Sign language is the normal mode of communication for many deafand non-speaking individuals. The problem with sign language is that to communicate, all persons involved must understand sign language and be in plain sight of each other. A systems is needed that can encode sign language and convert it to the English language text that it represents, so that remote signers and those not versed in sign language can communicate with each other. Digital Video Processing, Inc. will develop a sign language recognition and translation system to allow users versed in sign language to more readily communicate. The system will harness real-time motions and positions and encode this information. A sign language recognizer will interpret the data and relate it to a dictionary of segments (letter, words, etc.) and choose theappropriate segment. The Phase I effort will address the significant problems and opportunities associated with encoding the inputs, extracting the input's information content, and recognizing and translating particular segments. A complete hardware and software specification for the prototype systems will be detailed. Also, proposed algorithms for real-time sign language recognition will be described, and the ability to recognize sign language given the social, behavioral, and geographic differences inherent in any form of communication will be ascertained.
