The Equivalence Analysis of Image Matching Measure Functions in Digital Speckle Correlation Method

Abstract

Digital speckle correlation method (DSCM) is an import method of non- object image matching, has a lot of measure functions. The formulas have different physical meanings and application areas, each of which has good application examples. The author found, under certain conditions, some of the functions has obvious equivalence representation. In order to verify the above reasoning, the author makes a theoretical deduction from mathematics and establishes three equivalence classes of correlation coefficient formulas.. To validate the equivalence classes, a validation experiment was devised. Two parameters, namely the simple signal to noise ratio (SSNR) and the cross-sectional area at the bottom of the main peak (S_L), which most effectively reflect the shape of the measure function surface, were selected for the comparative analysis of the measure functions. The experimental results indicate that the two parameters, SSNR and S_L, of the measure functions within the same equivalence class are essentially equal and can be regarded as equivalent. The calculation times of each measure function within the same equivalence class were measured, and the optimal formula recommended for selection within the same equivalence class was presented. Based on the above conclusions, in the practices of image registration and image recognition, adopting the best function formula of the same type as the commonly used measure functions may potentially achieve the effect of a significant reduction in computational load while ensuring accuracy.