摘要
为了提高角点检测的准确性和快速性,通过分析穿过角点的直线所具有的特性,提出了一种基于非终结线数目的快速角点检测算法。该算法以当前像素为核心作用一个圆形模板,计算穿过核心的非终结线的数目换算值,当该值位于给定的阈值范围内时,则当前像素为角点。针对伪响应,提出了一种基于非终结线均匀程度的非极大值抑制方法。通过实验验证了新算法的可行性,以及在准确性和快速性上所具有的优点。
In order to meet the demand of high accuracy with low computational complexity in corner detection,by analyzing the properties of the lines that across the corner,this paper proposed a new algorithm based on the number of the end-line. The algorithm employsed a circle mask centered on the pixel in question which would be called the＂nucleus＂of the mask,then calculatesd the equivalent value of the number of end-lines which acrossed the nucleus,the pixel would be identified as the corner while the equivalent value between maximum threshold and minimal threshold. To suppress the false response,this paper put forward a non-maximum suppression method. Experiments demonstrate validity of the new method,and show its advantages both in accuracy and rapidity.
出处
《计算机应用研究》
CSCD
北大核心
2010年第10期3995-3997,共3页
Application Research of Computers
基金
黑龙江省教育厅科学技术研究项目(11521331)
关键词
角点检测
同值收缩核
非极大值抑制
特征检测
图像处理
corner detection
USAN（ univalue segment assimilating nucleus）
non-maximum suppression
feature detection
image processing
