基于像素-角度映射的摄像机标定算法

摘要

According to the fact that the linear imaging model cannot explain the lens distortion,and to overcome the problem that the ordinary calibration methods are not suit for the change of height and angle,based on the pin-hole model and the previous distortion correction,this paper builds up a mirror coordinates system to study the reflection relation of pixel and angle.By using trigonometric functions and distortion coefficient,the height parameter and angle parameter are introduced into the model,and then the non-linear model is obtained.Finally,under different kinds of situation,comparing with Tsai’s method,the algorithm validity is verified.After changing the camera angle and height,the algorithm,which does not require re-calibration,is still with good accuracy and robustness.Thus,the algorithm can be applied to the engineering field.%为克服线性模型不能解释镜头畸变的缺点,以及常规标定中图像成像转换以及标定不能适应高度和角度改变等问题,在小孔线性成像模型的基础上,通过前期畸变标定和建立镜像坐标系,研究像素与角度的映射关系.利用三角函数将畸变系数和高度及角度参数引入到标定模型中,得到基于像素-角度映射的非线性模型.在多种标定环境下与Tsai等算法进行实验对比,验证了算法的准确性.在改变摄像机的角度和高度后,进一步验证了基于像素-角度映射的摄像机标定算法的系数不需要重新标定,仍具有良好的准确性和鲁棒性,可推广到工程应用领域.