针对基本SUSAN算法存在漏检测、处理速度慢以及容易产生伪角点等问题,提出了一种改进的SUSAN角点提取算法。首先对图像的像素点按边缘点、内部点和背景点进行分类,在此基础上只对边缘点进行SUSAN算法处理,然后检查已经提取到的所有角点,剔除同一区域范围内的伪角点,从而实现对大部分角点的快速、准确提取。实验结果表明,该算法的运算速度比改进前提高约一倍,能够较准确地检测到目标角点。

To solve the problem of failing to detect corner points, slow speed of processing and the pseudo-corner, an improved SUSAN corner detection method is presented in this paper. This method firstly classifies pixels by its character into two parts. One includes edge points and another includes background points and inside points. On this basis, only edge points are processed with SUSAN algorithm, which can increase the processing efficiency. After that, it back-check all of the extracted corner-points, then gets rid of the false corners and saves the real corners, so as to detect most of corner points quickly and accurately. Experiment proves that this algorithm is quicker than before and it can detect the corners correctly.