基于区域生长重心的大型多阵列圆标定板基准点的提取技术

Extraction Technology of Datum Points for Large and Multi - array Circular Calibration Plate Based on Region Growing Barycenter Method

摘要:

针对视觉测量在轨道交通隧道大范围轮廓测量上的问题,研究设计一种大型多阵列圆标定板进行系统标定工作,提出区域生长重心法实现对标定图像上基准点的提取。即将区域生长法和灰度重心法相结合,通过粗定位寻找每个基准圆的种子点,考虑种子点像素的4邻域像素进行区域生长,然后利用灰度重心法提取生长区域的重心作为基准点。将非线性变化的基准点替代线性提取的基准圆的种子点,消除了图像畸变的影响,保证了特征检测过程的可靠性和鲁棒性,达到对于多阵列圆大型标定板基准点的提取。通过搭载鱼眼镜头拍摄6m的模拟隧道横断面轮廓,进行轮廓测量与还原。实验表明该方法提取的基准点像素坐标误差可达到2pixel以内,系统测量精度能够达到±5mm以内,完全满足轨道交通隧道横断面轮廓测量误差标准。

Abstract:

Towards to the problem of vision measurement on the large profile measurement of rail transit tunnel, research and design of a large and multi-array circular calibration plate for system calibration work. The method of region growing barycenter is proposed to extract the datum points on the calibration image. Combining the region growing method and the gray barycenter method, the seed point of each datum circle are searched by rough location and while 4 neighborhood pixels of the seed points’ pixel are considered for region growth. Then the gray barycenter method is used to extract the barycenter of the growing area as the datum point. In order to eliminate the effect of the image distortion, the nonlinear change datum points is substituted for the seed points of the linear datum circle. This method not only ensures the reliability and robustness of the feature detection process, but also can be used for the extraction of the datum points of the large and multi-array circular calibration plate. The profile of reduction experiment was carried out by carry a fisheye lens to shoot the simulation of tunnel cross section with 6 m. The experimental results show that the pixel coordinate error of the datum points extracted by this method can reach within 2pixel. The system precision can reach within ±5mm and fully meet the errorSstandard for the cross section measurement of rail transit tunnel.