An Automatic Analysis Method of Geometric Distortion for Image Quality Objective Evaluation Point Map

Abstract

The invention relates to an automatic analysis method of geometric distortion for objective image quality evaluation point diagrams. This method supports the automatic completion and search of internal residual defects, so that the overall target segmentation process and the point diagram sorting process form a closed loop to achieve the optimization of sorting. , which increases the robustness of the distortion analysis process to changes in point image segmentation results; considering that the point image image may have incomplete boundary points during the shooting process, resulting in inaccurate positioning of the boundary points during the sorting process, this method uses regional The growth method uniformly adjusts the overall incomplete boundary points, which increases the robustness of the distortion analysis process to the incompleteness of the point diagram boundary points.

Description

【Technical field】 [0001] The invention relates to the field of objective image quality evaluation, and can be used for an analysis method for automatically extracting and analyzing geometric distortion-related parameters based on dot-map charts in an image objective evaluation system. 【Background technique】 [0002] For the current objective image quality evaluation system, there are two main chart cards used for geometric distortion evaluation, namely Checkerboard and dot chart. This method only makes technical modifications for the detection errors that are prone to occur in the dot chart chart. Therefore, Checkerboard-based geometric distortion analysis is not considered. As the name suggests, except for the monochromatic background, the rest of the charts are circles with the same radius and uniform distribution in space. According to whether the background of the chart is dark or not, it is divided into two types: white dots on black background and black dots on white b...

AI Technical Summary

Problems solved by technology

[0009] 1) When the point number, size and other attributes of the point map change greatly, the result of point segmentation based on Hough transform may not be ideal, which is related to the setting parameters of the Hough transform method. If the point map is too dense or scattered, the segmentation Leakage points and noise points are easy to appear in the results (one point is recognized as multiple points);

[0010] 2) When there are incomplete and noise points, there will be obvious sorting errors when doing point sorting, resulting in inaccurate results from statistical analysis

Images