Memory-efficient circle and ellipse detection algorithm in digital images

The method for detection of circle presented in this paper uses a structured data array instead of the 3-dimensional parameter space used in Hough Transform techniques. The edge points in image are first examined against a set conditions. Data are written to the data array upon satisfying those conditions. The result is small memory requirement which facilitates fast and effective search to be made through the data space. A visual approximation model of ellipses using circular arcs has been proposed. Based on the proposed approximation model, the circle detection algorithm has been extended to effectively detecting ellipses. The method presented has been applied successfully to detection of circles and ellipses in digital images