摘要
SAR图像与光学图像有较大差异,为实现SAR图像与光学图像的匹配,提出了一种新的多子区SAR景象匹配算法。算法依据SAR图像噪声特性,提出了用于SAR图像边缘提取的基于小波包变换的改进断面检测法。匹配中将实时图分为多个子区,由子区灰度标准差和频谱直方图判别其可匹配性,用各子区位置分布关系约束匹配结果。匹配搜索策略为基于提升小波的金字塔分层搜索。用SPOT-5卫星光学基准图进行了大样本实验,实验结果表明了该方法是一种有效的图像匹配方法,具有鲁棒性,并且与中心点匹配方法相比,能为飞行器定位提供更多有效信息。
In order to match SAR image to optical reference image, a novel algorithm of scene matching with spatial relation constrain is presented. The edge feature of SAR image is extracted via an improved contour detecting algorithm, which computes the gradient of pixels using wavelet packet transform. The real-time image is divided into multiple subarea and each subarea is evaluated by the gray level standard deviation and the distance of spectral histogram. The spatial relationship of the subarea is applied to correct the matching result. The multiple-layer search strategy based on lifting wavelet transform is used to improve the performance of the algorithm. The experimental results of matching SAR real-time image to SPOT-5 optical reference image have veri- fied the effectiveness and robustness of our algorithms, which have outperformed the algorithm of matching with center point and can provide more information for flyer locating.
出处
《系统工程与电子技术》
EI
CSCD
北大核心
2007年第6期878-882,共5页
Systems Engineering and Electronics
关键词
SAR景象匹配
边缘提取
小波包变换
频谱直方图
多子区
SAR scene match
edge extraction
wavelet packet transform
spectral histogram
multiple subarea
