针对Harris角点检测算法中, 对于图像角点存在错检和漏检的问题, 提出了一种基于图像边缘的Harris角点检测方法。该方法首先采用二阶微分算子提取图像边缘, 然后在得到的图像边缘的基础上进行Harris角点检测, 最后记录检测出的角点的位置并在原图上进行标记。实验结果表明, 相对于传统的采用梯度进行角点检测的方法, 采用二阶微分算子提取边缘后的角点检测方法能够更多更准确地检测到图像的角点。

In order to solve the undetected and error detected problem in Harris corner detection method, an improved Harris corner detection method based on edge is put forward. The improved method firstly extracts edge of input image with second order differential operator and then detects the corners in the edge image with Harris corner detection method. Finally the positions of the detected corners are recorded and marked in the original image. Experimental results show that the improved corner detection method which extracts edge with second order differential operator is able to detect more corners more accurately than that with gradient.