为解决数字图像相关（DIC）方法测量帧间大变形时出现的退相关现象，提出一种基于双边特征匹配的DIC全场测量方法。首先，采用ORB（Oriented FAST and Rotated BRIEF）算法提取和匹配特征点对，保证快速提取到数目较多的匹配点对；其次，针对存在的错误匹配问题，在网格运动统计方法基础上提出双边特征匹配，剔除错误匹配点对；最后，独立估计每个感兴趣点的变形参数，为反向高斯-牛顿算法提供初值，并不断迭代获得最终变形结果。通过仿真和聚合物泡沫大压缩实验验证了该方法的鲁棒性和准确性。结果表明，该方法适用于测量帧间大压缩变形，可测实际变形量在40%左右，并且在测量其它帧间大变形方面具有广阔的应用前景。
To solve the decorrelation phenomenon of digital image correlation when measures large inter-frame deformation, a DIC full-field measurement method based on bilateral feature matching is proposed. Firstly, the ORB (Oriented FAST and Rotated BRIEF) algorithm is used to extract and match the feature point pairs to ensure that a large number of matching point pairs can be quickly extracted. Secondly, to deal with the problem of false matching, a bilateral feature matching is proposed based on the grid-based motion statistics method to eliminate false matching point pairs. Finally, the deformation parameters of each point of interest are independently estimated and provided to the inverse compositional Gauss-Newton algorithm to iteratively obtain the final deformation results. The robustness and accuracy of the proposed method are verified by simulation and large compression tests on the polymer foam. Results demonstrate that the real deformation can be measured by this method is about 40%., which is suitable for measuring large inter-frame compression deformation, and shows great potential in measuring other large inter-frame deformation.