基于BPT分割的SAR与可见光图像配准方法

SAR and Visible Light Image Registration Meth Based on the BPT Segmentation

提出一种基于BPT(二叉划分树)分割的SAR图像与可见光图像的配准方法,利用BPT的分层区域合并法将图像分成不同灰阶的ROI(感兴趣区域),分别提取重心当作控制点,然后采用独立距离作为相似性测度进行进一步的精校正,最后选择多项式模型以及双线性插值法进行匹配。实验表明:该算法较传统的边缘检测方法提取区域重心省去很多后续的形态学处理工作,具有更准确的定位,达到手动配准的精度。

Based on a kind of BFF （ binary classification tree） segmentation, SAR image and visible light image registration method was proposed. BPT of hierarchical region merging method is used to divide the image into different grayscale ROI （area of interested）. Then we extracted the center of gravity respec tively as control points, used independent distance as similarity measure for further correction, and finally chose matching polynomial model and bilinear interpolation method. Experiments show that the algorithm is to teach the traditional edge detection method to extract regional center of gravity, save a lot of subsequent morphology processing work, and have a more accurate positioning which achieves the accuracy of manual registration