Multi circle detection by using evidence accumulation

Hudai Fu, Jingang Gao

School of Mechatronics Engineering, Changchun Institute of Technology, Changchun 130012, JiLin, China

The traditional circle detection algorithm has complex computation, large memory space occupation and other deficiencies. It has low detecting efficiency and not suitable for the multi circle detection. So it proposes a multi circle detection method based on global search and evidence accumulation in the paper. The evidence accumulation and the weighted average are combined in the algorithm. The pseudo centre processed during the evidence accumulation are classified and analysed. Three kinds of pseudo centres are eliminated by class. Finally, the other circle parameters are calculated. It is proved that the algorithm has high precision, high efficiency, and low sensitivity to the defect of local information. Also the detecting time will not increase with the number of circles. The multi circle detection effect is obviously superior to the traditional randomized circle detection algorithm.