Vision system for hand gesture recognition

• Main Authors: Ilao, Joel P., Enero, J. M., Gonzalo, K. D., Salvador, R. P., Ticzon, E. I.
• Format: text
• Published: Animo Repository 2007
• Subjects: Gesture recognition (Computer science); Computer Sciences
• Online Access: https://animorepository.dlsu.edu.ph/faculty_research/6925
• Institution: De La Salle University

Vision system for hand gesture recognition (VISOR) is a system that recognizes a set of dynamic continuous gestures, using only a single hand, from a predefined vocabulary using computer vision algorithms. The system’s vocabulary is limited to five (5) primary gestures, each representing a command: forward, backward, left, right, and stop. VISOR is user-independent, able to contend with different backgrounds and does not require the user to wear a long-sleeved garment – a limitation commonly found in similar systems. A standard USB web camera is used to capture gesture sequences. The hand is detected and afterwards tracked in every frame captured. This paper describes the design and implementation of the hand gesture recognition through hand detection and tracking, and by extraction of temporal and structural features of the tracked hand. The developed system’s performance is then evaluated in an indoor environment. It ends by a discussion of test results and recommendations for future work.