光学学报, 2017, 37 (5): 0515003, 网络出版: 2017-05-05
车载前向运动视频的实时全景成像方法
Real-Time Panoramic Imaging Method for Train-Borne Forward Motion Video
机器视觉 全景成像 实时拼接 车载视频 前向运动 铁路环境 machine vision panoramic imaging real-time stitching train-borne video forward motion railway environment
摘要
传统的视频拼接模型和图像对齐方法中, 耗时的图像匹配和复杂的光流计算一直是实时拼接的性能瓶颈。提出了一种车载前向运动视频的实时全景成像方法, 该方法充分利用物理场景的几何结构和列车的运动信息来构造视频的拼接区域, 并建立几何模型实现了拼接区域的对齐。整个拼接避免了复杂的图像处理过程, 实现了铁路全景图的实时获取, 同时也提供了一种轻量级的视频全景索引方式, 降低了视频数据的存储和访问开销。
Abstract
In the traditional video mosaic model and image alignment method, the time-consuming image matching and complicated optical flow computation are the performance bottleneck of real-time mosaic. A real-time panorama imaging method is proposed, which makes use of the geometric structure of the physical scene and the motion information of the train to construct the mosaic region of the video and establishes the geometric model to achieve the alignment of the mosaic region. The entire stitching avoids the complex image processing, realizes the railway panorama real-time acquisition and also provides a lightweight video panoramic index method, which reduces the storage and access cost of the video data.
蒋欣兰, 王胜春, 罗四维, 黄雅平, 戴鹏. 车载前向运动视频的实时全景成像方法[J]. 光学学报, 2017, 37(5): 0515003. Jiang Xinlan, Wang Shengchun, Luo Siwei, Huang Yaping, Dai Peng. Real-Time Panoramic Imaging Method for Train-Borne Forward Motion Video[J]. Acta Optica Sinica, 2017, 37(5): 0515003.