Novel Technique for Background Removal from Sign Images for Sign Language Recognition System

Abstract

The Sign Language Recognition (SLR) system involves recognition of signs and their translation into normal spoken language. The hearing and speech impaired people are deeply associated with Sign Language as it is their fundamental medium of communication. Although such people can easily communicate amongst themselves, they face a serious challenge when they try to integrate into the educational, social and work environments around. The Sign Language Recognition system intends in breaking down the communication barrier between the people who use Sign Language as their only means of communication and others who do not know sign language. Non-uniform background in the edge images is a major challenge for object detection using Gradient operators. The paper discusses a novel technique for background removal in Sign Language Recognition System using the edge images of the ASL signs and morphological operations. Edge sign images are obtained by applying gradient masks (such as Sobel operator) and Slope Magnitude Method. Further with the help of these edge images and morphological operations background is removed. The proposed technique is tested on a generic image database with 312 images.