Indian Sign Language Recognition Using a Novel Feature Extraction Technique

Sign language is a means for communication between deaf mute people and people with normal speaking and hearing abilities. There is no way other than the sign language to be useful for deaf mute community. Roughly, there is around 5% of global community is using sign language; there is a need for automatic recognition of sign language. The Indian subcontinent community using sign language for communication purposes is known as Indian sign language. In this paper, a system for automatic conversion of isolated character-level ISL signs to text is presented. A total of 2600 input images, 100 images per ISL single-handed English character are used in the research. In experimental work, hierarchical centroid feature extraction technique with k-nearest neighbor and Naive Bayes classifiers is used. A satisfactory result of around 97% accuracy has been attended in the experiments.