The Statistical Clustering Method of Ellipse Detection
Ei Y · Microcomputer Information · 2007
This paper proposed a statistical clustering method of ellipse detection. This method avoided many invalid sampling data which restrict the sample points utilizing the geometric properties and shape size of ellipse. The authenticity of a potential ellipse can be con-firmed by the statistical value,which is the ratio of the perimeter of tangent rectangle of the ellipse to the amount of points with less dis-tance to the ellipse than the threshold. Compared with accumulated validation in the parameter space of RHT(Randomized Hough Trans-form),this method can simplify calculate at least cost,and it eliminates the need of allotting memory for the virtual ellipse and searching in the parameter space,moreover,it can quickly wipe off edge points in the validated ellipse and reduce invalid sampling.