Hand sign recognition from depth images with multi-scale density features for deaf mute persons

Abstract

Among many of the fastest growing research fields, sign language recognition is one of the top. Deaf and dumb community uses sign language to express their ideas or views. Sign Language is a methodical coded language where meanings are assigned to every gestures. Many techniques have been developed with the advancement of science and technology to minimize the problem for speech and hearing disabled. The mode of such communication is part of human computer interaction. Hand gesture plays an important role here. The interaction with computer through gesture removes the use of conventional input devices like mouse and keyboards. To create a strong interface between user and computer, recognition of gesture is important. In this paper, a hand gesture recognition method based on multiscale density features is proposed. Depth images of numerals of American Sign Language are considered in this work and recognition rate of 98.20% is obtained, which is comparable with related state-of-the art methods.