Method for evaluating objective quality of full-reference image

Abstract

The invention discloses a method for evaluating the objective quality of a full-reference image based on structure similarity, comprising the following steps: firstly, utilizing the space domain visual characteristic to obtain a visual perceptual map of an original image and solve for a position at which visual perception is remarkable; secondly, utilizing an evaluating method based on the structure similarity to solve for a structure similarity drawing SSIM (i, j) between the original image and a distorted image, calculating relative quality of the distorted image, and solving for a position of severe distortion; thirdly, defining the principle of visual attention focus transfer, ensuring a new visual attention focus, and producing a new visual perceptual map after the visual attention focus transfer; and fourthly, obtaining the objective evaluation of the image quality by using the weighted structure similarity of the produced visual perceptual map. The method is suitable for design of various image coding and treating algorithm as well as comparison of different algorithm effects, more accords with a human subjective evaluation on the aspect of an evaluating result of the image, and has a wide application prospect.

Description

technical field [0001] The invention relates to the technical field of digital video, in particular to a full-reference image objective quality evaluation method based on structural similarity. Background technique [0002] Digital images are widely used in multimedia products, which will cause quality loss in the process of acquisition, compression, storage and transmission. While using these multimedia products, human beings become the final recipients of digital images. Therefore, the human subjective quality evaluation (DMOS) of digital images is considered to be the most reliable. However, the process of subjective quality assessment is time-consuming and laborious, and the results are not reproducible. Therefore, scientists have done a lot of research on the objective quality evaluation methods of digital images for many years. According to the reference to the original image, the objective quality evaluation methods are divided into three types: full reference type...

AI Technical Summary

Problems solved by technology

[0010] The image quality evaluation method based on structural similarity proposed in literature [3] and [4] is based on the theory of literature [5], which more truly reflects the quality of visual perception. However, these two methods ignore the The effect of severe distortion on the choice of visual attention focus is proposed: in the process of observing images, the focus of visual attention is shifted by the control of human subjective will over time; the factors that determine the shift of visual attention focus can be significant Visual features or severe visual distortion; that is to say, usually the areas with prominent visual features in the image will attract people's attention first. Significant regions are shifted to heavily distorted regions

Images