基于两平行线及其线上三点的摄像机标定方法

Camera calibration based on two parallel lines and three collinear points

在大视场交通监控摄像机的标定中,由于不便于放置大尺寸标定靶标,常采用交通标线构成的特定几何图形替代标定靶标,但许多交通场景仅有一组平行的车道分界交通标线,难以获得传统标定方法所需的标定约束条件。针对此类交通场景,选取两平行交通标线和其中一条线上的三点作为标定参考物,依靠其中一条线上三点在单幅图像中的像素坐标、三点间的两个距离值及平行线间距作为约束条件来完成摄像机的标定。建立了图像坐标系与世界坐标系之间新的换算关系,推导出了摄像机内外参数的求解公式,探讨了旋转角对测量误差的影响。实验结果表明,在约束条件不足的情况下能够获得较好的摄像机参数近似解,沿道路方向上的测距精度优于对比文献的结果,具有标定参考物选取容易、操作简便、几何约束条件少等优点,适用于交通监控系统中大视场摄像机的快速标定场合。

Due to the characteristics of the traffic scenes, it is difficult to place calibration patterns in order to calibrate the roadside camera. Traffic lane marks are usually substituted for calibration patterns. However, there is only a set of parallel lane dividing lines in many traffic scenes. Calibration objects are not enough available in the traditional camera calibration methods. In the condition of insufficient constraints, two parallel lines and three collinear points on road lanes are used as calibration reference objects. The constraint conditions consist of the spacing of two parallel lines, two lengths between three collinear points and their corresponding coordinates in an image. The formulas are derived to solve camera parameters. The relationship between pixels in traffic image and points in the world coordinate system is built. The effect caused by ignoring swing angle is re.assessed. Experi- mental results show that approximate solutions of camera parameters are satisfying despite the lack of sufficient constraints. The measurement precision in the direction of the road is superior to Fung＇s camera calibration method. The advantages of the pro- posed method are the easy selection of the reference object, good flexibility and fewer constraints. It is very suitable for expeditious calibration of the camera in the transportation surveillance system.