Dynamic laser speckle analysis via normal vector space statistics

Dynamic laser speckle analysis is a useful measurement technique which can be used to analyze micro-motion on a sample surface via temporal statistics based on a sequence of speckle images. The existing analysis methods in the literature are based on intensity variation of the speckle images over time. It is observed that the robustness of these methods can be significantly influenced by the presence of non-uniform incidence and reflectivity, and unrelated image intensity variations. Hence, their applicable circumstances are limited. In this paper, a novel method for dynamic speckle analysis is proposed to address the interference of non-uniform incidence and reflectivity, and unrelated intensity variations. The dynamics of the speckle pattern at a pixel is measured by the temporal statistics for the directions of local normal vectors (rather the intensities) on every speckle image over time. The proposed method can effectively eliminate the aforementioned impacts in a dynamic speckle analysis process. Moreover, the local normal vector at a particular pixel in a speckle image also carries local geometry information regarding to its neighboring pixels. Three algorithms for computing the statistics of the directions of local normal vectors over time are subsequently presented. The effectiveness of the proposed method is demonstrated via the analysis of dynamic speckle patterns measured from a paint drying experiment. The robustness of the proposed method is compared with that of the typical existing methods through processing the dynamic speckle of simulation experiments under non-uniform and varying lighting conditions.