边缘检测中局部区域的动态阈值选取方法  

Dynamic threshold selection based on local region in edge detection

作　　者：王佳[1] 李波[1] 徐其志[1] 

机构地区：[1]北京航空航天大学计算机学院数字媒体室,北京100191

出　　处：《计算机应用研究》2010年第2期772-774,共3页Application Research of Computers

基　　金：武器装备预研基金资助项目;国家"863"计划资助项目

摘　　要：现有的边缘检测方法主要采用全局阈值选取方法,由于全局阈值的选取不当,易造成图像中重要而梯度变化较弱的边缘丢失,而局部阈值选取研究相对不足且实用性差。针对以上问题,提出了基于局部区域的动态阈值选取方法:以每个候选边缘点为中心,局部矩形区域内动态确定该点的双阈值,根据边缘点梯度模值与其双阈值大小关系选取边缘点。与Canny边缘检测中采用的阈值选取方法相比,本算法能够有效地提取图像中重要边缘,且抗噪声能力强。Most of the current edge detection methods adopt global threshold selection techniques. However, global threshold selection sometimes is not adaptive to detect out important edges, whose gradient intensity are below it. In recent years, local threshold selection methods has developed, but the investigation is still insufficient and complicated to apply. Thus, in order to solve these problems, this paper proposed a method of dynamic threshold selection based on local region ： each candidate edge point had two thresholds in local square region, where the point was in the center. Comparing the gradient of the edge point with the threshold, edge points＇can be picked out successfully. Experiments show that dynamic threshold selection has better results and little influence from noise than global thresholds selection used in Canny algorithm.

关 键 词：动态阈值选取 边缘检测 CANNY算法 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]