Image feature guided self-adaptive subset configuration for digital image correlation

Abstract

The size and shape of subset are critical factors influencing the accuracy and robustness of digital image correlation (DIC) measurements. In current DIC implementations, a subset with fixed size and shape is manually configured, highly dependent on user expertise. Moreover, this configuration often leads to suboptimal performance of DIC because it cannot fit the spatially varying speckle pattern quality or inhomogeneous deformation field. To overcome these issues, a self-adaptive subset configuration strategy is proposed that can dynamically optimize the subset at each point of interest (POI) to achieve optimal accuracy and resolution of measurements. In contrast to existing methods, our approach utilizes the information of image features to take both the localized speckle pattern quality for registration and the smoothness of deformation field into account. Experimental studies demonstrate that the method adapts effectively to various speckle patterns. It optimizes subset size and shape without manual intervention in measuring inhomogeneous deformation field and stereo reconstruction of complex surface topography. By enhancing the measurement accuracy and robustness while minimizing user dependence, this strategy is expected to facilitate the applications of DIC in diverse and challenging scenarios.