SAR海洋图像舰船尾迹检测和定位方法

SAR海洋图像中尾迹线性特征的检测和精确定位是个难点。常规的检测方法是利用Radon变换,在参数空间精细离散化时,可以达到较高的检测和定位精度,但计算量很大。针对这个问题,根据Radon变换的基本特性,提出了一种基于峰值波形多尺度匹配滤波的定位方法。该方法除了可以提供更加可靠的检测外,还可以使得估计的线性特征参数达到亚像素级精度,从而解决了由于精细离散化带来的大计算量的问题。仿真和实际数据处理的结果表明,该方法在参数估计误差和标准差方面明显优于其它算法。

Method of ship wake detection and positioning in SAR ocean imagery

The problem of the detection and accurate positioning of ship wake linear feature in SAR ocean imagery is one of great difficulty.The conventional method used to detect the linear feature is Radon transform,which can achieve higher detection and positioning accuracy at the expenses of increased computational burden in the case of fine discretization.According to the basic characteristics of Radon transform,an algorithm based on multiscale matched filtering of peak waveform is proposed.Besides providing more reliable detection,the method enables to estimate the linear feature parameters up to sub-pixel level resolution,thus overcomes the computational problem of very fine discretization.Simulation and real data processing results show that the proposed method is superior to others in the parameter estimation error and its standard deviation.