Evaluation of Feature Extractors for Sign Language Recognition

Abstract

Hand gestures play a very dominant role in human daily life to pass the information. The main application of hand gesture recognition is recognition of sign language. Sign language plays a crucial role to communicate with deafness and hearing-impaired community. Modern enhancement in machine learning and computer vision has given a number of techniques for modeling recognition system of sign languages. A work is produced here to evaluate feature extractors and classifiers for sign language recognition system. This recognition system recognizes alphabets (A to Z) presented by signer with different precision depends upon the pair of feature extractor and classifier. This work uses three feature extractors namely Wavelet Transform, Curvelet Transform, Contourlet Transform and two classifiers namely Neural Network and K-Nearest Neighbor which uses supervised learning to classify input data.