一种新的高速圆形匹配算法

A new algorithm for rapid circle matching

摘要:

针对现有圆形匹配算法无法同时满足高速度、低内存消耗以及高精度要求的情况,提出了一种基于击中率的新型圆形匹配算法．算法引入以轮廓作为匹配特征信息的圆环采样模板,匹配的结果由击中率表征,并进一步根据搜索目标与定位区域的灰度相关性剔除误检对象．实验表明,算法能够实现毫秒级快速定位,且在内存消耗和可靠性方面都获得较好的性能．

Abstract:

For the existing circle matching algorithms could not simultaneously meet the requirements of high-speed, low memory consumption and high accuracy, a new algorithm based on hit rate was proposed. The Ring Sample Template was introduced and its edge was used as matching information. The matching result was determined by hit rate, and the Grey Scale Correlation between source and destination was calculated to eliminate wrong destination. The result showed that the algorithm could get destination position in milliseconds, and had good performances on both memory consumption and reliability.