一种数字图像相关非均匀位移精度优化方法

A Simple Method for Improving the Nonuniform Displacement Accuracy of Digital Image Correlation

对一阶形函数进行分析,发现其不匹配所造成误差与同样窗口大小的均值滤波有着紧密的联系。探索一次和二次均值滤波对数据的平滑效应,发现多次滤波所造成的数据差之间存在一定的可类比性,因此将两次均值滤波形成的数据差作为一次均值滤波数据差的估计,以此为基础对原数据进行恢复,用来抑制由于形函数不匹配所造成的误差。分别将本方法应用到简单正弦函数和混合函数,以及另外模拟了数字散斑图像,并进行了数字图像相关运算,以本方法对位移数据进行恢复,取得了良好的效果。实验结果表明,本方法能够有效抑制由于形函数不匹配所造成的位移测量误差,有助于进一步提高数字图像相关技术对于非均匀变形的测量精度。

One of the main error source of the nonuniform measurement in digital image correlation(DIC)is the error from the undermatched shape function.The first order shape function is analyzed,and it is discovered that this error is closely connected with the mean filter with same window size.Exploring the mean effect of the mean filter,it is found that the effect between one and two times mean filter could be compared.So the data difference between two times mean filter is used as of the estimation of the first time difference.Based on that the original data could be recovered,and the error from the undermatched shape function is suppressed.This method is used in the sine and mixed simple functions,and the effectiveness is demonstrated.Also the simulated speckle patterns are generated,digital image correlation is employed to calculate the displacement.The proposed method is applied to the displacement data,and the experimental results shows that the systematic error due to undermatched shape functions is reduced,and the displacement measurement result of nonuniform deformation by digital image correlation is improved.