Indian sign language recognition in real time for deaf and dumb.

A system which recognizes hand poses and gestures from the Indian Sign Language (ISL) in real time using grid-based features is presented in this paper. An attempt is made to minimize the gap between the hearing and speech impaired and the rest of the people in the society. The solutions that are existing now neither provide relatively high accuracy nor do good work in real time. Good results on both these parameters are provided in this paper. Some gestures and 33 hand poses from the ISL can be identified using grid-based features. A smart phone camera is used to capture the sign language and are transmitted to remotely placed server in the form of frames. It is user friendly and external hardware usage is thus avoided. For hand detection and tracking, techniques such as skin color segmentation, face detection and object stabilization are used. Further to represent the hand's pose in the form of feature vector, the image grid-feature extraction technique is implemented on the image. K-nearest neighbors' algorithm is used to classify the hand poses. Hidden Markov Model chain is fed with the motion and observation sequence of intermediate hand poses to classify the gestures in which 12 gestures are predefined in ILS. [ABSTRACT FROM AUTHOR]