一种基于偏微分方程的形态学边缘检测新方法

Morphological edge detector based on partial differential equations

提出一种基于偏微分方程 (PDE)的形态学腐蚀算子 .在边缘检测中 ,直接对原图像求取一阶差分 ,利用该形态学腐蚀算子增强边缘梯度、降低噪声的影响 ,突破了传统的先低通滤波后计算梯度的边缘检测思路 .形态学运算不会带来边缘位置的偏移 ,因此能够对图像的边缘准确定位 ,同时具有抗噪性 .将新方法的边缘检测性能和Canny算子、Log算子做了比较 ,结果表明提出的边缘定位方法更加准确 。

A morphological erosion operator based on PDE was introduced. A new edge detecting scheme was proposed to filter gradient data directly with this new erosion operator instead of smoothing original image before edge detecting. Morphological operating will not lead to the shift of edge location so as to locate edge accurately, being robust against noise. The performance of the proposed algorithm is compared with Canny and Log operators. The experiments showed that the new edge detector had better performance in noise restraining and edge preserving.