基于曲线坐标系的图像边缘检测  

Image Edge Detection based on Curve Coordinate System

作　　者：顾晓东[1] 王晓明[1] 刘健[1] 

机构地区：[1]大连理工大学机械系现代制造研究所,大连116024

出　　处：《计算机工程与应用》2002年第3期36-37,共2页Computer Engineering and Applications

基　　金：国家自然基金项目(编号:59805001);辽宁省自然科学基金项目(编号:9810200102)

摘　　要：边缘检测是数字图像处理中一种重要的处理手段,目前普遍采用的方法是基于直角坐标系下的Laplacian算子,取其零交叉位置来检测。而现实中往往可见到许多曲面上的图像,比如柱面帖图、球面帖图,处理此类图像采用基于直角坐标系的微分算子显然不能胜任。该文采用基于微分几何曲线坐标系下普遍的Laplacian算子进行边缘检测,从结果可以看到处理效果明显优于普通的方法。Edge detection is an important process in image process,currently the popular method firstly to be used is the Laplacian transform which is based on the Cartesian coordinate system,then the zero-crossing of the Laplacian transform can be found as edge point.It is not conformable with many images on the surface such as cylindrical map image,spherical map image,which can be found in reality.This paper puts forward a united Laplacian transform which is based on the differential geometry curve coordinate system,the practical numerical experiment results show that this method has a strong edge detection ability and produces the better effect compared with old method.

关 键 词：曲线坐标系 LAPLACIAN算子 图像边缘检测 数字图像处理 

分 类 号：TN911.73[电子电信—通信与信息系统] TP391.41[电子电信—信息与通信工程]