RESUMO
Raw data compression is mandatory for monitoring of processes by dynamic speckle analysis when two-dimensional activity maps are built by pointwise statistical processing of correlated speckle patterns formed on the surface of diffusely reflecting objects under laser illumination. Coarse quantization of speckle patterns enables storage and transfer of a huge amount of images, but it may be inefficient at spatially varying speckle statistics, such as for patterns recorded at non-uniform illumination or reflectivity. We prove efficacy of coarse quantization of the raw speckle data with varying statistics for a normalized algorithm by simulation and a polymer drop drying experiment. Both uniform and non-uniform quantization are proposed for treating such data. Decreasing the bit depth from 8 to 3 is possible without worsening the quality of the activity map.
