Evaluation of structural similarity quality measures for wrapped recovered phase maps in optical metrology

Abstract

We propose and evaluate structural similarity quality measures for wrapped recovered phase maps in the range of [0, 2π). The proposed measures adapt the mathematical description of a structural similarity index employed in intensity images by using circular statistics, which are suitable for analyzing the wrapped phase images obtained in the application of optical metrology techniques. The developed framework avoids 2D unwrapping pre-processing techniques and obtains a quality measure that properly evaluates visual distortions that occur in the proximities of phase discontinuities that do not correspond to significant phase errors. Numerical experiments based on the assessment of simulated wrapped images indicate the suitability of the proposed quality measure when a reference phase map is introduced.