Abstract: This paper presents an automatic panoramic image stitching algorithm based on high speed railway forward motion video. According to the scene geometry properties of high speed railway in video, the algorithm firstly detects line segments and the vanishing point. And then, the appropriate sampling area is automatically constructed to achieve seamless stitching. Finally, automatic panorama stitching is realized. The experimental results show that the algorithm is faster and more accurate by comparing with the manual method to determine the sampling area.

Key words:  panorama, vanishing point, image stitching, line segments detection, railway environment

摘要： 提出了一种基于铁路环境视频的全景图自动拼接算法。根据铁路环境视频中的场景几何特性，检测场景中的线段及消失点，从而生成场景的空间布局关系；在此基础上，自动构建合适的采样区域，以达到无缝拼接的效果；最后实现全景图的自动拼接。实验结果表明，该方法实现了铁路环境视频的自动化拼接，且得到了更加令人满意的拼接效果。

关键词: 全景图, 消失点, 图像拼接, 线段检测, 铁路环境