stoBEST: an efficient methodology for increased spatial resolution in two-component molecular tagging velocimetry

Abstract

Abstract A new methodology to analyze two-component molecular tagging velocimetry image pairs is presented. Velocity measurements with high spatial resolution are achieved by determining grid displacements at the intersections as well as along the grid lines using a multivariate adaptive regression splines parameterization along the segments connecting adjacent grid intersections. The methodology can detect the orientation of the grid, contains redundant steps for increased reliability, and handles cases where parts of the grid are missing, indicating potential for automation. Initial demonstration of the algorithm’s performance was illustrated using synthetic data sets derived from Computational Fluid Dynamics simulations and compared to Hough-transform and cross-correlation methodologies. Besides providing comparable results in terms of precision and accuracy to previously reported methodologies, the analysis of images by the proposed methodology results in significantly increased spatial resolution of the flow displacement determinations along the grid lines with comparable precision and accuracy. This methodology’s ability to handle different grid orientations without modifications was assessed using synthetic datasets with grids formed by sets of parallel lines at 90, 45, and 30 degrees from the vertical axis. Comparable results in terms of precision and accuracy were obtained across grid orientations, with all uncertainties below 0.1 pixel for images with signal-to-noise levels exceeding 5, and within 0.5 pixel for the noisiest image sets.