基于Coherent Point Drift方法的SAR图像散射中心匹配

散射中心匹配是当前散射中心用于SAR图像目标识别的一个主要技术途径.散射中心匹配的难点在于散射中心特征存在的误差和缺失.Coherent Point Drift (CPD)方法从概率密度估计的角度解决点模式匹配问题,能够较好地考虑散射中心的误差和缺失.本文将CPD方法用于散射中心匹配,并在此基础上引入车辆目标SAR图像方位角估计先验信息和散射中心属性信息,以提高散射中心匹配的准确性和稳健性.MSTAR数据实验说明了该方法的有效性.

Coherent point drift method for SAR imagery scattering centers matching

Scattering centers matching is currently the primary technique in SAR Automatic Target Recognition(ATR) using scattering centers.The difficulties in scattering centers matching originate in the inexactness of scattering centers and missing scattering centers.The recently developed Coherent Point Drift(CPD) method solves point pattern matching from the perspective of probability density estimation.In the CPD method,the inexactness of scattering centers and missing scattering centers can be well considered.The CPD method was introduced in this research to solve scattering centers matching.In order to improve the accuracy and stability of scattering center matching,the CPD method was modified to incorporate the prior information of vehicles’ azimuth estimation and scattering centers attributes.Experiments on MSTAR dataset show the validity of the proposed method.