一种快速自适应的角点检测算法

A FAST ADAPTIVE CORNER DETECTOR

基于SUSAN原则提出了一种新的快速自适应角点检测算法,在几个方面进行了改进:以局部自适应阈值代替整个图像的固定阈值,提高了算法的自动处理能力;改进了响应函数,仅通过扫描模板边缘像素获取更多的角点信息,也简化了计算步骤;通过预处理,逐步缩小候选角点的搜索范围。实验证明,这是一种快速有效的角点检测方法。

A fast adaptive comer detector based on SUSAN algorithm is proposed in this paper. The main improved aspects are suggested just as follows：First,instead of one fixed threshold for the whole image ,the method adopts an adaptive threshold strategy based on local bright- ness. Second, more comer information is achieved by only scanning alone the mask boundary, it has the advantage of simplifying computational steps. In addition,in order to improving the efficiency,the method limits the search space for comer candidates to those pixels which are preprocessed. Experiments demonstrate that this comer detector is fast and efficient.