Real-Time Hand Gesture Communication System in Hindi for Speech and Hearing Impaired

Abstract

This paper proposes a novel real-time method for gesture to text (G2T) and text to gesture (T2G) in Hindi language, to assist the speech and hearing impaired people (deaf and dumb) to communicate with each other and also with normal people. In G2T system, the images of palm gestures are loaded on real time basis and hand region extraction is done on the basis of skin color segmentation. Feature extraction algorithms are used to identify 32 different gestures on the basis of 5 bit binary string. Gesture recognition algorithm is robust to hand orientation and scale changes and the proposed G2T system is able to recognize images with 98% accuracy. In T2G system, Hindi text in Devanagari is converted into binary string and their mapped corresponding gestures are displayed on screen.