基于图像的海上波浪场测量算法
An Algorithm for Image-Based Measurement of Ocean Wave Field
张睿雯 1魏汉迪 2肖龙飞1
作者信息
- 1. 上海交通大学海洋工程国家重点实验室,上海 200240;上海交通大学三亚崖州湾深海科技研究院,海南三亚 572024
- 2. 上海交通大学三亚崖州湾深海科技研究院,海南三亚 572024
- 折叠
摘要
为有效解决现有波场测量方法存在的效率较低和维护成本高的问题,提出一种可靠高效的广域波场测量方法,即基于图像的海上固定结构物周围波浪场测量方法.将图像中每个像素的深度分为平均波面深度和由波浪升高引起的深度变化,通过相机位置和图像信息对二者进行测算,进而通过像素深度重构三维波浪场.将采用该方法计算所得波浪高程时历和波谱结果与采用双目定距算法和浪高仪所得结果进行对比,得到谱峰周期和有义波高等波浪统计值的平均相对误差分别为0.92％和-2.22％,证明该方法是可行和有效的.
Abstract
Current approaches to measure wave fields have low efficiency and high maintenance costs;thus,a reliable and efficient measurement method for wide-area wave fields is urgently required.An image-based method for measuring wave field around fixed maritime structures has been proposed,where the depth of each pixel is separated into the depth of the mean water surface and the depth variation induced by wave elevation,which can be measured through camera position and image information,thereby reconstructing the 3D wave field by pixel depth.The results calculated by this method are compared with the time history and wave spectrum of binocular depth estimation algorithm and wave gauge,the average relative errors of the spectral peak period and significant wave height are 0.92％and-2.22％,respectively,which proves the feasibility and effectiveness of the method.
关键词
波浪测量/深度估计/自监督学习/机器视觉Key words
wave measurement/depth estimation/self-supervised learning/machine vision引用本文复制引用
基金项目
三亚崖州湾科技城专项(SCKJ-JYRC-2022-82)
国家自然科学基金项目(52031006)
国家自然科学基金项目(42206192)
海南省自然科学基金项目(521QN275)
三亚崖州湾科技城科研项目(SKJC-2021-01-003)
上海交通大学深蓝计划(SL2020PT301)
出版年
2024
NETL
NSTL
万方数据