数字图像相关方法的大变形初值估计

Large-Deformation Initial Guess for Digital Image Correlation Method

针对一般变形初值估计方法在被测物体表面存在较大刚体转动或大变形的情况下失效的问题,提出了一种自动的变形初值估计方法,即通过SIFT算法匹配得到变形前后图像计算子区的对应点,并利用最小二乘法获得变形初值估计作为Newton-Rapshon方法的迭代初值。对大变形下的模拟散斑图进行了仿真计算,结果显示该方法在30°大转角的情况下仍能有效测量位移。

For the problem of the failure of the commonly used initial guess method under the condi-tion that the test object has a big rotation angle or large deformation, a new initial guess technique is proposed. This technique obtains the reliable initial deformation guess by matching points between the reference image and the deformed image with SIFT algorithm. Then, reliable initial deformation guess can be determined with least square method and used as the initial iteration Newton-Rapshon method. The measured displacement fields from speckle diagram with 30° large rigid body rotation obviously demonstrate the validity of the technique.