Abstract
Hearing impaired individuals use sign languages to communicate with others within the community. Because of the wide spread use of this language, hard-of-hearing individuals can easily understand it but it is not known by a lot of normal people. In this paper a hand gesture recognition system has been developed to overcome this problem, for those who don't recognize sign language to communicate simply with hard-of-hearing individuals. In this paper a computer vision-based system is designed to detect sign Language. Datasets used in this paper are binary images. These images are given to the convolution neural network (CNN). This model extracts the features of the image and classifies the images, and it recognises the gestures. The gestures used in this paper are of American Sign Language. In real time system the images are converted to binary images using Hue, Saturation, and Value (HSV) colour model. In this model 87.5% of data is used for training and 12.5% of data is used for testing and the accuracy obtained with this model is 97%.
Highlights
Sign language could be a language that gives communication and permits people with hearing or speech impairments to communicate with one another or with other people within the community
As shown in the Fig.1 hand gesture recognition is divided into different steps, which are image acquisition, image pre-processing, feature extraction, classification, and recognition of hand gesture
Image Pre-processing is an important part of hand gesture recognition which is used for resizing and skin colour detection
Summary
Sign language could be a language that gives communication and permits people with hearing or speech impairments to communicate with one another or with other people within the community. In sign language every gesture contains a specific meaning. Complicated meanings will be justified by the assistance of combination of varied basic components. There are special rules and grammar's for expressing sign language effectively. There are two main gesture recognition approaches, image-based and device based. In those two approaches vision-based approach is mostly used, since there is no need of using any sensors or gloves to detect gestures. Revised Manuscript Received on March 11, 2020
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
