摘要
合成孔径雷达(SAR)图像自动配准问题是SAR图像解译的关键技术之一,长期以来未有得到很好的解决。SIFT(Scale-invariant feature transform)可以通过检测局部特征对图像进行配准,是一种比较成功的自动配准算法,但由于场景的复杂性,单纯的SIFT算法不能解决相似特征不同位置的问题,本文针对SAR图像的特点,基于Delaunay三角拓扑理论,提出了一种SIFT-Delaunay编码的自动配准算法,通过实测数据试验,该算法能够解决SAR图像仿射变换中SIFT算法配准中的问题,实现SAR图像自动配准。
SAR image automatic registration problem is one of key technology in SAR image interpretation and this problem is not well solved for long time. SIFT algorithm is a kind of effective automatic registration algorithm. Because of the complexity of scene, SIFT algorithm can＇t solve the problem of similar feature at different positions. In this paper, based on Delaunay triangle topology theory, a kind of automatic registration algorithm based on SIFT-Delaunay coding is proposed. According to the experimental result on measured data, this algorithm can solve the problem of SIFT algorithm registration in SAR image affine transformation, and accomplish SAR image automatic registration.
出处
《现代雷达》
CSCD
北大核心
2015年第4期20-23,30,共5页
Modern Radar
基金
国家自然科学基金资助项目(61172118)
中国博士后面上基金支持项目(2012M521830)
江苏省博士后基金支持项目(C671201026C)
