用基于种子点的三维图像相关法测量连续大变形

摘要

针对使用图像相关法测量物体发生的较大刚体旋转或变形容易失效的问题,提出了一种连续大变形的图像匹配方法.该方法首先利用相邻状态变形的连续性对种子点进行匹配,然后利用同一状态相邻点的变形连续性,通过种子点为其他点提供初值,并最终完成对所有点的匹配,从而在无需人工交互的情况下实现对大变形过程的高效测量.刚体旋转实验以及单向拉伸实验表明,对于40°的刚体旋转位移场以及高达113％的变形,测量结果良好,证明了本文方法的有效性.%To solve the problem of a large deformation measurement by using the Digital Image Correlation(DIC)method,a DIC method based on seed points is presented.After dividing subsets in the reference image,one point is chosen as the seed point and is matched firstly.According to the deformation continuity of adjacent stages of the image,an improved integer pixel search method is proposed to guarantee that the seed point can be matched successfully even in a large deformation situation.Once the seed point is matched,it can be used to calculate the initial correlation parameters of its four neighbor points according to the deformation continuity of neighbor points in one stage.The above progress repeats until all the points are matched.A rigid rotation experiment and a tensile test are carried out to verify the proposed method.Measurement results for 40° rotation and large deformation up to 113% confirm the validity of the method.