摘要
为了解决以双目视觉为主导的自主式水下机器人(AUV)对接回收过程中出现的单一摄像机遭受遮挡或因水下环境特殊而发生故障的问题,提出一种基于单双目切换的AUV视觉定位方法。该方法通过自适应阈值法区分图像中的光源与背景,通过中心加权质心检测法提取光源中心,并以光源中心点数量为判定条件选取单目或双目视觉定位,添加水体折射参数提高定位精度。试验结果显示,采用该方法的定位耗时相比以往减少约70%,定位结果输出稳定,对接精度在允许范围内。该方法在AUV回收对接过程中通过牺牲部分定位精度来提高对接的实时性和鲁棒性,保障AUV成功对接回收。
In order to solve the problem that a single camera is blocked or fails due to special underwater conditions in the docking and recovery process of autonomous underwater vehicle(AUV)dominated by binocular vision,an AUV visual positioning method based on monocular and binocular switching is proposed.In this method,the adaptive threshold method is used to distinguish the light source and background in the image,and then the center of the light source is extracted by using the center weighted centroid detection method.Taking the number of light source center points as the judgment condition,monocular or binocular visual positioning is selected,and water refraction parameters are added to improve the positioning accuracy.The experimental results show that the positioning time is reduced by about 70%,the positioning result output is stable,and its accuracy is within the allowable range of docking.This method sacrifices part of the positioning accuracy in the process of AUV recovery and docking,so as to improve the real-time and robustness of docking and ensure the successful docking and recovery of AUV.
作者
朱志鹏
张晓文
韩祥
ZHU Zhipeng;ZHANG Xiaowen;HAN Xiang(Marine Design and Research Institute of China,Shanghai 200011,China)
出处
《船舶工程》
CSCD
北大核心
2023年第2期131-139,144,共10页
Ship Engineering
基金
国防基础科研计划(JCKY2017414C002)
关键词
自主式水下机器人(AUV)
视觉定位
单双目切换
图像处理
AUV回收对接
autonomous underwater vehicle(AUV)
visual positioning
monocular and binocular switching
image processing
AUV recovery docking
作者简介
朱志鹏(1996—),男,硕士、助理工程师。研究方向:水下机器人精确自主对接技术。
