基于Hessian矩阵的角点自动检测算法

摘要

传统的角点识别方法存在一些缺陷,比如计算复杂度过高、误检漏检率较高等.针对这些问题,提出了一种基于Hessian矩阵的改进角点自动检测算法.该算法采用自适应阈值方法识别角点位置,同时采用圆形模版对称模版剔除伪角点,改进了传统固定阈值检测方法中对于伪角点剔除不够的缺陷.实验表明,提出的算法具有更好的灵活性和适应性,能够在不同工作环境下依然保持较高的焦点识别准确率.%Traditional corner point identification methods have some disadvantages, such as high computational complexity and high error detection rate. For these problems,this paper proposes an improved corner detection algorithm based on Hes-sian matrix. The algorithm adopts adaptive threshold method to identify angular point location and circular template symmetric templates to eliminate pseudo angular point. It improves the traditional fixed threshold detection method for missing pseudo Angle point. Experiments show that the proposed algorithm has better flexibility and adaptability, and can still maintain a high focus recognition accuracy under different working conditions.