| 一种基于偏微分方程的形态学边缘检测新方法 |
| |
| 引用本文： | 蔡超,刘买利,丁明跃,周成平. 一种基于偏微分方程的形态学边缘检测新方法[J]. 华中科技大学学报(自然科学版), 2003, 31(10): 1-3 |
| |
| 作者姓名： | 蔡超 刘买利 丁明跃 周成平 |
| |
| 作者单位： | 1. 中国科学院武汉物理与数学研究所,波谱与原子分子物理国家重点实验室
2. 华中科技大学图像识别与人工智能研究所,教育部信息处理与智能控制重点实验室 |
| |
| 基金项目： | 国家自然科学基金资助项目 (6 0 1 35 0 2 0FF0 30 4 0 5 )，中国科学院武汉物理与数学研究所波谱与原子分子物理国家重点实验室基金资助项目 (T1 5 2 1 0 1 ) |
| |
| 摘 要： | 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，具有更好的抗噪性能．
|
| 关 键 词： | 边缘检测 数学形态学 偏微分方程 |
| 文章编号： | 1671-4512(2003)10-0001-03 |
| 修稿时间： | 2003-03-06 |
Morphological edge detector based on partial differential equations |
| |
| Cai Cao Liu Maili Ding Mingyue Zhou Chengping Cai Cao Lect, Wuhan Institute of Physics and Mathematics the Chinese Academic of Science,Wuhan ,China.. Morphological edge detector based on partial differential equations[J]. JOURNAL OF HUAZHONG UNIVERSITY OF SCIENCE AND TECHNOLOGY.NATURE SCIENCE, 2003, 31(10): 1-3 |
| |
| Authors: | Cai Cao Liu Maili Ding Mingyue Zhou Chengping Cai Cao Lect Wuhan Institute of Physics Mathematics the Chinese Academic of Science Wuhan China. |
| |
| Affiliation: | Cai Cao Liu Maili Ding Mingyue Zhou Chengping Cai Cao Lect, Wuhan Institute of Physics and Mathematics the Chinese Academic of Science,Wuhan 430071,China. |
| |
| Abstract: | 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. |
| |
| Keywords: | edge detection mathematical morphology Partial Differential Equation |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
