摘要
不同于空气成像,水下全景受水压、暗光影响比较大,需要特殊的镜头封装和聚焦型水下照明,而水下多路图像处理可以借鉴、调整已有的图像处理算法,以解决偏色、噪点等成像问题;且基于全景视频传输的热点研究,水下ROV全景摄像系统有较好的发展空间。基于对水下全景摄像技术的研究,文章首先详细介绍了水下全景摄像的生成和传输方式,并以ROV平台为载体,从相机硬件的需求分析,说明分布式、集成式两种相机布局;接着,针对获得的多路图像数据,通过图像预处理、拼接融合生成全景图像;最后,考虑水下水上的远程视频传输,介绍了全景视频基于切片流(Tile流)划分的自适应多速率传输方案。从水下全景摄像技术的发展来看,随着图像无损拼接、影像处理器计算速度更迭、图像复原、超宽带连接和全景视频传输等技术的提高,ROV平台全景摄像系统将得到广泛应用。
Unlike imaging in air,underwater panoramic imaging faces great challenges arising from water pressure and diminished light,necessitating special lens packaging and focused underwater illumination.The processing of underwater images from multiple channels necessitates adaptations to existing image processing algorithms to mitigate color cast,noisy points,and other imaging defects.And based on the hot research of panoramic video transmission,the underwater panoramic imaging system on the ROV platform holds great development potential.Based on the technical research of underwater panoramic photography,this paper first elaborates on the generation and transmission of underwater panoramic images and discusses two camera layouts:distributed and integrated,utilizing the remotely operated vehicle(ROV)platform as the carrier,through an analysis of camera hardware requirements.Subsequently,it introduces the creation of panoramic images from multiple channels of image data through a process involving image preprocessing,stitching,and fusion.Lastly,taking into account of remote video transmission through water boundaries,a self-adaptive multi-rate transmission scheme is presented,based on the stream division into slices(Tile stream)for panoramic videos.From the development of underwater panoramic technology,with the improvement of panoramic technologies such as lossless image stitching,iterative computing speed of image processors,image restoration,ultra bandwidth connection,and panoramic video transmission,ROV platform panoramic camera systems will be widely applied.
作者
郭旭
涂绍平
徐蕾
GUO Xu;TU Shaoping;XU Lei(CRRC SMD(Shanghai)Ltd.,Shanghai 201306,China)
出处
《控制与信息技术》
2024年第1期109-115,共7页
CONTROL AND INFORMATION TECHNOLOGY
基金
国家重点研发计划项目(2022YFC2806901)。
