大变形下基于数字图像相关的改进分段位移传递法  

Improved Fractionized Displacement Transfer Algorithm Based on Digital Image Correlation in Large Deformation Applications

作　　者：叶南[1] 张丽艳[1] 

机构地区：[1]南京航空航天大学江苏省精密与微细制造技术重点实验室,江苏南京210016

出　　处：《光学学报》2010年第4期976-983,共8页Acta Optica Sinica

基　　金：国家自然科学基金(50875130);博士点基金(200802870016);江苏省支撑计划(BE2008136)资助课题

摘　　要：将数字图像相关方法用于物体表面发生大变形时的图像匹配时,经常因变形区域非线性变形和变形过大而遇到问题。借鉴已有的分段测量思想,提出了一种改进的位移测量方法。该方法根据图像间的相关程度将图像序列中的某一阶段图像重新选取为参考图像,然后将先前匹配得到的该图像上的亚像素子区中心圆整成整像素整坐标,从而可以直利用数字图像灰度分布作为该参考子区的灰度分布进行数字图像相关计算。对像素圆整过程中损失的精度,利用连续介质的变形连续性概念进行精度补偿。另外提出用生长法获取初始匹配点,提高了匹配点对生成效率,并且根据由生长法获取的待匹配点集与由相关计算得到的最优匹配点集之间的比率实现参考图像的自动选取。实验对比和精度分析显示所提方法有效、可靠,匹配精度在0.01 pixel量级。Difficulties often arise when digital image correlation（DIC） method is used for serial image matching in nonlinear and large deformation applications.An improved algorithm is proposed by utilizing the concept of fractionized displacement transfer（FDT）.For a series of image that records the deformation process,the matching algorithm automatically rechoose an intermediate image as the reference image when the correlation level between the current reference and target images drops below a given threshold.Then the sub-pixel corresponding points in the new reference image are rounded into intergral pixels to facilitate the subsequent DIC matching procedure.A method,which is specifically designed based on the continuum assumption,is proposed for compensating the loss of accuracy caused by the pixel rounding.In addition,a growing method for obtaining initial matching points is proposed,which proved to be able to improve the matching effect and efficiency.The reference images are chosen in terms of the ratio of the number of points in the initial matching set to the number of matched points obtained by the correlation method.Experiment and accuracy analysis demonstrate that the proposed technique is effective and reliable,the matching accuracy is on the order of 0.01 pixel.

关 键 词：机器视觉 分段位移传递 数字图像相关 亚像素匹配 大变形 板料成形 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]