基于三维视频融合的智慧港口全景建模研究
Study on Panoramic Modeling of Smart Port Based on 3D Video Fusion
陈立家 1王莹涛 1李胜为 1许毅2
作者信息
- 1. 武汉理工大学航运学院,湖北 武汉 430063;武汉理工大学内河航运技术湖北省重点实验室,湖北 武汉 430063
- 2. 武汉理工大学计算机与人工智能学院,湖北 武汉 430063
- 折叠
摘要
针对港区监控视频图像之间联动性差与缺乏三维立体视觉感受等问题,提出一种基于三维视频融合的港口全景建模方法.首先,基于倾斜摄影对港区进行三维重建,在重建过程中利用体素滤波法处理点云数据,以构建轻量化模型;其次,提出一种改进SURF(Speeded-Up Robust Features)算法进行图像相似配准,以求解相机姿态,实现三维模型与视频的融合;最后,以武汉某港口为实验区域,基于Cesium开源引擎实现了三维全景模型可视化渲染.结果表明,上述方法可以将港口监控视频流以纹理的形式投影到实景模型上,构建含有实时视频数据的三维全景模型,解决港口监控现存问题,为实现智慧港口立体监控提供一种新思路.
Abstract
Aiming at the problems of poor linkage between surveillance video images in port areas and lack of three-dimensional visual experience,a port panoramic modeling method based on three-dimensional video fusion is proposed.Firstly,based on oblique photography,the port area is reconstructed in 3D,and the point cloud data is pro-cessed by the voxel filtering method during the reconstruction process to construct a lightweight model.Secondly,an improved SURF(Speeded-Up Robust Features)algorithm is proposed to perform image similarity registration to solve the camera pose and realize the fusion of three-dimensional models and videos.Finally,taking a port in Wuhan as the experimental area,the visual rendering of the three-dimensional panoramic model is realized based on the Ce-sium open-source engine.The results show that the proposed method can project the port surveillance video stream into the real scene model in the form of texture,construct a three-dimensional panoramic model with real-time video data,solve the existing problems of port monitoring,and provide a new idea for realizing the three-dimensional moni-toring of smart ports.
关键词
三维重建/三维视频融合/图像相似匹配/全景建模Key words
3D reconstruction/3D video fusion/Image similarity matching/Panorama modeling引用本文复制引用
基金项目
国家重点研发计划项目(2019YFB1600603)
出版年
2024
NETL
NSTL
万方数据