The method to extract concentric ring radius and circle center based on image inversion

Aiming at the problem of accurate extraction of circle center and ring radius in non-localization interference concentric ring image, taking point light source non-localization interference concentric rings as the object, an algorithm based on image inversion for simultaneous caring of both dark and bright rings was proposed. The reference center was obtained using the Hough circle transformation after the pre-processing of smooth noise reduction. Then intensity outlines with an interval of 5 pixels in the range of [-50, 50] pixels in its x-direction and y-direction, respectively, were taken. With the inversion of each intensity contour line, accurate identification of dark and bright ring peak coordinates, respectively, was achieved. Finally, the accurate extraction of the circle center and radius of the image was achieved by averaging multiple circle parameters after circle regression fitting of obtained series circle center and radius. The results show that the center coordinates of the three bright rings and the three dark rings are very close to each other and have good reproducibility with the maximum deviation of -3.7 pixels to the mean value, i.e., a maximum relative deviation from the mean is -0.15%. The relative error of the radius-square ratio of two adjacent rings to the actual ratio varies between -4.18% and 0.36%, verifying the feasibility of the algorithm. This research is helpful in realizing automatic detection and improving measurement accuracy.