Efficient Monte Carlo simulation of spatiotemporal speckles and their correlations

Chen Bar, Ioannis Gkioulekas, Anat Levin. Optica 2023.


When viewed under coherent illumination, scattering materials such as tissue exhibit highly varying speckle patterns. Despite their noise-like appearance, the  temporal and spatial variations of these speckles, resulting from internal tissue dynamics and/or external perturbation of the illumination, carry strong  statistical information that is highly valuable for tissue analysis. The full practical applicability of these statistics is still hindered by the difficulty of simulating the speckles and their statistics. This paper proposes an efficient Monte Carlo framework that can efficiently sample physically correct speckles and estimate  their covariances. While Monte Carlo algorithms were originally derived for incoherent illumination, our approach simulates complex-valued speckle fields. We compare the statistics of our speckle fields against those produced by an exact numerical wave solver and show a precise agreement, while our simulator is a few orders of magnitude faster and scales to much larger scenes. We also show that the simulator predictions accurately align with existing analytical models and simulation strategies, which currently address various partial settings of the general problem.