Skip to main content
. 2018 Jul 17;18(7):2312. doi: 10.3390/s18072312

Figure 7.

Figure 7

Ground-truth verification of our algorithm compared against different techniques on SynVideo2 in presence of nonlinear large motion. (a) Top: Original video frame. Bottom: A spatio-temporal YT slice of the video along the red line marked on the input frame. (b,c) Linear EVM [6] and phase-based motion magnification [12]. (d) The ground-truth. (e,f) Motion acceleration magnification [17] and our algorithm. Our algorithm reveals the small motion of interest and best resembles the ground truth, while other techniques generate significant blurring and artifacts.