摘要
合成孔径雷达(synthetic aperture Radar,SAR)影像的精确配准是使用SAR影像准确分析矿区变形的前提。虽然目前已有的影像配准算法很多,但是直接应用于SAR影像配准的效果还不够理想。为此,提出了一种集成互补不变特征的配准方法。该方法首先利用Canny边缘检测算法对影像进行区域分割,利用获得的分割区域进行粗匹配,然后应用改进Canny边缘特征的SIFT算法进行精匹配,最终获得精确配准的SAR影像。该方法能够降低由单纯使用SIFT特征进行匹配所产生的巨大计算量。通过实验分析可知,该方法能够准确配准矿区SAR影像,为矿区变形分析和综合治理提供高质量的影像数据。
The accurate synthetic aperture Radar (SAR) image registration is the prerequisite for exact analysis of mine deformation. Many image registration algorithms have been proposed, but the results are not satisfactory when these registration algorithms are directly applied to SAR image. In view of such a situation, the authors put forward an integrated registration approach in this paper. The first step of this approach is the coarse matching with Canny edge for region division; then the fine matching is performed by SIFT algorithm with improved Canny edge features; finally, the accurate registration SAR image is obtained. This approach has fewer computations than that simply using SIFT feature matching. Experimental analyses with SAR images demonstrate the efficiency and accuracy of this approach for mine SAR image registration, which provides high -quality image data for comprehensive management in mining areas.
出处
《国土资源遥感》
CSCD
北大核心
2014年第1期52-56,共5页
Remote Sensing for Land & Resources
基金
欧空局CAT-1项目(编号:8371)
江苏省博士后科研资助计划项目(编号:1101109C)共同资助