Abstract:
Demonstration of sign languages with the computer is a potentially useful learning aid for sign language learners. If implemented as a part of a learning tool, one that includes sign recognition as well, it will invaluable for providing feedback to the learners, a most needed contribution. Human body animation and motion capture technologies have reached a point where realistic virtual actors can perform plausible human movements in realtime. For this, the motion can be defined on a virtual human skeleton, either by design or by motion capture methods, and then displayed over the skeleton which drives a realistic skin model, visualizing the human body. In this work we capture Turkish sign language finger spelling alphabet and semiautomatically translate it into a visually appealing model. For capturing the sign language we use a magnetic motion capture system. Then, a playback tool generates sign language demonstrations interactively and in realtime in 3D.
