COHST and Wavelet Features Based Static ASL Numbers Recognition

Bridging communication gap between the deaf and dumb people with the common man is a big challenge. A sign language recognition system could provide an opportunity for the deaf and dumb to communicate with non-signing people without the need for an interpreter. Research in the area of Sign language...

Full description

Saved in:
Bibliographic Details
Published inProcedia computer science Vol. 92; pp. 455 - 460
Main Authors Thalange, Asha, Dixit, S.K.
Format Journal Article
LanguageEnglish
Published Elsevier B.V 2016
Subjects
American Sign Language numbers
Combined Orientation Histogram and Statistical (COHST) Features
Gesture Recognition
Neural Network
Orientation Histogram
Statistical Features
Wavelet Features
Combined Orientation Histogram and Statistical (COHST) Features
Gesture Recognition
Wavelet Features
Neural Network
American Sign Language numbers
Statistical Features
Orientation Histogram
Online AccessGet full text

Cover

More Information
Summary:Bridging communication gap between the deaf and dumb people with the common man is a big challenge. A sign language recognition system could provide an opportunity for the deaf and dumb to communicate with non-signing people without the need for an interpreter. Research in the area of Sign language recognition has become very significant due to various challenges faced while capturing of the sign. Not a single efficient methodology or algorithm is developed which overcomes all the difficulties and recognizes all the signs with cent percent accuracy. This paper proposes two new feature extraction techniques of Combined Orientation Histogram and Statistical (COHST) Features and Wavelet Features for recognition of static signs of numbers 0 to 9, of American Sign Language (ASL). The system performance is measured by extracting four different features of Orientation Histogram, Statistical Measures, COHST Features and Wavelet Features for training and recognition of ASL numbers individually using neural network. It is observed that COHST method forms a strong feature than the individual Orientation Histogram and Statistical Features giving higher average recognition rate. Of all the System designed for static ASL numbers recognition, Wavelet features based system gives the best performance with maximum average recognition rate of 98.17%.
ISSN:1877-0509
1877-0509
DOI:10.1016/j.procs.2016.07.367