In this work, the efficiency of a feature-based DIC （digital image correlation） algorithm for measuring high strain gradients was investigated by means of numerical and actual experiments. The so-called SIFT-Meshles...In this work, the efficiency of a feature-based DIC （digital image correlation） algorithm for measuring high strain gradients was investigated by means of numerical and actual experiments. The so-called SIFT-Meshless method consisted of a novel formulation involving the SIFT （scale-invariant feature transform） feature detector with a self-adaptive meshless formulation. Whereas the numerical experiments aimed to evaluate the accuracy and the spatial resolution, the actual experiments aimed to demonstrate in practice the above findings. A stereoscopic system and a micro-stereoscopic system were used to perform high strain gradient measurements in notched specimens of different materials and notch sizes. This paper concludes that the feature-based algorithm is able to provide accurate strain measurements at high strain gradient regions, even under conditions of plasticity. Moreover, the algorithm showed its efficiency to capture the peak strain near the notch boundary. Lastly, a spatial resolution study proposes a link between the desired accuracy and the pixel resolution required to perform accurate measurements of high strain gradients.展开更多
文摘In this work, the efficiency of a feature-based DIC （digital image correlation） algorithm for measuring high strain gradients was investigated by means of numerical and actual experiments. The so-called SIFT-Meshless method consisted of a novel formulation involving the SIFT （scale-invariant feature transform） feature detector with a self-adaptive meshless formulation. Whereas the numerical experiments aimed to evaluate the accuracy and the spatial resolution, the actual experiments aimed to demonstrate in practice the above findings. A stereoscopic system and a micro-stereoscopic system were used to perform high strain gradient measurements in notched specimens of different materials and notch sizes. This paper concludes that the feature-based algorithm is able to provide accurate strain measurements at high strain gradient regions, even under conditions of plasticity. Moreover, the algorithm showed its efficiency to capture the peak strain near the notch boundary. Lastly, a spatial resolution study proposes a link between the desired accuracy and the pixel resolution required to perform accurate measurements of high strain gradients.
