A New Approach to the Phase-Based Video Motion Magnification for Measuring Microdisplacements

Abstract

Video motion magnification (VMM) techniques have recently gained attention as a low-cost accurate method for measuring microdisplacements. VMM techniques are known to have incorrect frequency responses when handling dynamic measurements consisting of multiple frequencies. New phase-based VMM method utilizing radial fractional steerable 2-D Hilbert transform is proposed and contrasted, in terms of the frequency response, with two other established VMM methods. Typical VMM workflow includes amplifying phase changes between subsequent video frames on many spatial frequencies separately using image pyramids, which are then collapsed forming new, motion-magnified video. In this paper, Hilbert transform is used to extract local phase changes. The modified Kullback–Leibler divergence was calculated between the spectral estimates of the time series of vibrating plate measured using a high-speed video camera and a reference time series synchronously acquired with high-precision air-coupled ultrasonic transducer. Time series consisted of micrometer range multicomponent vibrations (one, two, and three sine frequencies) recorded with different motion magnification factors. The results show that in the majority of cases, the proposed method, out of three considered, resulted in the most uniform frequency response and generated the lowest amount of unwanted frequencies. To conclude, a novel approach to VMM is proposed, which substantially improves frequency response and makes this technique more suitable for the measurements of phenomena consisting of multiple frequencies.