A computer vision-based video extensometer technique for geosynthetics tensile testing

This study presents a low-cost strain measurement technique that combines Python-based computer vision tools with a compact wide-angle action camera. The proposed method serves as an alternative to crosshead displacement, extensometer-based measurements, and conventional digital image correlation (DIC). Its accuracy is evaluated using in-isolation tensile tests conducted on two types of high-density polyethylene (HDPE) geomembranes and three types of nonwoven polypropylene (PP) geotextiles. Quantitative comparisons with existing methods are made using error metrics, highlighting the differences between the proposed technique and traditional approaches. The effects of testing speed and material characteristics on measurement performance are also examined. Consequently, the results demonstrate that the proposed method provides repeatable and accurate displacement measurements within a predefined gauge length using a series of image processing steps. In addition, complementary DIC analyses are used to identify regions of strain localization in the tested specimens.