简易高精度的平面五点摄像机标定方法

发布时间：2018-04-11 08:34

本文选题：相机标定 + 矩形　；参考：《光学精密工程》2017年03期

【摘要】：为了进一步提高相机参数精度,简化标定过程,在张正友平面标定算法的基础上,提出了一种较简单的5点标定算法。本文利用一个矩形作为标定模板,将其4个角点和中心点作为待匹配点。仅利用不同角度拍摄的10(10~20幅均可)幅图像,即可求解全部相机内外参数。首先,为了减小数据对结果的影响,对获取的5个点的图像坐标进行归一化,将点坐标控制在以原点为中心,2~(1/2)为半径的圆内;其次,相机成像模型未考虑畸变,考虑到单应矩阵可能会有零元素,故按照9个自由度来求解;再次,求解出相机的内外参数,将求得结果的平均值作为非零因子λ的值;最后利用Levenberg Marquarat(LM)算法对所有内外参数进行整体优化。选用不同噪声水平的模拟实验和真实图像进行测试,结果显示该算法比张正友棋盘格算法的精度要高。
[Abstract]:In order to further improve the precision of camera parameters and simplify the calibration process, a simple five-point calibration algorithm is proposed on the basis of Zhang Zhengyou plane calibration algorithm.In this paper, a rectangle is used as the calibration template and its four corners and center points are used as the matching points.Only 10 or 10 or 20 images taken from different angles can be used to solve all the internal and external parameters of the camera.Firstly, in order to reduce the effect of the data on the result, the image coordinates of the five points obtained are normalized, and the coordinates are controlled in a circle with a radius of 2 / 2 at the center of the origin. Secondly, the camera imaging model does not take distortion into account.Considering that the homography matrix may have zero elements, it can be solved according to 9 degrees of freedom. Thirdly, the internal and external parameters of the camera are solved, and the average value of the result is regarded as the value of non-zero factor 位.Finally, the Levenberg Marquarat LM algorithm is used to optimize all the internal and external parameters.Simulation experiments with different noise levels and real image tests show that the accuracy of this algorithm is higher than that of Zhang Zhengyou checkerboard algorithm.
【作者单位】：北方工业大学图像处理与模式识别研究所;
【基金】：国家科技重大专项课题(No.2014ZX02502003) 国家自然科学基金资助项目(No.61170327)
【分类号】：TP391.41

【相似文献】