Sign language recognition using color means of gradient slope magnitude edge images

Abstract

Sign Language is a method of communication for hearing and speech impaired people in which hand movements, gestures and facial expressions are used to convey messages. The hearing and speech impaired people are deeply associated with Sign Language as it is their fundamental medium of communication. American Sign Language (ASL) is a complete, visual-gestural language that employs signs made by moving the hands combined with facial expressions and postures of the body. The paper discusses novel Sign Language Image Retrieval techniques using the edge images of the ASL signs. Edge images are obtained by applying gradient masks and Slope Magnitude Methods. The proposed image retrieval techniques are tested on generic image database with 312 images. Feature vector of sign images are extracted using color averaging techniques. In all 5 techniques are experimented and sign images are compared using 5 masks (Prewitt, Robert, Sobel, Laplace, and Canny) and 5 averaging techniques. The GAR (Genuine Acceptance Ratio) values indicate the best performance values.