A fast and accurate circle detection algorithm is presented. The proposed algorithm satisfies the application requirements on accurate circle localization for porous components with excellent real-time performance. The core idea of the algorithm is to cluster circle edge points by an easy-to-implement image analysis technique. Specifically, the original image is firstly down sampled with uniformly spaced grid to form an image with smaller size. The new image is then analyzed, pixel by pixel, to find the candidate grid points located inside the circle on the original image. Secondly, candidate grid points are evaluated to eliminate the ones which are not inside real circles. Thirdly, grid points located inside one circle are regrouped according to minimum distance between circles. These remaining grid points are the reference points to cluster circle edge points. Finally, parameters for a specific circle including radius and center coordinates are calculated based on the edge points belonging to that circle with Random Sample Consensus (RANSAC) algorithm. Experiment results demonstrate the efficiency of the proposed circle detection method.
You can submit articles in our peer reviewed Journal at http://scitechnol.com/submitmanuscript.php