Algorithm Comparisons among Space Time Volume Velocimetry, Horn-Schunk, and Lucas-Kanade for the Analysis of Water Surface Velocity Image Sequences
Bernadus Herdi Sirenden, Aniati Murni Arymurthy, Petrus Mursanto, Sensus Wijonarko · 2019
Performance of three optical flow algorithm variants has been evaluated to obtain the most optimal measurement for water surface velocity. The algorithms are Space Time Volume Velocimetry (STVV), Horn-Schunk (HS) and Lucas-Kanade (LK). The algorithms were applied to the same video of waterflow, whose direction was rotated by 0°, 90°, 180°, and 270°. Since the water velocity in the video are unknown, the evaluation parameter is varian of estimated water velocity and their repoducibility when the direction of water flow is changed. The results show that the STVV had the smaalest varian 7% to 13%, HS is 11% to 41 % and LK is 17% to 30%. However, the STVV has the most disadvantages compared to the other two methods in terms of reproduciblity when the direction of water flow is changed. The reproducibility of STVV result when using En number as testing method could reach 1,7 while HS and LK is 0 for all experiment.