Camera calibration is an important part of computer vision, and the accuracy and real-time performance of calibration directly affect its industrial applications. In order to realize fast and high-precision camera calibration,the im-aging model and the solving method of model parameters were studied. During the studied,it is found that the distor-tion of area near the center of the camera plane can be neglected. In view of this discovery,a simple and fast camera stereo calibration method is proposed. Firstly, the points near the center of the imaging plane were used to be feature point to solve the internal and external parameters of the camera. Then,with these parameters known,the 3D coordi-nates and pixel coordinates of the remaining points were used as input to fit the distortion parameters through multiple linear regression minimum two multiplication optimization. The experimental results show that the calibration accuracy of this method is high, the average pixel error can reach 10-2 orders, and the computing time of the algorithm is short-ened obviously,which can better meet the needs of real industrial applications.%相机定标是计算机视觉中极为重要的环节,定标的准确性和实时性直接影响其工业应用.为了实现快速、高精度的相机定标,对相机的成像模型及模型参数求解进行了研究,发现相机成像平面中心附近区域的畸变基本可以忽略.针对这一发现,提出一种简易、快速的相机立体定标方法.首先以成像平面中心附近的点作为特征点,用来求解相机的内外参数,再将求得的内外参数作为已知带入畸变模型,以剩余点的三维坐标及像素坐标作为输入,用多元线性回归最小二乘法对畸变参数进行最优化拟合.实验表明,该方法的定标精度高,其平均像素误差可达到10-2量级,同时算法运算时间明显缩短,能更好地满足实际工业应用的需求.
展开▼
