基于移动双目视觉的铁路建筑物限界快速自动检测方法

Fast and Automatic Railway Building Structure Clearance Detection Technique Based on Mobile Binocular Stereo Vision

采用高速、高分辨率CCD相机以优于10-6s时间同步精度获取铁路建筑物限界的移动双目立体图像。提出了基于移动双目视觉测量模型,通过自检校多传感器系统标定和DGPS/IMU集成处理计算各双目图像严密的内、外方位元素,构建立体图像中各要素的空间相对关系。提出基于自动拟合计算的线路中心线空间位置的建筑物限界计算模型,与车辆限界相结合计算建筑物限界参数。设计并开发了一个基于双目视觉的铁路建筑物限界自动检测试验系统,进行铁路建筑物限界自动检测试验,对其精度进行分析和讨论,结果表明可达到标准规定的技术指标。

The synchronization imaging technique with high speed and high resolution CCD cameras is presented to obtain the mobile binocular stereo image of the clearances of railway building structures. With the unit timing board, the binocular images can obtain synchronized time to GPS/IMU sensors less than 10-6 s. The photogrammetry surveying model based on mobile binocular vision is proposed. With the sensor calibration parameters based on self-calibration and the train position and attitude from DGPS/IMU integration, all the objects in the stereo images can be restricted within a precise formulation. The intelligent image processing algorithm is designed to automatically extract railway from the binocular sequence images to calculate the railway centerline. The railway building structure clearance calculation model with the railway centerline is presented to calculate the building constructure clearance parameters within the railway clearance standards. The experiment system of railway building structure clearances is designed with the proposed methodology for project applications. The accuracy of building structure clearance detection is less than 0.03m which satisfies the railway clearance measurement standards.