基于Sobel边缘检测的圆周Harris角点检测算法

(西安科技大学 计算机科学与技术学院,陕西 西安 710054)

角点检测; 改进Harris算法; 旋转不变性; 圆周窗口

Circular Harris corner detection algorithm based on Sobel edge detection
DONG Li-hong,PENG Ye-xun,FU Li-mei

(College of Computer Science and Engineering,Xi'an University of Science and Technology,Xi'an 710054,China)

corner detection; improved Harris algorithm; rotation invariance; circumferential window

DOI: 10.13800/j.cnki.xakjdxxb.2019.0227

备注

特征点的选取是视觉研究的基础,其选取的精确度直接影响到视觉信息处理的准确性与可靠性。传统Harris角点检测方法具有计算简单、易于实现的优点,但该算法效率低下、检测精度较低、抗噪性差、存在角点簇。提出一种基于Sobel边缘检测的圆周Harris角点检测算法:首先采用Sobel边缘检测进行角点预筛选,本质上提高检测效率; 随后采用圆周窗口模板对筛选后的角点进行非极大值抑制,减少漏检点与伪角点的个数; 最后采用临近点剔除法,保留非极值最大点作为角点,解决角点簇问题。实验结果表明,文中角点检测算法运行效率得到了极大提高,减低漏检率的同时,旋转不变性亦得到了改善。

The selection of feature points is the basis of visual research,and the accuracy of its selection directly affects the accuracy and reliability of visual information processing.The traditional Harris corner detection method has the advantages of simple calculation and easy implementation.However,the algorithm has low efficiency,low detection precision,poor noise resistance and corner clusters.In this paper,a circumferential Harris corner detection algorithm based on Sobel edge detection is proposed.Firstly,Sobel edge detection is used to perform corner point pre-screening,which improves the detection efficiency in essence.Then,the circumferential window template is used to suppress the non-maximum value of the selected corner points to reduce the number of missed detection points and pseudo corner points.Finally adopt the adjacent point culling method,retain the non-extreme maximum point as the corner point,and solve the corner point cluster problem.The experimental results show that the running efficiency of the corner detection algorithm is greatly improved,and the rotation invariance is also improved while reducing the missed detection rate.