基于SAR图像的自动目标识别系统设计与实现

设计并实现了一套完整的、基于合成孔径雷达（SAR，synthetic aperture radar）图像的自动目标识别（ATR，automatictargetrecognition）系统。该系统包括降相干斑处理、恒虚警率（CFAR，con-stantfalsealarmrate）检测、强散点聚类与鉴别、感兴趣区域（ROI，rangeofinterest）分割和基于模板的最大相关匹配等五大部分。SAR图像匹配模板库来自于对典型目标的电磁仿真，待检测SAR图像来自某机载SAR系统对某地面区域的成像实验。测试表明所介绍的ATR系统在满足工程性能要求的意义上，实现了在典型的地貌环境下对典型军事目标的实时自动识别。

Design and Implementation of SAR Image Based Automatic Target Recognition System

An automatic-target-recognition(ATR) system based on synthetic aperture radar(SAR) image processing is presented.The integrated system is comprised of five parts: wavelet de-noising of raw image,constant false alarm rate(CFAR) detection,speckle clustering and discrimination,segmentation of region of interest(ROI) followed by template based optimum correlation matching.The templates used in the matching process are computed from scattering data of the targets through electromagnetic simulation,while the measured SAR images are generated by an airborne SAR system in ground imaging experiments.This ATR system has reached the goal of real time recognition of military targets of interest in typical land environments in the sense of satisfying engineering requirements. SAR;ATR;de-noise;CFAR;cluster;image segmentation