本文介绍了一种新的摄像机标定算法,它的内部参数和外部参数可以通过从标定板的任何视点的图像来进行特征提取,然后将这些特征映射到无畸变和校正图像的对应点来估计。多视点校正是一个非线性最小二乘优化问题,它通过简化数学模型建立一个参考图像,初始化连同成本函数的梯度矢量的计算允许夸尔特算法以最快的速度收敛到最优解。 This paper presents a new camera calibration algorithm whose internal parameters and external parameters can be estimated by extracting features from the image of any view point on the calibration plate and then mapping these features to the corresponding points of the image without distortion and correction. Multi-view correction is a nonlinear least-squares optimization problem. It creates a reference image by simplifying the mathematical model. Initialization of the gradient vector with the cost function allows the Quartet algorithm to converge to the optimal solution as quickly as possible.