基于地形起伏和SVM的SAR景象MSA选取

SAR scene MSA selection based on topographic relief and SVM

为提高合成孔径雷达(synthetic aperture radar,SAR)景象匹配区(matching-suitable areas,MSA)的匹配概率和匹配精度,提出基于支持向量机(support vector machine,SVM)的SAR景象MSA选取方法。首先结合SAR图像特性,选取基于数字高程模型、基于灰度和基于特征的3级特征参数;然后基于综合特征量,利用SVM训练样本数据,得到MSA决策函数。该方法充分考虑了SAR图像信息对MSA选取结果的影响及特征参数间的相关性,能够有效规划出高性能的匹配区。

In order to improve the matching probability and matching accuracy of synthetic aperture radar(SAR)scene matching-suitable area(MSA),a SAR scene MSA selection method based on support vector machine(SVM)is proposed.Firstly,digital elevation model-based,intensity-based and feature-based measure parameters are selected based on the characteristics of SAR images.Then based on the integrated feature vectors,SVM is employed to training sample data,and decision function which could separate suitable and unsuitable matching area class is obtained.This method fully considers the impact of SAR image information on the results of the selecting MSA and the correlation between the characteristic parameters.Experimental results show that this method can effectively single out high-performance MSA.