Abstract: There is often some extent distortion existing in the short focal length and wide angle lens camera which is used for machine vision detection，in order to carry out high precision measurement，the lens distortion must be corrected first.A new method for distortion correction is proposed.It obtains the initial distortion coefficients by combining the optical image theory with distortion correction model at first，then the optimization function is used to find the optimal distortion coefficients，cubic B-spline interpolation method is adapted for gray rebuilding at last.Experiments indicate that this method shows perfect results，the radial correction RMS is 0.45 pixel and the gray rebuilding RMS is 0.36 pixel.

Key words: image processing, geometric distortion, correction algorithm, gray rebuilding

摘要： 在机器视觉检测中大视场短焦距摄像机镜头一般都存在一定程度的光学畸变，在高精度测量中必须对摄像机镜头畸变进行校正。提出利用光学成像规律和镜头畸变校正模型相结合的畸变校正方法求出初始畸变系数，然后通过优化目标函数求出最优畸变系数，最后采用三次B样条插值对畸变图像进行灰度重建。实验结果表明该方法在不依赖摄像机内部参数的前提下，校正后径向均方根误差为0.45个像素，灰度重建后径向均方根误差为0.36个像素。

关键词: 图像处理, 几何畸变, 校正算法, 灰度重建