摘要
文中通过傅里叶变换（Ｆｏｕｒｉｅｒｔｒａｎｓｆｏｒｍ ），在频率域，通过求垂直于边缘的一阶方向导数，并检测其极值（极大或极小）点来确定边缘．同时，参考一阶方向导数的数值来排除一些非边缘点（一阶方向导数的数值为常数者，零也为常数）．实验证明该方法检测边缘的准确度高，而且不用选取门限值（阈值）．
In this paper, a new edge detection method——method using extremum of first directional derivative is proposed, based upon a general gradient edge model. Edge lines are extracted automatically from image, by acquiring points of extremum of first directional derivative vertical with edge line with Fourier transform in frequency domain. In detecting edges, non\|edge points are rejected using values of first directional derivatives vertical with edge line. Experiments involving both test and real image are conducted. Acquiring second directional derivative vertical with edge line is the main character of the method.
出处
《计算机研究与发展》
EI
CSCD
北大核心
2000年第2期244-247,共4页
Journal of Computer Research and Development
基金
吉林大学符号计算与知识工程教育部开放研究室实验室
国家自然科学基金项目!(项目编号69883003)
关键词
边缘检测
一阶方向导数
傅里叶变换
图像处理
edge (boundary) detection, first directional derivative, extremum, Fourier transform
