边缘直线拟合确定鱼眼镜头光心算法

An algorithm to determine the fish-eye lens optical center using edge straight-line-fitting

为了去除鱼眼图像存在严重的畸变,需要对鱼眼相机进行标定,本文在已知相机参数表的前提下,提出了一种边缘直线拟合来标定最重要的镜头光心坐标算法.该算法利用像素点在图像平面上的物理尺寸及相机焦距,确定相机标定内参矩阵中水平轴和竖直轴的尺度因子.在部分内参已知的情况下,对光心的横、纵坐标在给定范围内进行遍历,校正点到拟合直线加权距离和最小的点,即为所求的光心坐标值.实验结果表明:与常规的标定算法相比,该算法得到的光心坐标减少了由于不精确参数使用带来的累积误差,提高了相机标定结果的精度.

In order to remove the distortion in fish-eye image,the camera calibration is required.We put forward an algorithm to determine fish-eye lens optical center using the edge straightline-fitting technology with a given camera parameters table.This algorithm utilizes the physical size of each pixel in the image plane as well as the camera focal length.Therefore,the scale factor in horizontal and vertical axes of the reference matrix can be determined.With some provided parameters,the optic center can be determined by traversing the pixel coordinates within given range.The optic center,which made the weighted distance between corrected points and fitted straight lines smallest,is needed.Experimental results show that compared with the conventional camera calibration algorithm,this algorithm reduces the accumulation of errors due to the using of inaccurate parameters and makes the calibration results more accurate.