An Algorithm for Fast Circular Object Detection Based on Discrete Evolute Distribution

Based on the features of circle-like edge in digital image, an algorithm is presented in which the evolute of edge is calculated and from its distribution, the circles are located. The edge of target is first extracted and traced. Then the centers of curvature are calculated by least square fitting in fraction of edge and thus the evolute of edge is worked out. From the evolute distribution and the fitting degree, the circles can be recognized and their radii and centers are calculated. Experiments show that by the proposed method, circle-like target in image can be detected efficiently and accurately. And it is effective even for the distorted and deformed circles.