基于单目视觉测距在调车作业中的应用研究

Research and Application of Distance Measurement Based on Monocular Vision in Shunting Operation

为实现铁路调车作业视频监控和车体间距测量,本文基于单目视觉测距技术设计调车作业中的图像测距系统,实时采集已挂车厢和待挂车厢之间的视频信息并测算距离。在Windows平台和MATLAB图像处理系统下,首先利用形态学处理技术对目标图像进行增强,对背景图像进行弱化处理,然后根据自适应阈值对图像进行二值化,利用Harris角点检测方法提取目标特征点,最后根据在序列图像中车体成像面积和距离成比例的原理,建立车体特征点与摄像机光心直线测距模型实现距离测量。现场试验结果表明,距离测量精度提高了50%,满足实际应用需求,大大改善人工瞭望进行调车作业。

In order to realize the monitoring of railway shunting operations and the measurement of vehicle body distance,this paper designs an image ranging system based on monocular vision ranging technology for shunting operations,the video information between the suspended carriage and the waiting carriage is collected in real time and the distance is measured.Under the Windows platform and MATLAB image processing system,firstly,the collected image is processed by morphology,the target image is strengthened,the background image is weakened,and then the image is binarized based on threshold,the feature points of the target are extracted by Harris corner detection method,at last,according to the principle that the imaging area and distance of the target in the sequence image are proportional,the distance measurement is realized by establishing the linear distance measurement model between the target image feature point and the camera optical center.The field experiment results show that distance measurement accuracy is improved by 50%,which meets the actual application requirements and greatly improves the manual observation for shunting operation.