Full reference image quality assessment method in consideration of gradient direction difference

Abstract

The invention discloses a full reference image quality assessment method in consideration of gradient direction difference. The method comprises the following steps: firstly carrying out graying treatment on a reference image and a distortion image to obtain a luminance component and eliminating color information; carrying out two-dimensional average low-pass filtering on a luminance component, carrying out 1 / 2 spatial down-sampling on width and height to obtain a corresponding image with width and height being half of the size of an original image; calculating gradient intensity of the reference image and the distortion image after down sampling in the horizontal direction, the vertical direction, the main diagonal direction and the minor diagonal direction at each pixel position by using a Prewitt gradient operator; calculating the difference of the gradient intensity of the reference image and the distortion image in the four directions at the each pixel position and counting to obtain a maximum value of the difference of the gradient intensity in four directions at each pixel position; finally accessing the overall image quality by using counting average of the maximum value of the difference of the gradient intensity at all the pixel positions.

Description

Technical field [0001] The present invention relates to the technical field of image quality evaluation, in particular to a full-reference image quality evaluation method that considers the difference in gradient directions. Background technique [0002] Image quality evaluation has important applications in many image processing fields. At present, image quality evaluation methods are mainly divided into subjective evaluation and objective evaluation. Subjective evaluation is performed by a group of human observers under certain environmental and equipment conditions by subjectively scoring distorted images, and finally obtaining image quality indicators through statistical methods. Since most of the final recipients of image processing systems are humans, subjective evaluation is the most accurate; however, subjective evaluation methods are complicated in process, time-consuming and labor-intensive, and most importantly, it is impossible to evaluate and monitor image quality i...

AI Technical Summary

Problems solved by technology

However, most of the current evaluation algorithms only use the intensity information of the gradient, and rarely consider the difference in the direction information of the gradient, and the distortion will change the intensity and direction of the gradient at the same time.

Images