Vision-Based Of Continuous Sign Language Recognition

Sign Language Express ideas by Shape of the hand, arm movement and face Expression, lip movement and other body potential. It also have standardized syntax, clear semantics and complete vocabulary system. Sign Language Commonly used among the deaf for information exchange and communication. For Ordinary people, they do not understand sign language, so there are obstacles in the communication with deaf people. we would develop a system based on visual of sign language recognition to remove the obstacles. We translated the sign language into text and voice in really-time, so people will Understanding sign language by the deaf people. Research work includes the following aspects:1. Find the basic elements of sign language:Chinese sign language as an important branch of the sign language, divided into two categories:Finger language and Sign Language, for finger language is just Phonetic Composed by alphabet, so the most research work is on the Sign Language.2. Gesture segmentation:We will use HOG-SVM for the hands segmentation in the image.3. Static gesture feature extraction and recognition:we will get the HOG Value, than use the SVM distinguish the static sign language alphabet.4. Dynamic Gesture Tracking:Hand is the non-rigid objects that will have deformation in moving, TLD algorithm because of it excellent online learning ability so it can track non-rigid objects. so we use the TLD for the Dynamic Gesture Tracking.5.Classification of sign language:We train the HMM model based on the word. than recognition of sign language by comparison the HMM model.