基于轮廓曲线的角点识别方法

Identification Method Based on the Corner of the Contour Curve

Harris角点检测算法是一种经典的角点检测算子,针对其过于依赖阈值选取及计算量大的缺点,论文提出一种改进的基于轮廓曲线的角点识别的方法,通过判断边缘点方向性,先确定出粗角点,再采取由小尺度到大尺度逐步剔除伪角点的方法,最终实现角点识别。实验结果表明,文中方法对角点提取的时间仅为Harris算法的5%,而且角点定位精确,漏检率低。

Harris corner detection algorithm is a classical corner detection operator, for the shortcomings of over-reliance on the threshold and lots of calculation, this paper presents an improved method of identi- fication based on the corner of the contour curve, by judging the direction of edge points, determining the rough corners firstly, then gradually eliminating the pseudo-corner from small scale to large scale, finally achieving corner distinguished. The experimental result shows that the time of this method on corner ex- traction is just 5 % of the Harris algorithm,and locating corner precisely with low undetected.