相似像素的Harris角点检测改进算法

An improved Harris corner detection algorithm based on similar pixels

针对Harris算法提取的角点对尺度变化较敏感,且运行速度慢的问题,该文提出了一种基于相似像素的Harris角点检测改进算法。受SUSAN算法启发,改进算法首先计算目标像素8邻域内与之相似的像素数目,并据此筛选出候选角点;然后利用候选角点的相似像素数目改进角点响应函数;最后进行局部非极大抑制确定最终角点。实验结果表明,与Harris算法相比,改进算法所提取的角点位置更加准确,重复率较高,且角点检测时间仅为原算法的26.63%。本文所提算法提高了Harris算法的角点检测效率和稳定性。

Aiming at the problem that the Harris algorithm was sensitive to scale changes and ran slowly,an improved Harris corner detection algorithm based on similar pixels was proposed.Inspired by SUSAN algorithm,the improved algorithm first calculated the number of pixels within the8-neighborhood template of the target pixel that were similar to the target pixel,and selected candidate corners by its value;then,the corner response function was improved by the number of similar pixels of candidate corners;finally,the local non-maximal suppression was performed to determine the final corners.The experimental results showed that,compared with Harris algorithm,the improved algorithm achieve better location of corners and higher repeatability rate,and its corner detection time was only26.63%that of the original Harris algorithm.The proposed algorithm could improve the corner detection efficiency and stability of Harris algorithm.