基于二维几何特征的微地形快速识别算法

Fast Recognition Algorithm of Micro-terrain Based on 2D Geometric Features

为满足水下超声波准确、实时探测海底微地形的需要,研制出一种摆动式单波束探测系统。该系统在探测实验过程中,因出现高程奇异值而造成数字"伪"地形产生,由此提出一种基于二维几何特征的水平、斜坡和凹凸不平三种微地形模式的快速识别算法。该算法建立在现代统计分析与模糊聚类的理论基础上,具有准确性高、实时性好以及抗干扰能力强等优点;并由物理模拟实验证明,利用该算法能有效消除高程奇异值的影响,识别出来的微地形模式准确可靠,为下一步快速重构逼真的数字微地形提供了保障,能够应用于真实的海底微地形探测。

In order to explore benthonic micro-terrain by the underwater ultrasonic truly and real-timely, a pendulum single beam bathymeter was devised. Because the digital fake micro-topography was worked by the exceptional elevation values in course of detecting the simulative micro-terrain, a fast recognition algorithm of three patterns of micro-terrain, namely the level, the gradient and the uneven, based on 2D geometric features was put forward. It establishes the theory foundations of modern statistics analysis and fuzzy clustering, and has some virtues such as high veracity, good real-time and strong power of anti-jamming etc. This algorithm proved by the physical simulation test at lab, can eliminate effectively the influences of those exceptional data as well as recognize well and truly the actual micro-terrain, furthermore afford a safeguard to restructure fleetly the lifelike micro-terrain, so can be used on the actual exploration of benthonic micro-terrain.