Colour-based gesture recognition for American Sign Language via Hidden Markov Models
Sara Greenberg, Jennifer Blight, Alexander Wong · Vision Letters · 2015
We present a new approach to gesture recognition for use in a sign language learning environment. This method utilizes inexpensive cloth gloves to alleviate the difficulty of hand detection and to allow for feature creation. Salient colours identify the glove base and fingertip markers, which are then used to extract a hand centroid and a convex hull describing the fingertips for each hand. A Hidden Markov Model is created for each sign, as well as an additional threshold model created from all signs. When a candidate sign is performed, the sign of the HMM that produces the greatest likelihood is matched, provided it also exceeds the threshold model likelihood. Isolated recognition testing of the training library indicated 76% accuracy, and continuous recognition testing showed 60% accuracy.