一种基于水下机器人的构筑物场景复原优化方法  

A kind of scene reconstruction optimizationmethod for building on the basis of remotely operated vehicles

作　　者：郭云翔[1] 周军[1] 

机构地区：[1]河海大学机电工程学院,江苏常州213022

出　　处：《锻压装备与制造技术》2016年第6期118-123,共6页China Metalforming Equipment & Manufacturing Technology

基　　金：江苏省科技支撑项目(BE2012096)

摘　　要：为了克服水下构筑物场景复原过程中的累积误差,有效还原水下构筑物的表面结构信息,本文提出了一种基于水下机器人的构筑物场景复原优化方法。通过构建最大生成树,采用基本的图算法寻找基准坐标系以及完成坐标系的转换,并建立了误差方程,选取L-M迭代算法实现变换矩阵的最优化,采用对比实验实现了对该方法的可靠性检测。实验结果表明,采用该优化方法后可有效反映整个探测过程的图像信息,为后续的机器视觉处理建立了有效基础。In order to overcome the cumulative error in the reconstruction process of underwater building and effectively restore the surface structure information of underwater building, a kind of scene reconstruction optimizationmethod for building on the basis of remotely operated vehicles has been put forward in the text. The maximum spanning tree has been established. The basic graph algorithm has been used to search reference coordinate system and complete the transformation of coordinate system. The error equation has been established. The optimum transformation matrix has been realized by use of L-M iterative algorithm. The reliable detection has been conducted to this method through contrast experiment. The experimental resuits show that the image information of the whole detection process has been effectively reflected by adopting this ootimized method. It provides effective basis for follow-up machine vision processing.

关 键 词：水下构筑物 累积误差 优化方法 最大生成树 变换矩阵 

分 类 号：TP391.7[自动化与计算机技术—计算机应用技术]