Comparative Analysis of Indian Sign Language Recognition System

Abstract

Sign language is a way of communication among Hearing and Speech Impaired Persons. Normal peoples can not understand sign language and it is not feasible for deaf-dumb people to bring translator with them in every place. So, for bridging this gap many systems have been developed. Sign Language recognition systems which can convert Sign into text or Speech and vice-versa. Sign language recognition system work in five steps are: data acquisition, pre-processing, feature extraction, classification and recognition. This paper discussed the Indian sign language recognition system. In this paper, a comparative analysis of various gesture recognition techniques involving Artificial Neural Network, Convolutional Neural Networks Hidden Markov Model and PCA has been discussed with its accuracy. This comparative study came across that much work has been done in alphabet and numeric level but work in word and sentence level is less. Sign language recognition for static signs has been done by many researchers but dynamic sign recognition systems have scope of development. A Comparative study is utilized to find out research gaps in existing systems and give inspiration to develop interpreters for Indian Sign Languages.