基于定点旋转的遥操作前端视觉自标定

为解决遥操作过程中当前端视觉内参数改变时的标定问题,特别是在一些危险的或是人无法到达的环境下,在前端放置标定物用传统的方法进行标定是不可能的,因此采用了一种基于定点旋转的自标定技术来进行标定.对基于定点旋转的自标定方法进行了理论推导,得到了标定步骤.在试验过程中,摄像机固定在一种旋转装置上进行拍摄,采用SIFT(Scale Invariant Feature Transformation)算法对拍摄的图像进行特征点检测和匹配,采用牛顿算法进行计算.为和传统的标定方法进行比较,对自制的平面模板进行拍摄,利用Harris算子对平面模板图像进行角点检测,用传统的标定方法重新对摄像机进行标定.结果表明两种方法得到的答案相近,用传统标定方法标定结果是接近摄像机真实参数值的,说明这种新的标定方法是一种有效的标定方法.

Camera self-calibration technique for vision system in front for teleoperation system based on pointing rotation

Camera self-calibration technique for vision system in front for teleoperation system based on pointing rotation is an effective method.The problem that the calibration object can not be put in front of the camera,when the teleoperation is going on and the camera parameters need to be modified can be solved by using the self-calibration.The theoretic of self-calibration was analyzed firstly,correlative examination was done secondly.In the testing process,the camera was fixed on the rotation device.The scale invariant feature transformation(SIFT) was used to solve the problem of the feature points detection and matching,and the Newton method was used to calculate.Another camera calibration method was used to calibrate the same camera,and compared with the self-calibration.The corner point of the calibration plane was detected by using the Harris corner detector.The result of the two methods is similar,so the self-calibration is effectual.