基于数值散射模拟与模型匹配的SAR自动目标识别研究

该文提出并实现了一种基于模型的SAR自动目标识别算法,该算法用实验室开发的BART进行离线电磁散射计算,系统参数设置和MSTAR数据库的参数完全一致,对待测图像和电磁散射数据所成的图像分别进行特征提取,然后进行搜索匹配。该文通过MSTAR 3类目标3种型号的实测数据和BART仿真数据分别验证了算法的可行性和准确性。该算法简单易实现,运行时间短,目标分类识别的效果较好。 

SAR Automatic Target Recognition Based on Numerical Scattering Simulation and Model-based Matching

This study proposes a model-based Synthetic Aperture Radar (SAR) automatic target recognition algorithm. Scattering is computed offline using the laboratory-developed Bidirectional Analytic Ray Tracing software and the same system parameter settings as the Moving and Stationary Target Acquisition and Recognition (MSTAR) datasets. SAR images are then created by simulated electromagnetic scattering data. Shape features are extracted from the measured and simulated images, and then, matches are searched. The algorithm is verified using three types of targets from MSTAR data and simulated SAR images, and it is shown that the proposed approach is fast and easy to implement with high accuracy.