摘要
图像在应用时往往需要物体在图像中准确的位置,但由于物体在从世界坐标系到摄像机坐标系变换时存在平移和旋转,所以拍摄的图像会存在非线性几何畸变。为此针对广角镜头非线性畸变,在分析多项式变换的基础上,提出多项式分区域校正方法。首先检测标准网格和畸变网格的坐标,计算出镜头的畸变系数;然后拟合多项式进行坐标修正,重建图像;最后对重建的图像进行分块,计算每个小块的畸变系数并再次重建图像。实验结果表明,提高了图像校正的精度和智能化,从而达到高质量校正的目的。
We usually want to know the accurate position of the object in the image,but a rotation and a translation exists in the transformation of the object from a world coordinate system(WCS) to a camera coordinate system(CCS),which result in nonlinear distortion between real images and ideal images to some extent.For nonlinear distortion of wide-angle lens,an approximation algorithm is proposed in this paper.We firstly detect corners to get coordinates of standard grid model and distortion,and calculate distortion parameters of the camera lens.Then we use a second-order polynomial to rectify coordinates.Finally,we divide the image into rectangles according to corners and calculate distortion parameters of each rectangle.This algorithm can maximize the calibration precision and intelligentize,moreover,accomplish high quality calibration.
出处
《计算机仿真》
CSCD
北大核心
2011年第7期268-271,共4页
Computer Simulation
关键词
角点检测
双线性插值
几何畸变
广角镜头
Corner detection
Bilinear interpolation
Geometrical distortion
Wide-angle lens