Parameter based synthesis of sign language

The primary objective of this project is to demonstrate synthesis of signs in American Sign Language with a finite number of parameters. The parameters in this method include 22 key locations, six location modifiers, six orientation indicators, 46 hand shapes, six predefined movements and seven possible end effectors. These parameters can be used in conjunction with a computer script and they maximize precision and versatility while minimizing development time and resources. The parameters for each sign are stored in an individual text file averaging approximately half a kilobyte. The success of this project is demonstrated using UGS's Jack Software, Version 4.0. Jack is a tool for the animation of human movement. The human Jack is modeled using joint angle limits and inverse kinematics methods when moving any limb to a destination. A Python interface to Jack can take the English equivalent of the sign as an input, call its parameters and command Jack to sign. A graphical user interface (GUI) was constructed using MATLAB 6.5 as a user friendly tool to help grow the database of parmetrically coded signs. The GUI takes all the parameters identified for each sign and prints them to a file. The GUI is easily expandable and new parameters will be implemented in the future. The parameters of these signs will be used in the future as the foundation for sign recognition technology. Animation of sign language is a significant step toward a fully integrated ASUEnglish translator.

Name thesis:degree_name

Master of Science in Biomedical Engineering - (M.S.)

Discipline thesis:degree_discipline

Biomedical Engineering

Year

2005

Author dc:creator

• Irving, Amanda

Contributors dc:contributor

• Richard A. Foulds
• William Corson Hunter
• Sergei Adamovich

• American Sign Language
• Synthesized sign language
• Biomedical Engineering and Bioengineering

Repository record dc:identifier

https://digitalcommons.njit.edu/theses/463

OAI identifier oai:identifier

oai:digitalcommons.njit.edu:theses-1462