Systematic interrogation area optimization in large-scale particle image velocimetry using information entropy

Abstract

This study introduces an Information Entropy-based method for determining optimal Interrogation Area (IA) size in Large-Scale Particle Image Velocimetry (LSPIV), a crucial factor for enhancing non-contact surface flow measurement accuracy. By analyzing entropy in particle images across variable IA sizes, we assessed 48 synthetic and 2 experimental flow scenarios. The method demonstrated superior accuracy, achieving Vector Correlation Coefficients up to 1.916 and Root Mean Square Errors as low as 1.113 and 2.444 pixels/frame in synthetic cases, and accuracy rates of 90.89% and 97.23% in experimental cases, rivaling traditional empirical approaches. Incorporation of surrounding pixel intensity data resulted in a 48–52% improvement in particle information quantification. Expanding the range of IA sizes from 5 to 8 significantly reduced measurement errors to below 0.7 and 1.0 pixels/frame. These findings suggest that the Information Entropy method offers a robust framework for systematic optimization of IA selection in LSPIV, promising enhanced measurement accuracy through further refinement of convergence criteria and noise reduction techniques.