抗差估计在多波束测深异常值探测中的应用

Application of robust estimation for detecting outliers of multibeam data

针对传统最小二乘趋势面滤波法在多波束测深异常值探测过程中存在的不足,提出一种基于抗差估计的选权迭代趋势面拟合滤波法,利用等价权将抗差估计理论与最小二乘趋势面拟合有机结合起来,以趋势面拟合系数求解模型作为判别异常值的迭代推值模型,介绍了多波束测深异常数据探测及滤波的基本步骤,结合仿真和实测数据采用不同滤波方案进行分析比对.研究结果表明:经选权迭代趋势面拟合法滤波后的海底地形表面较为平滑,同时微小地形得以保留,跳点以及孤立点得到剔除,滤波效果较为理想.

Considering the defects of traditional least square trend surface filtering method on detecting outliers of multibeam data, a method based on robust estimation with variable weights iteration on trend surface fitting filtering is presented, which combines robust estimation theory with least square trend surface fitting by equivalent weight. Regards the trend surface fitting coefficient solving model as the distinguish outliers iteration extrapolated model, this paper presents the basic steps for outlier detection and fitting, and filtering with different schemes under simulation and examples. The experiments show that after filtering with the method, the seafloor is smooth and the microrelief is reserved, the jump points and isolated points are rejected, and the filtering effect is ideal.