| 多分辨率形态学目标检测 |
| |
| 引用本文： | 余莉,王润生,韩方剑. 多分辨率形态学目标检测[J]. 计算机辅助设计与图形学学报, 2006, 18(6): 849-853 |
| |
| 作者姓名： | 余莉 王润生 韩方剑 |
| |
| 作者单位： | 国防科学技术大学ATR国家重点实验室,长沙,410073;国防科学技术大学ATR国家重点实验室,长沙,410073;国防科学技术大学ATR国家重点实验室,长沙,410073 |
| |
| 摘 要： | 用形态小波变换得到图像的塔式表示,并从最低分辨率到原始分辨率由粗到精地提取目标区域.在每个分辨率上,先应用分水岭变换分割该分辨率下的低频分量（或上一级分辨率得到的标记图）,得到一个标记图;再用一个区域搜索策略来更新该标记图.对多类目标的实验结果验证了该算法具有速度快、精度高和对噪声不敏感的优点.
|
| 关 键 词： | 目标检测 多分辨率分析 形态小波变换 分水岭变换 |
| 收稿时间： | 2005-05-16 |
| 修稿时间： | 2005-09-05 |
Multiresolution Morphological Object Detection |
| |
| Yu Li,Wang Runsheng,Han Fangjian. Multiresolution Morphological Object Detection[J]. Journal of Computer-Aided Design & Computer Graphics, 2006, 18(6): 849-853 |
| |
| Authors: | Yu Li Wang Runsheng Han Fangjian |
| |
| Affiliation: | ATR National Laboratory, National University of Defense Technology, Changsha 410073 |
| |
| Abstract: | First, a pyramid representation of images is created using morphological wavelet transform. Second, a coarse-to-fine analysis scheme is implemented from the lowest-resolution to the full-resolution. At each resolution, watershed transform segments the image of the pyramid (or the previously obtained result) to generate a labeled image, which is relabeled by a region searching scheme. Experimental results with various kinds of image data demonstrate the algorithm's efficiency, precision, and robustness to noise. |
| |
| Keywords: | object detection multiresolution analysis morphological wavelet transform watershed transform |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
