Research On Key Technology Of Target Classification Based On Ashore And Underwater Integrated Measurement System

Marine surveying and mapping is the basic work of the state development of the ocean.The surveying and mapping targets include islands(reefs),coastal zones and port fairways.In the application,the classification of mapping targets is a constructive result of surveying and mapping.Form by the identification of objects and underwater landform classification,classification of objectives play a positive reference role in the marine infrastructure construction,resource development and management.In the context of the implementation of the strategy of ocean power,the importance of marine surveying and mapping work has become increasingly prominent,mapping equipment is also constantly updated,multi-beam and ship borne laser scanner to be promoted.Using high-density data obtained by multi-beam and laser scanner,combined with the space and attribute information data in the target,the target classification,the integration of the two can play a greater role,but also one of the current trend of marine mapping.This study focuses on the use of integrated equipment for accurate and rapid classification of ashore and underwater target classification,mainly to the following research work:(1)This study systematically introduces the composition and principle of multibeam system and ship-borne laser scanning system and the research status of the system in target classification.On the problem of system assembly and integration,the solution of system space and time synchronization problem is given by using the software rigid installation parameters and the PPS time synchronization pulse data in GPS by means of commercial software QINSy.(2)In the classification study of multi-beam,the idea of clustering analysis is introduced,and the classification model based on multi-beam depth and echo intensity is established.The multi-beam data is pre-processed in advance,including the correction of coarse errors and system errors,to achieve the purpose of homing the multi-beam point cloud in order to better classify.In the specific classification method,the multi-beam echo intensity is first processed by using the piecewise function(linear function and Lambert function),and then the K-means clustering analysis method is used to analyze the echo intensity and the water depth corresponding to its coordinates As the feature quantity,the average value and the variance are respectively classified as the characteristic parameters of the clustering,and each of the clustering results has similar spatial and attribute characteristics.In the experiment of this study,after comparison analysis,when K = 5,the classification effect is better.(3)In the classification of laser,based on the spatial geometric characteristics of point cloud,an automatic classification method is proposed,and the laser point cloud is classified by two methods: height difference threshold method and stratified density method.The Prior to the classification of the pretreatment work is mainly point cloud filtering,using TIN based on the progressive encryption algorithm,the ground point and non-ground point separation,based on TIN progressive encryption algorithm in the experimental flat terrain effect is very good.In the non-ground point on the point cloud for identification: first with the height difference threshold method will meet the conditions of low vegetation and tall buildings separated,and then use the stratification density method will be separated between the two points cloud,Divided into two types of general trees and low construction.The method is simple to calculate and has a good classification effect in typical terrain(low vegetation,general trees,tall buildings).