Continuous Number Signs Recognition

Sign Language is the language of gestures used by mute people for non-verbal communication. This paper is about a novel approach for the recognition of number signs of Pakistan Sign Language. An image sign is captured from running video when signer nods head. Deep pixels-based analysis of the scanned image is pursued for the recognition of fingers (from index to small finger) while thumb´s position is find out through Template Matching. After fingers identification, Dynamic signs are recognized based on fingers´ state of being raised or down. For quick detection, signs are categorized into four groups which the algorithm identifies following a model of seven phases.