Abstract: Communication is the process of exchanging information, views and expressions between two or more persons, in a verbal and non-verbal manner. Hand gestures are the non verbal method of communication used along with verbal communication. A more organized form of hand gesture communication is known as sign language. Physically disabled persons like the deaf and the dumb use this language to communicate among themselves and with others. The main aim of this paper is to design a system that can modify the sign language accurately so that the less fortunate people may communicate with the outside world without an interpreter. By keeping in mind the fact that in normal cases every human being has the same hand shape with four fingers and one thumb, this paper aims at designing a real time system for the recognition of some meaningful shapes made using hands. In this work Tamil signs are synthesized by combining the texture and the shape extracted from the already trained model. K-nearest neighbor (KNN) classifier, Sequential forward and backward features are used for the synthesis.Experimental results demonstrate the effectiveness of the proposed work (recognizing efficiency 91%).
Keywords: Active shape model, Landmark points, Sign Language, Tamil Language.