一种基于Delaunay剖分三角形内插值的摄像机标定方法  

A Camera Calibration Method Based on the Interpolation of Delaunay Triangles

作　　者：徐侃[1] 李文龙[1] 尹周平[1] 

机构地区：[1]华中科技大学数字制造装备与技术国家重点实验室,湖北武汉430074

出　　处：《装备制造技术》2013年第2期13-17,29,共6页Equipment Manufacturing Technology

基　　金：青年科学基金项目(51105155);国家科技重大专项课题(2009ZX02021-004);中央高校基本科研业务费资助(HUST2012QN251)

摘　　要：基于Delaunay三角剖分性质,应用特征点集Delaunay剖分三角形内插值运算,提出了一种CIDT摄像机标定方法。提取特征点像素坐标,按像素坐标对特征点集进行Delaunay三角剖分,并依据定义的条件筛选计算有效Delaunay剖分三角形;通过仿真实验讨论了有效剖分三角形区域的内插值线性关系不变性,并基于该性质,提出了完成变形前像素点在变形后图像上对应位置映射运算的新方法,从而代替了传统方法中的二元高次多项式系数拟合计算。经过与MIL软件包进行对比测试,证明了CIDT方法能够达有效地去除图像中的非线性畸变,且可满足电子封装设备中摄像机标定与坐标转换精度。Based on the law of Delaunay triangulation, this paper proposed a method called CIDT to finish camera calibration by using Interpolation of Delaunay Triangles. Pixel coordinates of the feature points was calculated, and Delaunay triangulation was applied on these feature points by their pixel coordinates. A restriction defined in this paper was used to remove the redundant Delaunay triangles and remain the effective Delaunay triangles. The invariance property of the effective Delaunay triangles＇ interpolation between undistorted image and distorted image was demonstrated by simulation experiment. Based on the this property, this paper proposed a new method to connect the pixel points of undistorted image to their position coordihates in distorted image. This new method used the interpolation instead of the fitting of polynomial. CIDT was proved to have a good result in image correction and coordinate transform, by the contrast experiment with Matrox Imaging Library.

关 键 词：DELAUNAY三角剖分 三角形内插值 摄像机标定 电子封装设备 

分 类 号：TN948.41[电子电信—信号与信息处理]