钻孔摄像机器人视频序列拼接

Mosaics of video sequences based on the borehole televiewer robot

针对检测过程中视频序列展开拼接过程中标定难、且匹配效率低及耗时长的缺点,提出了一种新的钻孔内壁视频序列拼接的方法.首先对读取的视频均匀采样,得到视频序列;然后对取样的图像帧进行圆心提取以及环带图像选择,完成环带图像展开,重投影为矩形图像;采用棋盘格标定的方法,设置畸变校正参数,对展开后的矩形图像进行畸变校正;采用基于模板匹配的特征匹配策略进行拼接,得到钻孔内壁展开图像;最后根据内壁图像,生成立体内壁图像模型.结果表明该方法数据量小且处理效率高,满足可视化需求,可用于岩体的质量评价,保障施工安全.

For a commonly used method for rock detection, this paper proposes a new one that uses a borehole televiewer robot for video shooting inside the borehole. Based on the video-shooting data, this paper designs a related vision method to process it. After loading the video, the method extracts the image frame at a certain time interval to get the target video sequence. The sampled fame will be extracted at its enter and re-projected from the belt image to the rectangular image. This paper uses a method based on template matching to match the adjacent images, figuring out the inner-wall expanded image of the borehole. According the expanded image, the system can build a 3D model for the borehole to evaluate the construction safety.