Evaluation of activity from binary patterns in dynamic speckle analysis

Abstract

Pointwise intensity-based statistical processing of dynamic laser speckle patterns formed on the surface of a diffusely reflecting object is perspective for monitoring of ingoing processes of physical or biological activity within the object. The output of the measurement is a two-dimensional map, which shows qualitatively regions of higher or lower activity on the object surface. In the paper, we have proposed transforming the captured 8-bit encoded speckle patterns into binary patterns by applying a pointwise threshold. Motivation behind this study is the need for an algorithm, which is robust to non-uniform illumination and offers lowered computational complexity as time and memory consumption. We showed by simulation and experiment that quality of the map obtained by processing of correlated binary patterns approached that achieved with 8-bit encoded patterns. We analyzed statistical behavior of the estimates built from binary patterns and proved applicability of a non-stochastic threshold as e.g. a constant intensity level at uniform illumination. The proposed algorithm was applied for processing polymer drop drying experimental data.