Resilient Real Time Hand Gesture Recognition System

Abstract

Gesture based Human Computer Interaction (HCI) is one of the most natural and intuitive modes of communication between individuals and machines. Intelligent gesture recognition systems pave the way for a new era where the focus lies on the task itself, as it should be, and not on the interaction modality. Gestures of the hand itself can be used to communicate effectively as compared to computer peripherals, thereby empowering the visual and speech impaired population. In this project, we propose a technique that employs vision based approach together with dynamic gesture recognition techniques. The prototype architecture of the application comprises of a 2D webcam which continuously records the trajectory of the hand extracting information using a suitable colour model. This project focuses on developing a prototype of a robust, real-time hand gesture recognition system which works well under different degrees of background complexity and illumination conditions for dynamic gestures. The system is cost effective and works efficiently while interacting with objects in virtual environment using hand gestures. As an application, features of media player were controlled with hand gestures and an overall success rate of 97.5 per cent was achieved. While conducting the experiments, we have taken into account the challenges faced due to complex background, presence of non-gesture hand motions and different illumination environments in order to develop a system which can be used for innumerous further applications.