Indian Sign Language Recognition

Abstract

:- This paper introduces a system that can recognize poses and hand gestures of Indian Sign Language in real time using grid-based features. This system tries to reduce the communication gap between listening and speaking disabled and the rest of society. Existing solutions either provide relatively low precision or do not work in real time. The system provides good results in both parameters. Sign Language is captured from a smartphone camera and its frames are sent to a remote server for further processing. Techniques such as Face detection, object stabilization and skin color are used. Segmentation is also used for hand detection and tracking. The image undergoes a gridbased function extraction technique representing the placement of the hand in the form of Function Vector, Hand postures are classified using the k-Nearest Neighbors algorithm. However, by gesture classification, movement and intermediate positions of the hand observation sequences are entered into the chains of the hidden Markov model corresponding to the pre-selected gestures defined in Indian Sign Language.