基于边缘特征的多源图像分层匹配算法研究

The hierarchical image matching algorithm based on edge features for multi-source images

针对灰度相关的匹配算法不能适用于多源图像的匹配问题,提出了一种基于边缘特征的分层匹配算法.该算法首先提取出两种非同源图像的边缘特征作为匹配的特征空间,并且为了提高匹配算法的速度,主要采取了以下几种措施:采用粗匹配和精匹配相结合的分层序贯相似度检测算法(SSDA)作为搜索策略;在粗匹配阶段运用跳跃式的搜索策略和亚抽样模板.然后在精匹配阶段为了兼顾匹配的精度采用逐点扫描的全像素点匹配.通过对真实合成孔径雷达(SAR)图像和光学图像的仿真,结果表明该算法能够较好地适用于多源图像的匹配,并且大幅提高了匹配算法的速度.

Aiming at the problem that the gray-level correlation matching algorithm can not be applied to the image matching of multi-source images, a hierarchical image matching algo- rithm based on edge features is proposed. At first, in this algorithm, the edge images of the different source images are extracted to be as the feature space of the image matching, and moreover, in order to raise the speed of the matching algorithm, several measures are adopt ed. firstly, the hierarchical sequence similarity detection algorithm（SSDA） combining of the rough matching and the accurate matching is used as the search strategy; secondly, in the rough matching stage, the pixel-jump searching strategy and the sub-sample template are a- dopted. Then, in the accurate matching stage, in order to take into account the accuracy of the matching algorithm, the pixel-by-pixel scanning searching algorithm is brought into use. At last, this matching algorithm is tested on the real synthetic aperture radar （SAR） image and the optical image through the simulation experiments. The results show that this matc- hing algorithm is suitable for the multi-source images matching, and also highly raise the matching speed faster.