基于渐进三角网的多波束滤波算法研究

Research on automatic removal of multi-beam outliers by progressive triangulation

多波束测深过程中,由于受到各种要素干扰,数据会生成大量异常值。异常值(通常称为噪声)剔除是多波束数据处理过程中的关键。当前常用的趋势面滤波算法主要是建立水下地形的曲面,对于噪声点与所建立曲面对比完成多波束噪声的过滤。针对多波束噪声剔除问题,提出了渐进三角网加密(progressive TIN densification algorithm,简称PTD)算法,选取最低水深点。利用Grubbs算法选取最低水深点,通过Delaunay三角剖分建立三角网构筑海底模型,利用三角网边长、角度与距离作为判断阈值,分离噪声点与水深点。以温州海域航道水深测量为例,与Caris软件中CUBE算法处理结果以及传统趋势面算法进行对比,验证渐进三角网加密算法的实用性和优缺点。

A large number of outliers are generated in the data processing due to the interference of various elements in the process of multi-beam bathymetry.Outliers(often referred to as noise)rejection is critical in multibeam data processing.The commonly used trend surface filtering algorithm is mainly used to establish the surface of the underwater terrain and to filter the multi-beam noise by comparing the noise point with the established surface.This paper attempts to use the form of constructing a triangular network to complete multi-beam filtering.At present,manual intervention is mainly used,and the operators′expertise significantly impacts the data results.This paper proposes a PTD aiming at the problem of multi-beam noise elimination.The Grubbs algorithm is used to select the lowest water depth point.In this paper,the seabed terrain points are used to establish a triangulation network through Delaunay triangulation to construct a seabed model.The edge length,angle and distance of the triangulation are used as the judgment thresholds to separate noise points and water depth points.Taking the measurement of channel depth in Wenzhou sea area as an example,advantages and disadvantages of this algorithm are verified by comparing with the artificial processing results of the CUBE algorithm in Caris and the traditional trend surface algorithm.