第19卷第1期 2011年1月
文章编号
1004-924X(2011)01-0161-07
光学精密工程
Optics and Precision Engineering
平面几何测量中的图像畸变校正
苏成志，王恩国，郝江涛，曹国华，徐洪吉（长春理工大学机电工程学院，吉林长春130022）
Vol. 19No.1
Jan.2011
摘要：针对图像畸变对平面图像儿何线度精密测量精度的影响，提出一种直接利用标准网格板作为测量基准的畸变校正方法。根据待测物体与网格板处于相同物面时，其图像畸变与网格板图像畸变相同，待测点在网格板图像中相对网格的几何位置不变这一性质，提出直接使用发生畸变的网格板图像作为校正基准来代替通过建模将外部标准转换为摄像机内部基准的畸变校正恩路。首先，确定待测点在网格板畸变图像中的初始位置：然后，根据平行线分线段成比例定理确定待测点在网格板畸变图像网格内的精确位置，对两者求和完成待测物体上任意两点的实际几何线度测量。实验证明，当校正网格板间距为1mm，精度为0.2um时，使用提出方法得到的畸变误差是现有校正方法的20%，校正精度可达4
m。该方法省去了建模过程，其校正精度仅与网格板精度有关，具有更高的精度和适应性。关键词：平面图像测量；图像时变；校正基准；建模校正
中图分类号：TP391.4
文献标识码：A
doi:10.3788/OPE,20111901.0161
Distortioncorrectionforimagesinplanarmetrology SU Cheng-zhi,WANGEn-guo,HAO Jiang-tao,CAOGuo-hua,XUHong-ji(Collegeof Mechanical andElectricEngineering，ChangchunUniversityof
Science and Technology,Changchun 130022,China)
Abstract: In consideration of the effect of the distortion error of an image on the accuracy of planar ge-ometrical measurement in the precise visual metrology,a method to correct the image distortion by u-sing standard grid board directly as measurement calibration is proposed. As the position of an under tested point is unchanged relative to that of grid board when an under-tested object and a grid board lay on the same object plane, the grid board is directly used as the correcting calibration of image dis-tortion instead of the way that converts the external standard into the intrinsic parameter of a camera by modeling . Firstly,the primary position of the under-tested point in the image of grid board is deter mined;then,its fine distance is decided according to the proportion theorem of line segment divided by parallel line, Finally, planar geometrical measurement is fulfilled by calculating the sum of both dis-tances,Experimental results show that the distortion error by the proposed method has reduced to 2o% that of the traditional method and the correctoin accuracy of image has reached 4 μm or higher, while the distance of intersection point of the calibrated grid board is 1 mm and its accuracy is O. 2 μm. The method is more applicable and has high accuracy,for it omits the modeling and its accuracy only
收稿日期：2009-11-26；修订日期：2010-03-29
基金项目：吉林省科技发展计划基金资助项目（No.20070304)