| 摘要: |
| 运用水下机器人(ROV、AUV)配备的高清数码相机拍摄深海底的视频图像(视频、照片)有其自身的特点，如获取难度较高、覆盖面积小、光照不均匀、特征点少、数据量大等。为了获取整体大范围的海底影像，使海洋科研人员更直观、全面地认识海底微地形地貌，对其底质状况进行分析，以更详细地了解海洋底栖生物的生态环境，需要对这些拍摄的海底视频图像进行拼接处理。本文提出一种基于水下机器人的视频图像拼接流程和处理方法，首先对水下机器人拍摄的海底视频数据进行关键帧提取，得到带时间戳的序列图像；然后导入姿态数据中的偏航角对序列图像进行方位角恢复；最后基于定位数据批量地对经过方位角处理后的序列图像进行拼接和融合匀色，最终获得拍摄范围内的海底全景拼接图像。基于本文提出的图像拼接方法，对某科考航次中获取的视频图像进行拼接处理实验，取得了较好的拼接效果，验证了本文提出的方法的可行性。 |
| 关键词: 水下机器人 海底探测 视频图像 图像拼接 匀色 |
| DOI：10.13656/j.cnki.gxkx.20230710.013 |
| 投稿时间：2022-10-29修订日期：2022-12-14 |
| 基金项目:中国科学院战略性先导科技专项项目(E0B435101T，XDA22050502)，东华理工大学博士启动基金项目“基于水下机器人的深海底视频影像拼接关键技术研究”(DHBK2019196)和东华理工大学江西省数字国土重点实验室开放研究基金项目(DLLJ202005)资助 |
|
| Discussion on Seabed Video Image Mosaic Technology Based on Underwater Robot |
|
Xiansan LIU1,2, Zhendong LUAN2,3, Duan HUANG1,2
|
| (1.Key Laboratory for Digital Land and Resources of Jiangxi Province, East China University of Technology, Nanchang, Jiangxi, 330013, China;2.School of Surveying and Geoinformation Engineering, East China University of Technology, Nanchang, Jiangxi, 330013, China;3.Key Laboratory of Marine Geology and Environment, Institute of Oceanography, Chinese Academy of Sciences, Qingdao, Shandong, 266071, China;4.Center for Ocean Mega-Science, Chinese Academy of Sciences, Qingdao, Shandong, 266071, China) |
| Abstract: |
| The use of high-definition digital cameras equipped with underwater robots (ROV, AUV) to capture video images (videos, photos) of deep seabed has its own characteristics, such as high difficulty in obtaining, small coverage area, uneven illumination, few feature points, and large amount of data.Obtaining a wide range of seabed images can help marine researchers understand the seabed micro-topography more intuitively and comprehensively, analyze the sediment conditions, and understand the ecological environment of marine benthic organisms in more detail, it is necessary to mosaic these captured seabed video images.In this article, a video image mosaic flow and processing method based on underwater robot is proposed.Firstly, the key frame is extracted from the seabed video data captured by the underwater robot, and the sequence image with time stamp is obtained.Then, the yaw angle in the attitude data is imported to recover the azimuth angle of the sequence image.Finally, based on the positioning data, the sequence image after azimuth processing is stitched and fused in batches, and the panoramic mosaic image of the seabed within the shooting range is finally obtained.Based on the image mosaic method proposed in this article, the video images obtained in a scientific expedition are stitched, and a good stitching effect is obtained, which verifies the feasibility of the method proposed in this article. |
| Key words: underwater robot seabed detection video image image mosaic uniform color |
