摘要
利用多圆阵列平面标定板,以其阵列圆心作为世界坐标参考点,采用基于轮廓跟踪的区域标记算法提取图像上多圆区域的形心作为图像像素坐标系参考点,利用同一标定板不同位置拍摄的多幅图像,借助每幅图像的单应性矩阵线性标定出摄像机的内外参数并作为初值,建立非线性优化模型采用最小二乘法进一步优化参数与畸变系数。利用优化的参数对空间点进行投影得到投影平均误差小于0.05个像素。
In order to get precise camera parameters, one planar template including multi-circles whose centers were regarded as the reference points was designed. Using region labeling arithmetic based on the contour tracking the corresponding shape centers in images were automatically extracted. Every 3D points in the template and their image points produced one Homography matrix, using the multi-images＇ Homography matrixes the camera intrinsic parameters and external parameters were estimated by solving the closed-form. Finally the estimates were fed to the optimizer as initial guess parameters to get the precise calibration parameters through least square method. The tests indicated that using the final calibration parameters the projection error was not more than 0.05 pixels.
出处
《机电工程技术》
2015年第12期99-101,共3页
Mechanical & Electrical Engineering Technology
关键词
摄像机标定
单应性矩阵
线性标定
最小二乘法
camera calibration
Homography matrix
linear calibration
least square method
