基于SVR及阴影信息的SAR目标方位角估计

目标成像方位估计是合成孔径雷达（synthetic aperture radar，SAR）自动目标识别中一个重要的预处理过程。提出一种基于支持向量回归机（support vector machine for regression，SVR）并结合SAR目标阴影信息的方位角估计方法。首先通过长直边拟合法与脊波变换法对目标图像进行方位的粗估计．再由SVR完成精确估计。利用“运动与静止目标的获取与识别”（moving and stationary target acquisition and recognition, MSTAR）项目组提供的实测数据所做实验表明，此方法可以有效估计SAR目标方位角，精度高、泛化能力强．特别是在水平成像方位附近利用了目标的阴影信息，明显提高了相应区间的方位角估计精度。

SVR-Based SAR Target Azimuth Estimation Combined with Shadow

Imaging pose estimation is a key pre-processing step in SAR automatic target recognition. This paper presents a novel SVR-based target azimuth estimation method combined with the shadow information of target. First,target and shadow azimuth is roughly estimated by two aspect estimation approaches respectively, i.e. long-edge fitting and ridgelet transform. Then they are fused into features to train SVR estimators,which are later applied to estimate target azimuth exactly. Experimental .results with MSTAR data set show the proposed method is effective and of good generalization, especially, shadow information of target is beneficial to improve the azimuth estimation accuracy at horizontal aspect.