Open AccessAn Optical Flow-Based Gesture Recognition Method
Author(s) -
Dániel Zoltán Nagy,
Imre Piller
Publication year - 2020
Publication title -
acta marisiensis. seria technologica
Language(s) - English
Resource type - Journals
ISSN - 2668-4217
DOI - 10.2478/amset-2020-0005
Subject(s) - optical flow , workflow , robustness (evolution) , computer science , gesture , gesture recognition , computer vision , artificial intelligence , representation (politics) , variance (accounting) , pattern recognition (psychology) , image (mathematics) , database , biochemistry , chemistry , accounting , politics , political science , law , business , gene
The efficient human-machine interaction is an essential and current problem of computer science. The paper presents a gesture recognition method which applies optical flow calculation and an aggregation for obtaining a heatmap-like representation of the motion trajectories. After the overview of the image processing workflow, the paper introduces six symbols for providing some measurements. The described experiments show the robustness of the method against color, shape and time variance.