基于边缘导向的直边图像亚像素定位方法  

An Edge-Directed Subpixel Localization Method of Straight Edge Image

作　　者：于新瑞[1] 汪国宝[1] 王石刚[1] 柯晓丹[1] 蒋志辉 

机构地区：[1]上海交通大学机械与动力工程学院,上海200030 [2]三一重工股份有限公司,长沙410100

出　　处：《上海交通大学学报》2004年第6期866-869,873,共5页Journal of Shanghai Jiaotong University

基　　金：国家自然科学基金资助项目(50390064;50375099);2002年度博士点基金资助项目(20020248048)

摘　　要：通过边缘检测,把图像分为边缘区域和非边缘区域,然后在边缘区域内进行边缘定位.根据局部区域内边缘的直线特性,求得小邻域内直线段的高精度位置;再根据边缘区域内边缘的全局直线特性,用线段的中点来拟合整个直线边缘,得到亚像素精度的图像边缘.在拟合的过程中,根据直线段转角的变化剔除了噪声点,提高了定位精度.并且,根据角度和距离区分出不同直线和它们的交点,给出了图像精确的矢量化结果.The image is divided into an edge region and non-edge region through edge detection, and then the edge is localized in the edge region. According to the straight property of the edge in the local region, the line segment is first localized with high precision in small neighborhood. Since the edge also has global straight property in the whole edge region, the midpoint of the line segment is used to fit the straight edge by least square method to get subpixel edge location. During the edge fitting, irregular points are eliminated by the angle of the line segment to advance the precision. Different edges and their intersections are also distinguished by their angles and distances, and the vectorial result of the image edge is given with high precision.

关 键 词：边缘导向 直线边缘 亚像素 图像定位 微电子制造 

分 类 号：TH741[机械工程—光学工程]