SAR/INS组合导航中基于SURF的鲁棒景象匹配算法

在SAR/INS组合导航系统中,由于合成孔径雷达采用正侧视成像工作方式,会引起SAR图像的严重变形,而且获取的SAR图像还可能存在严重的斑点噪声。为了适应SAR图像的几何畸变和高斑点噪声影响,需要提取出的图像特征具有较高的鲁棒性。本文提出了基于SURF的导航用鲁棒景象匹配算法,算法首先针对惯性组合导航的工作特点,对SURF特征匹配进行了改进和优化设计,然后用RANSAC方法过滤掉错误和低精度的匹配点,最后,进行最小二乘精确匹配获取航向和位置偏差信息。通过仿真分析了算法对SAR图像的适应性、抗斑点噪声性能,匹配精度以及实时性,并与基于SIFT特征的景象匹配算法进行了对比。仿真结果表明,所提出算法性能优越,在匹配适应性、鲁棒性、匹配精度及匹配速度方面都优于SIFT算法,可以满足SAR/INS组合导航系统图像匹配修正的高性能要求。

Robust Scene Matching Algorithm for SAR/INS Integrated Navigation System Based on SURF

In the SAR/INS integrated navigation system,the positive side-looking synthetic aperture radar(SAR) imaging methods can cause serious distortion and speckle noises of SAR images.Image features must have high robustness to adapt speckle noises of SAR image.The robust scene matching algorithm is proposed based on speeded-up robust feature(SURF) for navigation.Firstly,the process of SURF feature matching especially for the inertial integrated navigation is optimized.Then,RANSAC algorithm is used to remove the ...