Real-Time Gesture Recognition for Human-Computer Interaction Using Hidden Markov Models

International Journal of Computer Science and Technology | Vol. 14, No. 2, February 2012 | pp. 8–14
DOI: 10.46882/2012/IJCST/000120

Research Article

Title: Real-Time Gesture Recognition for Human-Computer Interaction Using Hidden Markov Models

Names of Authors: Oliver G. Bennett¹, Chao-Chuan Lin²

Authors’ Affiliations: ¹Department of Computer Science, University of Bristol, Bristol, UK; ²Institute of Computer Science, National Taiwan University, Taipei, Taiwan

Abstract: Natural Human-Computer Interaction (HCI) systems require real-time hand gesture tracking algorithms to operate without wearable sensor devices or specialized marker gloves. Developing robust systems is challenging due to varying illumination, hand size differences, and the complexity of spatial tracking over time. This paper details a Real-Time Gesture Recognition Framework (RTGRF) designed for standard webcam video inputs. The architecture handles skin color segmentation using an adaptive color space model, tracks centroid positions, and extracts spatial motion vectors. These vectors are subsequently classified into discrete gesture commands using multi-state Hidden Markov Models (HMMs). We validated the RTGRF against a dataset of 12 distinct hand gestures executed by 40 participants. The experimental findings show a mean classification accuracy score of 95.6%. The system maintained low processing latencies (latency <= 18 ms), which supports real-time performance at 30 frames per second on standard consumer hardware. The framework functions reliably under ambient lighting shifts up to ±20%, offering a practical tracking solution for accessible kiosks, surgical displays, and interactive gaming environments.

Keywords: Human-Computer Interaction, Gesture Recognition, Hidden Markov Models, Computer Vision, Real-Time Systems

Manuscript Timeline: Received October 20, 2011; Revised December 03, 2011; Accepted January 05, 2012; Published February 24, 2012.

Citation: Bennett, O. G., & Lin, C. C. (2012). Real-Time Gesture Recognition for Human-Computer Interaction Using Hidden Markov Models. International Journal of Computer Science and Technology, 14(2), 8–14.

Subscribe to read the full article: https://www.internationalscholarsjournals.org/subscribe-to-read