International Journal on Advanced Science, Engineering and Information Technology, Vol. 9 (2019) No. 6, pages: 2182-2192, DOI:10.18517/ijaseit.9.6.6267

A Gesture-based Recognition System for Augmented Reality

Vinothini Kasinathan, Aida Mustapha, Asti Amalia Nur Fajrillah


With the geometrical improvement in Information Technology, current conventional input devices are becoming increasingly obsolete and lacking. Experts in Human Computer Interaction (HCI) are convinced that input devices remain the bottleneck of information acquisition specifically in when using Augmented Reality (AR) technology. Current input mechanisms are unable to compete with this trend towards naturalness and expressivity which allows users to perform natural gestures or operations and convert them as input. Hence, a more natural and intuitive input device is imperative, specifically gestural inputs that have been widely perceived by HCI experts as the next big input device. To address this gap, this project is set to develop a prototype of hand gesture recognition system based on computer vision in modeling basic human-computer interactions. The main motivation in this work is a technology that requires no outfitting of additional equipment whatsoever by the users. The gesture-based had recognition system was implemented using the Rapid Application Development (RAD) methodology and was evaluated in terms of its usability and performance through five levels of testing, which are unit testing, integration testing, system testing, recognition accuracy testing, and user acceptance testing. The test results of unit, integration, system testing as well as user acceptance testing produced favorable results. In conclusion, current conventional input devices will continue to bottleneck this advancement in technology; therefore, a better alternative input technique should be looked into, in particularly, gesture-based input technique which offers user a more natural and intuitive control.


hand gesture; e-learning; human-computer interaction.

Viewed: 386 times (since abstract online)

cite this paper     download