Improved randomized algorithm for circle detection
Aijun Chen · Guangdian gongcheng · 2006
An improved randomized algorithm is presented for circle detection. Edge pixels are tracked by a 4-neighbor-priority eight-direction tracking method and series of edge chains are formed. Then three edge pixels are randomly selected from every edge chain and a distance criterion is defined to determine whether there is a possible circle. An evidence-collecting process is applied to further determine if the possible circle is a true one when a possible circle is found. During the course, simple comparison operators, not complex square operators are used when edge pixels outside of the circum-rectangle of the possible circle are dealt with. Results obtained from experiments on real images demonstrate that the proposed algorithm performs well and fast.