Research Article Open Access

Robust Hand Gesture Recognition for Human Machine Interaction System

Abstract

Hand gesture is the form of non verbal communication to convey particular message by using the visible movements and posture of hand. It is interpreted by using a recognition system that can be used for interfacing between humans and computer devices. The interfaces based on hand gesture recognition (HGR) can be used for a wide range of applications like sign language recognition, virtual gaming, automation and security. The present work represents a technique for human machine interface (HMI) using HGR system. This system is able to recognize five different hand gestures with remarkable accuracy. The proposed system is tested for five different environmental and physical conditions. Best techniques of image processing implemented to make system more and more robust. The segmentation scheme used in this system was based on grey level thresholding. The Y component reduced from test image and Cb, Cr components separately used to extract the particular hand area. This helped in making system more robust for light varying environment. Further binary image has been used to extract contour of hand area. For this canny edge detector was used. After successful edge detection contour scale space used to find out the raised finger count, hence hand gesture recognized according to the number of fingers raised. On MATLAB successful implementation of proposed system yielded overall average accuracy of 95.2% varying from 96% for good light and 92% for bad light conditions. The whole system took 0.8 sec to recognize a complete hand gesture in MATLAB 2012b.

To read the full article Download Full Article