluation is not only carried out on the recovered image, but the perception difference value between the distorted recovered image pairs can be quantified, so that the prediction on the recovered image is more reliable and effective.

Description

technical field [0001] The invention belongs to the field of computer vision, and in particular relates to an image restoration quality evaluation method based on multi-level difference learning. Background technique [0002] Image restoration (IR) technology refers to restoring low-quality images into clear images. However, there are few studies on how to objectively benchmark these algorithms, which is a challenging problem that may hinder the rapid development of image restoration techniques. The image quality evaluation technology designed for the image restoration scene can be used to sort the performance of the image restoration algorithm, select the parameters and guide the design of the image restoration algorithm, which can solve this problem well. [0003] At present, most of the non-reference image quality evaluation methods based on neural networks are designed for traditional distortion types. Although good prediction results have been achieved on CSIQ, LIVE an...

AI Technical Summary

Problems solved by technology

[0004] In summary, the existing technologies have the following problems: 1. The existing image restoration quality evaluation methods cannot be applied to all restoration scenarios, and the generalization ability is relatively weak; 2. The existing image restoration quality evaluation methods cannot quantify the distortion of the restored image. mass change between

Images