Spatio-temporal analysis of dynamic speckle patterns using singular value decomposition

The ability to non-destructively visualize transient phenomenon via spatial and temporal statistics of fluctuating speckle patterns is determined primarily by the contrast between regions of differing activity levels. We demonstrate a singular value decomposition-based method for extracting spatio-temporal correlation parameters from temporal sequences of dynamic speckle patterns. Each pattern in the temporal sequence is reduced to a representation using column vectors which are subsequently grouped to form a matrix encoded with localized spatio-temporal speckle intensity data. A correlation metric is defined using the singular values of this matrix and is subsequently utilized for generating a correlation map by interpolating the correlation computed at each individual patch. Via a comparison with synthetic data that we generate using known decorrelation parameters, we are able to show that our algorithm produces activity maps with enhanced contrast that are closest to the ground truth.

Automated summary

A new method is proposed in this study, which extracts spatio-temporal correlation parameters from temporal sequences of dynamic speckle patterns using singular value decomposition and generates correlation maps, showing enhanced contrast compared to other methods.