3D Signing Avatar for Sinhala Sign Language

Abstract

Animating signs of a sign language requires on demand bone movement of a signing avatar based on single posture, multi posture signs for the words with a known sign, and fingerprinting signs for representing unknown words character by character. Ordering and sequencing of different types of signs corresponding to words/phrases of a typical spoken/written language are the most essential features of a 3D signing avatar. Typical techniques of signing avatar animation are video sequencing of signs of corresponding words of a sentence, or replaying movements of signs with motion-captured animation sequences. This paper presents a multi-facet 3D avatar and an animation framework that supports the definition and animation of sign gestures without motion capture hardware. The system is developed to accommodate any sign language, but a prototype has been built for the Sinhala sign language (SSL). The avatar system is initially built with 10 bones per arm. It is then extended to 29 bones per arm to improve flexibility in palm and fingers. The avatar animation framework is capable of signing 200 plus gestures of both single and multi posture SSL signs and 40 fingerprinting SSL signs. Speed and uniformity of the sign gesture animation is achieved by the automatic calculation of intermediate sequences of arm movements of signs within a given number of frames, or with a user defined increment value per bone. Animation system also supports the definition of fingerprinting signs, with a user defined tag set corresponding to the alphabet of any given sign language.