基于图论的数字图像边缘检测算法

Digital Image Edge Detection Algorithm Based on Graph Theory

基于图论的图像边缘检测技术是近年来国际上图像分割领域的一个新的研究热点。介绍了用图论的知识来检测图像边缘的方法,将图像映射为带权的有向图,根据图像边缘的特点——灰度级间断、亮度不连续性这一特点作为权重函数,通过搜索低开销路径来检测图像边缘,在搜索起始点的过程中可以去除孤立噪声,使检测出来的图像边缘清晰。与经典的边缘检测算子—Laplace算子进行比较,并经过仿真实验可知基于图论的边缘检测算法是一种全局性与局部性相结合,且具有抗噪性和时效性的边缘检测方法。

Based on graph theory technique of edge detection image segmentation in recent years the field of international to a new research focus.Introduced with the knowledge of graph theory methods to detect image edge,the image is mapped to the right with a directed graph,according to the image edge features-gray-scale discontinuity,the continuity of this feature is not as light weight function,by searching low-cost path to detect the image edges,the starting point in the search process can remove isolated noise,the detected image edge clear.With the classical edge detection operator-Laplace operator to compare,and after simulation shows that graph-based edge detection algorithm is a combination of global and localized,and has anti-noise and edge detection timeliness.