No-reference image quality evaluation method based on attention positioning network

Abstract

The invention discloses a no-reference image quality evaluation method based on an attention positioning network. The method is specifically implemented according to the following steps: inputting a training image into a VGG network to obtain global depth features; adding an attention positioning network to the last layer of the VGG network, and obtaining the position coordinates of the concerned area through the attention positioning network; cutting and magnifying the original image to obtain an attention attention image; inputting the attention attention image into the VGG network, and extracting local depth features; fusing the global depth features and the local depth features; carrying out regression training on the fused features and subjective MOS values, and establishing an image quality evaluation model; and inputting the to-be-detected distorted image into the image quality evaluation model, extracting image features according to the trained parameters, and obtaining an image quality score, thereby solving the problems of neglect of visual characteristics of human eyes and low accuracy when an image quality algorithm is constructed in the prior art.

Description

technical field [0001] The invention belongs to the technical field of image processing and image quality evaluation methods, and relates to a reference-free image quality evaluation method based on attention positioning network. Background technique [0002] With the advent of the 5G era and the rapid development of multimedia, image processing, and communication technologies, humans can disseminate and obtain multimedia data more conveniently and quickly. Because image data has the characteristics of rich content and simple expression, it can express information more intuitively than text, so image has great advantages as an information carrier. The use of images as information carriers shows a high rate of growth and is widely used in all aspects of life. Image quality has a great influence on the acquisition of human visual information. High-quality images are what users desire, because high-quality images carry more information. However, in the process of image acquis...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to provide a non-reference image quality evaluation method based on attention positioning network, which solves the problem of ignoring the visual characteristics of human eyes and low accuracy when constructing image quality algorithms in the prior art

Images