加权SUSAN角点检测算法稳定性改进研究

SUSAN算法是比较稳定的角点检测算法,但是也存在着漏检和误检的情况。对经典SUSAN算法研究后针对其缺陷进行了改进,通过提出加权SUSAN算法,提高了角点检测的精度。同时,在USAN判决准则上进行了改进,新的准则增加了模板内和核心点灰度相似像素需要与核心点连通,并且USAN区域重心需要远离核心点的约束条件来进一步筛选候选角点。通过改进,减少了原算法的漏检和误检问题,实验结果也证明了改进方法的有效性。

Research on improving stability of weighted SUSAN corner detection algorithm

The traditional SUSAN algorithm is one of the most excellent corner detection methods.However,it could not detect all the corners and sometimes gets some false corners or missing detection.To solve these problems,an improved SUSAN corner algorithm,weighted SUSAN,is proposed after researching the classical method.Comparing with the traditional algorithm,the new method has a significant improvement in accuracy.There is an improvement in the USAN criterion.The new method calculates the number of the pixels in the adjoining nucleus and connected segment.In addition,it uses constraint condition that the USAN regional gravity should be away from the nucleus for further screening of candidate corners.Due to this improvement,it can avoid the missing detection and reduce false corner detection.Experiment results show that this Weighted SUSAN algorithm is accurate and efficient.