一种模拟纠正影像的SAR立体影像匹配方法

SAR stereo image matching method based on simulated correction image

针对常规SAR影像匹配方法在地形起伏较大区域无法获取理想的SAR立体影像匹配结果的问题,该文提出了一种基于模拟纠正影像的SAR立体影像匹配方法。利用观测区域粗分辨率DEM进行SAR影像模拟,进而获取几何纠正SAR影像,在几何纠正SAR影像上实现影像匹配,等效于常规基于灰度影像匹配中进行的影像粗配准工作,并且相比于常规粗配准,该等效粗配准过程能够均衡地顾及到整幅影像的像元,最终实现了地形起伏较大区域的高精度SAR立体影像匹配。采用COSMO-Skyped影像数据进行了实验,实验结果表明,匹配精度达到了3个像素左右,比常规的基于灰度影像匹配方法有较大提高,能够有效解决地形起伏较大区域SAR立体影像匹配问题。

Aiming at the proplem that common SAR stereo image matching method can not obtain ideal stereo matching result in the large terrain area, a new method that SAR stereo matching based on simula- ted correction image was proposed in this paper. Geometric correction image was got by using auxiliary DEM for image simulation, and then the geometric correction image was matched, this process was equiv- alent to the coarse registration of common image matching method and had a better effect because it could get a better result of image matching by taking every pixel of the image into account. The proposed method was tested by COSMO-Skyped images, the experimental result showed that this method could get 3 pixel or so precision matching result, which was better than common methods and solved the matching problem of large terrain area efficiently.