针对常见的几何形状匹配算法对目标遮挡较为敏感, 提出了一种基于角点匹配的几何形

状定位。 该方法首先根据边缘曲率提取图像的角点, 然后采用基于改进的投票策略的角点匹配算法对检测图与模板图进行匹配, 最后通过 Ransac算法去除错匹配。 实验表明, 该算法定位效果良好,有效地解决了目标部分遮挡问题。-A noval geometry shape position algo rithm based on point feature matching is proposed to solve the problem that geometry shape matching in common is sensitive to object occusion .A corner

detector based on edge curvature is applied in feature extraction and a i mproved voting-strategy-based corner matching algorithm is employed to match the model image and test

image . Ransac is used to reduce the m is match points . The experimental resu lts show that the efficiency of position is good , and the part occusion is effect ively solved.