{"title":"Design and Realization of Underwater Target Vision 3D Reconstruction System","authors":"Bing Li, Hengtao Ma, Jiashuai Li","doi":"10.1109/ICMA54519.2022.9856180","DOIUrl":null,"url":null,"abstract":"In order to meet the needs of high precision and fine 3D digitization of underwater targets, a visual 3D reconstruction system for underwater targets is designed. The system consists of underwater image sequence acquisition, underwater image preprocessing and 3D reconstruction. The acquisition of image sequences is based on the dynamic positioning platform with visual positioning and underwater camera. The underwater image preprocessing adopts the underwater image enhancement algorithm based on dark channel prior and color balance fusion to process the underwater image, improve the clarity and contrast of the sampled image, and correct the color deviation problem of the underwater image. The motion structure recovery and multi-view stereo vision algorithm are used to recover the three-dimensional point cloud information of spatial objects from two-dimensional images. The underwater image sequence is reconstructed by COLMAP + OpenMVS + Meshlab to obtain the reconstructed model data. The experimental results show that the designed underwater target visual three-dimensional reconstruction system has good three-dimensional reconstruction accuracy. After the underwater image enhancement method proposed in this paper, the measurement error of the reconstructed model is reduced from 5.3 % to 3.0 %, and the details of the model are more abundant. The accuracy and completeness of the reconstructed model are improved. This lays a foundation for large-scale underwater geological exploration and the establishment of large-scale underwater three-dimensional digital map.","PeriodicalId":120073,"journal":{"name":"2022 IEEE International Conference on Mechatronics and Automation (ICMA)","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2022-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2022 IEEE International Conference on Mechatronics and Automation (ICMA)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/ICMA54519.2022.9856180","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
Abstract
In order to meet the needs of high precision and fine 3D digitization of underwater targets, a visual 3D reconstruction system for underwater targets is designed. The system consists of underwater image sequence acquisition, underwater image preprocessing and 3D reconstruction. The acquisition of image sequences is based on the dynamic positioning platform with visual positioning and underwater camera. The underwater image preprocessing adopts the underwater image enhancement algorithm based on dark channel prior and color balance fusion to process the underwater image, improve the clarity and contrast of the sampled image, and correct the color deviation problem of the underwater image. The motion structure recovery and multi-view stereo vision algorithm are used to recover the three-dimensional point cloud information of spatial objects from two-dimensional images. The underwater image sequence is reconstructed by COLMAP + OpenMVS + Meshlab to obtain the reconstructed model data. The experimental results show that the designed underwater target visual three-dimensional reconstruction system has good three-dimensional reconstruction accuracy. After the underwater image enhancement method proposed in this paper, the measurement error of the reconstructed model is reduced from 5.3 % to 3.0 %, and the details of the model are more abundant. The accuracy and completeness of the reconstructed model are improved. This lays a foundation for large-scale underwater geological exploration and the establishment of large-scale underwater three-dimensional digital map.