Abstract: According to the different degree distortion of CCD lens used in machine visual detection and high precision measurement, this paper proposes a combined model-based for CCD lens distortion correction. The algorithm uses the classical model to correct the distortion image initially, the polyhedral function is employed for second high precision correction, and cubic B spline function is used for grade rebuilding. Experimental results indicate that the algorithm can promote the precision and strengthen the robustness compared with other single lens distortion correction model, and the radial correction RMS is 0.3 pixel.

Key words: lens distortion, correction algorithm, combined model

摘要： 针对机器视觉检测和高精度图像测量中使用的CCD镜头都存在不同程度光学畸变的问题，提出基于混合模型的CCD镜头畸变校正算法。用经典模型对畸变图像进行初次校正，用多面函数拟合法进行二次精校正，用三次B样条函数进行灰度重建。实验结果表明，该方法在不依赖摄像机内部参数条件下，相比单一的镜头畸变校正模型，精度提高，鲁棒性增强，校正后径向均方根误差为0.3个像素。

关键词: 镜头畸变, 校正算法, 混合模型

CLC Number: 

• TP312