成像系统中的图像畸变校正技术  

Image Distortion Correction Technology in Imaging Systems

作　　者：陈盈锋[1] 李大海[1] 章辰[1] 王琼华[1,2] 刘曦[1] 李萌阳[1] 李洪[1] 

机构地区：[1]四川大学电子信息学院,四川成都610065 [2]视觉合成图形图像技术国家重点学科实验室,四川成都610064

出　　处：《光学学报》2012年第F12期119-126,共8页Acta Optica Sinica

基　　金：国家自然科学基金（60877004）资助课题.

摘　　要：针对成像系统中图像畸变对测量精度的影响，提出了基于Zernike矢量多项式的图像畸变校正方法。将基准图与畸变图中标记点的坐标用相同的归一化方法在单位圆内进行归一化，运用Zernike矢量多项式拟合基准图与畸变图的映射关系，利用得到的映射关系对畸变图进行校正并插值，得到校正图。然后对校正效果进行了评价，并与基于Matlab的相机标定工具箱方法进行了比较。实验结果表明：对径向畸变，Zernike矢量多项式方法与后者相比，均方根（RMS）误差减少近50％；对梯形畸变，Zernike矢量多项式校正方法可使RMS误差减少至10^-3量级。In consideration of the effect of the image distortion on the measurement accuracy in imaging system, a method of image distortion correction based on Zernike vector polynomials is proposed. The coordinates of the fiducial points both in reference map and distorted map are normalized in a unit circle, and the mapping relation between reference map and distorted map is fitted with the Zernike vector polynomials. The correction of the distorted image is implemented by using the mapping relation which is discovered by use of the Zernike vector polynomials fitting method. In order to reduce the data discontinuity, the corrected map is also interpolated. Then the correction results are evaluated and compared with the camera calibration toolbox for Matlab method. The experimental results indicate that compared with the latter, using the Zernike vector polynomials method can reduce the root mean square （RMS） error due to the radial distortion by nearly 50 %, and can reduce the RMS error due to the keystone distortion to 10^-3 magnitude order.

关 键 词：成像系统 畸变校正 相机标定工具箱 Zernike矢量多项式 径向畸变 梯形畸变 

分 类 号：O435.2[机械工程—光学工程]