Abstract: During similar simulation experiments in geotechnical engineering, cracks and speckles falling off appear on the surface of the equivalent material model, and invalid areas will form at the excavation zones and model surface penetrations. However, the traditional digital image correlation (DIC) method had some limitations in these cases. To improve the applicability of the DIC method in similarity simulation experiments, a novel method was proposed based on the image feature matching algorithm. In the proposed method, firstly, reliable subsets were selected by feature matching of images before and after deformation; secondly, deformation parameters of reliable subsets were obtained by correlation operation; finally, deformation parameters of monitored points were determined by the least absolute deviation fitting of displacements of reliable subset center points in certain areas around the monitored points. To validate the proposed method, deformations of the similar simulation experiment of mining-induced fault slip were measured using the proposed method, and the results were compared with those of the traditional method. Results show that cracks and invalid areas can be accurately identified by the proposed method, and displacements measured by the proposed method are more complete than those by the traditional DIC method in the vicinity of these areas. The proposed method has strong stability for speckles falling off on the surface of the model.

Key words: geotechnical engineering, similar simulation experiment, DIC method, feature matching, least absolute deviation fitting, fault