摘要
提出一种基于圆盘结构元的形态学角点检测算法,使用形态学方法,在圆盘结构元高帽运算后剩余部分中确定一个三角形区域,通过检测三角形的顶点获得图像中的角点.在检测角点的同时,利用三角形区域的几何特征完成检测角的角度和朝向估计,并引入结构元尺寸自学习机制自动确定结构元尺寸,完成角检测.与其他6种角检测方法的对比实验证明了文中方法的优越性,并对其他角检测方法中很少涉及的角度和朝向估计给出了实验结果.
A new morphological corner detector is proposed. At first, a triangle area is determined in the residual of the disk element tophat operation, then corners are detected by the convex corner of the triangle area. The proposed detector can additionally estimate the corner angle and orientation from the triangle geometric features and is also able to auto-adjust the disk element size during the detection process. Six other detectors are experimentally compared and our proposed detector manifests some appealing superiority.
出处
《计算机辅助设计与图形学学报》
EI
CSCD
北大核心
2007年第10期1342-1347,共6页
Journal of Computer-Aided Design & Computer Graphics
基金
公安部重点项目(20029322301)
黑龙江省自然科学基金(F0318)
关键词
形态学
角点检测
角估计
自适应
mathematical morphology
corner detection
angle estimation
adaptation
