三维数字图像相关法的初值估计和仿真图研究

摘要

Three-dimensional digital image correlation (3D-DIC) method has been increasingly employed to measure 3D surface profile and displacement.The key technologies applied in this method include stereo camera calibration,digital image correlation,reconstruction,and 3D displacement/strain field computation.This paper mainly studies on an initial guess technique based on SIFF matching and linear interpolation to improve the accuracy of 3D-DIC.Additionally,a new method based on reverse mapping is proposed to generate the 3D simulated speckles for testing the accuracy of the initial guess and digital image correlation technique.The numerical simulations reveal that the error in Z-direction is about 2 × 10-4％ of the object distance.Test and analysis of the in-plane displacement error is within 8 μm under small displacement.Results show that the 3D simulated speckles are useful and effective for the verification of 3D-DIC method and show that the initial guess technique can achieve high accuracy.%三维数字图像相关法主要用来测量物体表面的三雏形貌和变形,其关键技术主要包括相机标定、图像匹配、三维重建和三维位移场和应变场的计算等.针对现有的初值估计方法在计算速度和精度上的局限性,提出基于sift的形函数和线性插值的初值估计方法,有效地提高了计算的速度和精度.传统的三维算法精度分析只能借助于高精度的仪器测量对比,为此提出了基于反向映射法的三维仿真散斑图分析方法.仿真结果表明:在小变形情况下z方向的误差在2×10-4％左右,面内位移测量误差在8μm以内.验证了三维仿真散斑图的有效性和三维初值估计策略的可行性.