摘要
提出一种基于相位信息的对称性检测新方法,将对称性检测问题转化为频域中的相位分析.文中的可行性分析、PSD的定义及其合理性证明为该算法奠定了理论基础.实验证明:该算法可直接应用于原始图像,不需要图像的任何先验知识,不需分割等任何预处理;具有旋转、亮度和对比度不变性;可以同时检测镜像对称、旋转对称、曲线对称等多种对称性.
A new method of symmetry detection, based on phase information, is proposed in this paper. In this way, symmetry detecting is converted to phase analysis in the frequency domain. The feasibility analysis, Phase-based Symmetry Detection （PSD） definition, and rationality demonstration establish the theoretic foundation for this algorithm. Experimental results show that it can apply directly to original images and it need no segment and any other preprocessing. And it is invariant to rotation, brightness and contrast. Finally, it can detect bilateral symmetry, rotational symmetry, curve symmetry and others.
出处
《计算机学报》
EI
CSCD
北大核心
2005年第11期1929-1933,共5页
Chinese Journal of Computers
基金
天津市高等学校科技发展基金(20041304)资助
关键词
相位信息
对称性检测
计算机视觉
phase information
symmetry detection
computer vision
