机器视觉测量中相机两步标定法的改进

在机器视觉中为了确定空间三维模型和图像模型之间的几何坐标关系,必须要建立相机的图像几何模型及其参数,这个过程被称为相机的标定。作为最广泛使用的标定方法之一,两步标定法使用了网格板来进行相机的标定。在标定中,相机的多数参数都是通过求解分析方程得到的,然后经过反复迭代得到其他标定参数,使用这种标定方法需要在多个位置拍摄多个图像。本文针对两步标定法进行改进。在改进的方法中,用到了一个呈直角的L形标定板,标定时得到一副包含两个垂直标定图像的图片,之后使用两步标定法进行必要的相机内、外部参数的计算。标定的结果显示误差低于0.7%,能满足要求。

Improvement of Two-step Calibration Method in Machine Vision

In the machine vision we must establish the imaging geometric model of the camera and its parameters which are measurement parameters of the camera in order to determine the geometry coordinate of relationship between points in the spatial three-dimensional model and corresponding ones in the image.And this process is called as camera calibration which can be classified to two categories： the traditional method of camera calibration and camera self-calibration method.As one of the most common traditional method,two-step calibration method uses grid board to calibrate camera.In the calibration,the majority of calibration parameters will be obtained by solving analytical equations set,and then use the iterative solution to obtain other parameters,but there need multiple images at different locations at the time of calibration.The paper show an improvement of the traditional two-step calibration method.In this method we use two calibration images which are perpendicular with each other and obtain a single image which includes two perpendicular calibration images,and then use the two-step calibration method to establish the required internal and external parameters of the camera.The results of calibration show error is less than 0.7% and meet the requirements.