In digital holographic interferometry, reliable estimation of phase derivatives from the complex interference field signal is an important challenge since these are directly related to the displacement derivatives of a deformed object. In this paper, we propose an approach based on deep learning for direct estimation of phase derivatives in digital holographic interferometry. Using a Y-Net model, our proposed approach allows for simultaneous estimation of phase derivatives along the vertical and horizontal dimensions. The robustness of the proposed approach for phase derivative extraction under both additive white Gaussian noise and speckle noise is shown via numerical simulations. Subsequently, we demonstrate the practical utility of the method for deformation metrology using experimental data obtained from digital holographic interferometry.

No abstract

No abstract

The paper presents a method for automated defect identification from fringe patterns. The method relies on computing the fringe signal’s Wigner–Ville distribution followed by a supervised machine learning algorithm. Our machine learning approach enables robust detection of fringe pattern defects of varied shapes and alleviates the limitations associated with thresholding-based techniques that require careful control of the threshold parameter. The potential of the proposed method is demonstrated via numerical simulations to identify different types of defect patterns at various noise levels. In addition, the practical applicability of the method is validated by experimental results.

No abstract