自由双目立体视觉摄像机动态外参数的获取
发布时间：2019-01-07 06:58
【摘要】：针对自由双目立体视觉中由于摄像机旋转导致的摄像机外参数变化的问题,提出一种基于旋转轴标定的动态外参数获取方法。在多个不同位置,立体标定得到多组旋转平移矩阵,利用最小二乘法求解旋转轴参数;结合初始位置左右摄像机的内、外参数及旋转角度,实时获取左右摄像机的外参数。利用所提方法获取动态外参数,并对棋盘角点进行三维重建,平均误差为0.241 mm,标准差为0.156 mm;与基于多平面标靶的标定方法相比,精度高且操作简单。所提方法无需实时标定,可完成摄像机旋转情况下动态外参数的获取。
[Abstract]:A dynamic external parameter acquisition method based on rotating axis is proposed to solve the problem of camera external parameter change caused by camera rotation in free binocular stereo vision. In many different positions, the rotation translation matrix is obtained by stereo calibration, and the rotation axis parameters are solved by the least square method, and the external parameters of the left and right camera are obtained in real time by combining the internal and external parameters and the rotation angle of the camera at the initial position. The proposed method is used to obtain the dynamic external parameters and reconstruct the corner points of the chessboard with an average error of 0.241 mm, standard deviation of 0.156 mm;. Compared with the calibration method based on multi-plane target, the proposed method has high accuracy and simple operation. The proposed method does not need real time calibration and can obtain dynamic external parameters under camera rotation.
【作者单位】： 天津大学精密仪器与光电子工程学院;光电信息技术教育部重点实验室;中国民航大学电子信息与自动化学院;
【基金】：国家自然科学基金资助项目(61535008)~~
【分类号】：TN948.41
，
本文编号：2403330
[Abstract]:A dynamic external parameter acquisition method based on rotating axis is proposed to solve the problem of camera external parameter change caused by camera rotation in free binocular stereo vision. In many different positions, the rotation translation matrix is obtained by stereo calibration, and the rotation axis parameters are solved by the least square method, and the external parameters of the left and right camera are obtained in real time by combining the internal and external parameters and the rotation angle of the camera at the initial position. The proposed method is used to obtain the dynamic external parameters and reconstruct the corner points of the chessboard with an average error of 0.241 mm, standard deviation of 0.156 mm;. Compared with the calibration method based on multi-plane target, the proposed method has high accuracy and simple operation. The proposed method does not need real time calibration and can obtain dynamic external parameters under camera rotation.
【作者单位】： 天津大学精密仪器与光电子工程学院;光电信息技术教育部重点实验室;中国民航大学电子信息与自动化学院;
【基金】：国家自然科学基金资助项目(61535008)~~
【分类号】：TN948.41
，
本文编号：2403330
本文链接：https://www.wllwen.com/kejilunwen/xinxigongchenglunwen/2403330.html
