基于无人机的船舶水尺图像识别系统设计  

Design of Vessel Draft Image Recognition System Based on UAV

作　　者：沈益骏 刘新 黄昱洪 刘冲伟 朱纯瑶 张悠格 沈昌宁 SHEN Yi-Jun;LIU Xin;HUANG Yu-Hong;LIU Chong-Wei;ZHU Chun-Yao;ZHANG You-Ge;SHEN Chang-Ning(Comprehensive Technical Service Center of Zhoushan Customs,Zhoushan 316000;Comprehensive Technical Center of Zhangjiagang Customs,Zhangjiagang 215600;Shengsi Customs,Zhoushan 202450;Longkou Certification&Inspection Co.,Ltd.,Longkou 265700;Nanhai Experimental School,Zhoushan 316021)

机构地区：[1]舟山海关综合技术服务中心,舟山316000 [2]张家港海关综合技术中心,张家港215600 [3]嵊泗海关,舟山202450 [4]龙口检验认证有限公司,龙口265700 [5]南海实验学校,舟山316021

出　　处：《中国口岸科学技术》2023年第12期29-35,共7页China Port Science and Technology

基　　金：杭州海关重点课题(2017-ZKZ-04);舟山市科技项目(2022C31051);南京海关科研项目(2023KJ36)。

摘　　要：为解决船舶水尺计重中存在的耗时长、准确度低、智能化程度不高等问题,设计了基于无人机的船舶水尺图像识别系统。该系统运用无人机采集船舶水尺图像,在现有图像识别技术基础上,改进算法,做到精准识别水尺图像,给出准确吃水值。该系统经水尺区域搜索、吃水线识别、水尺刻度识别及读数等必要环节,最终输出船舶吃水值,实现水尺计重自动化作业。通过验证实验,证明了该套系统真实有效。In order to solve the problems of time-consuming,low accuracy and low intelligence in vessel draft survey,a ship draft image recognition system based on unmanned aerial vehicles(UAV)is designed.The system uses UAV to capture draft image of vessels.Based on the existing image recognition technology,the algorithm is improved to accurately identify draft images and give accurate draft values.The system finally outputs the results through the necessary links such as darft marks area search,waterline identification,draft marks identification and reading,so as to realize the automatic operation of vessel draft survey.Through the verification experiment,it is proved that the system is true and effective.

关 键 词：无人机 水尺计重 图像识别 感兴趣区域(ROI) 算法 水尺读数 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] U692.73[自动化与计算机技术—计算机科学与技术]