Computer Recognition of Indian Sign Language

Sign language is the only medium of communication for hearing-impaired people and hard hearing community. This is also helpful in communication between speaking community and who cannot speak. In India the sign language is known as Indian Sign Language (or ISL). English and Hindi are officially accepted sign languages in India. The focus of this research is on computer recognition and translation of ISL in English language domain. A sign language recognition system can be based on a device based approach or vision based approach. Due to problems in device based recognition systems the work presented in this thesis is on vision based computer recognition of ISL. Some pre-processing steps and feature extraction techniques (histogram features structural features and feature vector inherited directly from an image matrix) in the pattern recognition process for performance improvement are devised. A new pattern recognition process for video data set is proposed and is useful in recognition of dynamic gestures of ISL. The thesis contains six chapters.