According to the problem of estimation precision low to SAR imagery of vehicle target, especially 0 degrees and 180 degrees, this paper proposed a combined pose estamation method. First, analyzed the SAR imagery features of vehicle target in different pose and the advantages and disadvantages of typical targets pose estimation method. Then, through judging the shadow and area characteristics of vehicle target in SAR image and associating Hough transform method with extracting axis method, estimated the pose using the corresponding estimation strategies. Finally, took the pose estimation experiment for the MSTAR database. The results of experiment show that correctly pose estimated rate in the rang of 5° absolute error is more than 89％, that error mean value is less than 4°, and that the pose estimation algorithm is effective and feasible.%针对SAR图像地面车辆目标方位角估计精度不高、尤其是0°和180°估计误差大的问题,提出了一种地面SAR图像目标方位角联合估计方法.分析了地面车辆目标在不同角度的成像特点和典型目标方位角估计方法的优缺点,通过判断当前目标成像所具有的特点,利用目标阴影特征与目标轮廓特征,并结合目标主轴提取方法和Hough变换方法对SAR图像目标方位角进行联合估计,利用MSTAR目标切片数据对该方法进行了验证实验,绝对误差在50范围内的准确估计率均在89%以上,目标误差均值都在40以内.实验结果表明,该算法的方位角估计精度比较高,算法是有效和可行的.
展开▼
