Error reduction in molecular tagging velocimetry via image preprocessing

The effect of preprocessing molecular tagging velocimetry (MTV) images to reduce measurement error was studied using simulated and experimental images with signal-to-noise (SN) ratios of SN = 2–16. The results of the simulations showed that image filtering reduced the measurement error by up to 30 % for conditions typically seen in real-world MTV experiments. Under some conditions (i.e., thin lines or large spatial filters), filtering was found to increase the measurement error. Experiments confirmed the simulation results, although the actual error levels were higher. The use of an averaged initial or “undelayed” image, instead of individual undelayed images, was also investigated. This strategy increased the SN of the undelayed image by averaging out the random noise. It was shown that the use of an averaged undelayed image reduced error for low SN images but potentially increased error for high SN images.