Algorithm for circle subpixel location detecting based on point Hough transformation

In order to improve location precision and speed of base circle in PCB (Printed Circuit Board) detecting, an algorithm for circle subpixel location detection was presented. At first, the edge points could be obtained from edge detecting the filtered image using canny operator, then the circle approximately position and its parameter could be gained through dealing with the edge points using point Hough transformation; secondly, the circle's edge points could be got through deleting big noise points according to the circle's parameter, then the circle's subpixel edge points could be obtained from subdividing the circle's edge points; finally the circle's precision location and its parameter could be obtained through fitting the circle's subpixel edge points using least square method. The experiments demonstrate that this algorithm can detect the base circle's subpixel location and its parameter. The center location precision is up to 0.5 pixel, and the radius detection precision is up to 0.3 pixel, and the speed is very high.