首页 | 期刊介绍 | 编委会 | 道德声明 | 投稿指南 | 常用下载 | 过刊浏览 | In English
引用本文:
【打印本页】   【HTML】   【下载PDF全文】   查看/发表评论  下载PDF阅读器  关闭
←前一篇|后一篇→ 过刊浏览    高级检索
本文已被:浏览 338次   下载 0  
分享到: 微信 更多
多波束声呐水体影像沉船自动提取方法
李东辉, 郑双强, 陈建兵, 刘洪霞, 阳凡林
山东科技大学
摘要:
多波束声呐系统记录的水体数据携有覆盖扇区内完整的时序观测信息,具有三维成像能力,被广泛应用于水下目标探测。目前水体数据分析与目标识别主要采取人工浏览、选定的方式,耗时且无法保证结果的可重复性,限制了数据的利用率和定量分析能力。通过分析水体数据的成像机理,提出一种基于单帧水体影像自动提取沉船目标的算法,利用噪声抑制、数学形态学方法检测目标边缘,根据海底统计特性识别海底,再结合深度特性实现海底与沉船目标分离。文中使用实测数据验证了本文算法,且抗噪性较强。通过水体影像提取的目标信息能够弥补水深点目标模型分辨率不足的缺点,为后续目标三维精细建模奠定基础。
关键词:  多波束系统  水体数据  水体影像  目标提取  边缘检测
DOI:10.11759/hykx20180722001
分类号:
基金项目:国家重点研发计划项目;山东省重点研发计划项目
Automatic extraction of wreck based on multibeam water column image
lidonghui, zhengshuangqiang, chenjianbing, liuhongxia, yangfanlin
Shandong University of Science and Technology
Abstract:
The water column data recorded by MBES has a complete time series information covering the sector, and has the ability of three-dimensional imaging. It is widely used in underwater object detection. At present, water column data analysis and target recognition mainly take the form of manual browsing and selection, which is time-consuming and can not guarantee the repeatability of the results, limiting the utilization of data and the ability of quantitative analysis. By analyzing the imaging mechanism of water column data, an algorithm for automatic extraction of wreck based on single Water Column Image (WCI) is proposed. After detecting the edge of the object by noise suppression and mathematical morphology, the seafloor, which needs to be removed from WCIs with depth, can be identified according to the statistical characteristics. The efficiency of the method has been assessed with the measured data, and its noise immunity is strong. The object information extracted from the WCI can make up for the drawback of the insufficient resolution of the depth model, and lay the foundation for the detailed three-dimensional modeling of the object.
Key words:  multibeam echosounder system  water column data  water column image  object detection  edge detection
版权所有 《海洋科学》 Copyright©2008 All Rights Reserved
主管单位:中国科学院 主办单位:中国科学院海洋研究所
地址:青岛市市南区福山路32号  邮编:266071  电话:0532-82898755  E-mail:marinesciences@qdio.ac.cn
技术支持:北京勤云科技发展有限公司