一种有效的基于Freeman链码的拐角检测法

拐角检测在模式识别和机器视觉中有重要作用。本文提出了一种有效的用于检测和确定拐角位置的方法，它能显著地改善现有拐角检测法的性能。该算法基于Ｆｒｅｅｍａｎ链码，它包括三个步骤：（１）排除掉在一条直线上的非拐角点，同时提取出必是拐角的点。（２）估算剩下可疑点的曲率，把曲率大于一阈值的点做为可能的拐角点。（３）挑出每组临近可能拐角点中曲率最大的点为真实拐角点，将其它曲率较小的可能拐角点作为伪拐角点排除掉。实验结果表明本文提出的方法优于现有的拐角检测法，它能准确地检测出同被测对象相一致的拐角点。另外，该方法的处理速度也比现有的拐角检测法快。

An Efficient Corner Detection Method Based on Chain Code

Corner detection plays a very important role in Pattern Recognition and Computer Vision. In this paper, an efficient method for detecting and locating corners is proposed. The proposed method can greatly improve the conventional corner detection methods. The proposed method is based on Freeman code and consists of three procedures: (1)exclude those points that are on a straight line and extract those points that surely are corner points. (2)estimate the curvature of the remaining doubtful points. The points whose curvatures are greater than a threshold value are considered as the potential corners. (3 ) locate the points with maximum curvatures in a set of close points as the true corners and hence elimiate the detection of spurious corners. Experimental results show that the proposed method is superior to the conventional methods. The corner points according to the human expectations can be properly detected by the proposed method. What is more, the proposed method can be processed very quickly.