Sign language recognition using a cheremic architecture

In this paper we analyse the gestures that compose British Sign Language (BSL), where an immediate response or detection of a single gesture is not required but instead a sequence must be translated with a small delay to allow for contextual information to be considered. BSL is a natural language that uses two communication channels comprising hand/body postures (manual channel) and facial expressions (non-manual channel) to communicate information. In this paper, analysis is restricted to the manual channel only