In this thesis, the real-time Korean Sign Language(KSL) recognition system is presented. Sign language, which is usually known as a set of natural language with formal semantic rules, is a large set of hand gestures that are daily used to communicate with the hearing impaired. KSL can be analyzed statistically according to 4 basic elements in a gesture : 24 motion-direction, 23 posture, 10 orientation and 21 location. The semantics of a sign language vocabulary can be different when gesticulating in different locations relative to the human body. Hand locations can be recognized using two approaches. One is color cording and color transform and the other is model-based searching and tracking. Glove-based device system and vision-based device system are combined to KSL recognition system. Glove-based device gets motion- direction, posture and orientation elements and vision-based device gets location element. About 450 Korean signs and 31 Manual alphabets can be recognized. The average recognition rate for KSL is about 85.1%.
