Method for evaluating quality of full reference color image based on quaternion

Abstract

The invention discloses a method for evaluating the quality of a full reference color image based on quaternion. The method includes the following steps: constructing a brightness channel component template of any pixel point; respectively calculating the local variance of an original reference color image and a to-be-evaluated distorted color image; obtaining a local variance V, a brightness channel L, a first color channel A and a second color channel B; taking V as a real portion of a quaternion matrix, and L, A, and B as virtual portions of the quaternion matrix, respectively constructing the quaternion matrix of the original reference color image and the to-be-evaluated distorted color image; conducting singular value decomposition on the quaternion matrices to obtain corresponding singular value characteristic vector; and calculating the Euler's distance of the singular value characteristic vectors, and taking the Euler's distance as the distortion level metric of the to-be-evaluated distorted color image. According to the invention, the method can better evaluate the distortion of the color image, requires less calculation, and can effectively increase the relativity between a subjective evaluation result and objective perception, and produces the result that has the precision as human eyes.

Description

technical field [0001] The invention belongs to the field of color image quality evaluation, and more specifically relates to a quaternion-based full-reference color image quality evaluation method. Background technique [0002] In today's society, informatization, digitization, and network technology are developing rapidly, and at the same time, the application of digital products such as digital images, digital videos, and digital TVs is becoming more and more common. As the carrier of visual information, images play an increasingly important role in sharing, expressing, mining and exchanging information in people's daily life. At the same time, people's requirements for image quality are getting higher and higher. Image distortion will bring difficulties to subsequent image processing, analysis and understanding, which is not conducive to people's correct understanding of the objective world. Therefore, how to realize the effective evaluation of digital image quality bec...

AI Technical Summary

Problems solved by technology

[0009] (2) This patent document does not take into account the problem of non-uniform distortion when measuring the quality of distorted images, and regions with different degrees of distortion in the image have different effects on image quality perception

This patent document treats pixels with low distortion and high distortion equally, which will lead to inconsistency in the subjective and objective evaluation of non-uniform distortion

Images