Turkish Sign Language Animation with motion capture

摘要

In this paper, using motion capture, we study on Turkish Sign Language Animation. A markerless system, Microsoft Kinect sensor and a professional optical system are compared. Besides, we use the motion capture data in two different approaches; data driven and real time. To animate the sign language, we merge the motion capture output with a cartoon character. In the first approach, which is data-driven, motion capture data is recorded, processed and converted to a character animation off-line. This data is stored in a database and it can be used in mobile devices or any other necessary ways. Real-time approach on the other hand converts the motion capture data into a character animation in real-time. This animation can be used in place of a real signer, which would attract the attention of people easily. TV channels, conferences, cinemas are some examples of its usage. Both methods can use the optical system and the markerless system. However, we see that Microsoft Kinect sensor is not sufficient to track the fingers of the signer, so this is not an appropriate solution yet. On the other hand, optical systems give accurate results, which can be used for Turkish Sign Language animation in both data-driven and real-time approaches.