A real-time precise vision measurement method for thin material tensile tests at full strain-rates

Abstract

Vision measurement has extensive potential applications in the thin material stress-strain tests. The traditional single vision measurement methods of sub-pixel location based on feature extraction (FE) or digital image correlation (DIC) are defective in their full strain-rate ranges. FE exhibits a satisfactory stability at dynamic strain-rates, but yields a low accuracy because of the imprecise artificial mark, especially for materials with a rough surface texture. DIC, on the other hand, requires time-consuming computation, and exhibits an unstable accuracy at a high strain-rate because an increasing deformation can decrease the correlation of pixel subsets. To address these issues, a real-time precise measurement method combining DIC and FE is presented for thin material tensile tests, which can satisfy the test requirements region-by-region in the full strain-rate ranges. To reduce the effect of imaging distortion, a projection model based on the local sub-plane mapping was designed. Real experiments and analyses were performed to evaluate the performance the proposed method. The results showed that this method yields an improved performance at dynamic strain-rates.