Stereo-rectification and homography-transform-based stereo matching methods for stereo digital image correlation

Abstract

The measurement accuracy and efficiency of stereo digital image correlation (DIC) greatly depend on the stereo matching which is to find the corresponding points (CPs) between the left and right images captured from different perspectives. Hence, it is significant to enhance the accuracy and efficiency of stereo matching in stereo DIC. In this study, a stereo-rectification-based stereo matching (SRSM) method and a homography-transform-based stereo matching (HTSM) method are presented to search the CPs between the left and right images. Although HTSM is only suitable to the planar object, it still has many applications. The presented methods simplify the coarse searching process of stereo matching and alleviate the influence of the shape change caused by the difference of perspectives (DOP). The performances of SRSM, HTSM, and the conventional stereo matching (CSM) method are compared in both simulation and experiment. It is shown that the accuracy and efficiency of HTSM are the highest among the three methods if the object has a planar surface, and the efficiency of SRSM is higher than that of CSM.