Content-prioritised video coding for British Sign Language communication.

Video communication of British Sign Language (BSL) is important for remote interpersonal communication and for the equal provision of services for deaf people. However, the use of video telephony and video conferencing applications for BSL communication is limited by inadequate video quality. BSL is a highly structured, linguistically complete, natural language system that expresses vocabulary and grammar visually and spatially using a complex combination of facial expressions (such as eyebrow movements, eye blinks and mouth/lip shapes), hand gestures, body movements and finger-spelling that change in space and time. Accurate natural BSL communication places specific demands on visual media applications which must compress video image data for efficient transmission. Current video compression schemes apply methods to reduce statistical redundancy and perceptual irrelevance in video image data based on a general model of Human Visual System (HVS) sensitivities. This thesis presents novel video image coding methods developed to achieve the conflicting requirements for high image quality and efficient coding. Novel methods of prioritising visually important video image content for optimised video coding are developed to exploit the HVS spatial and temporal response mechanisms of BSL users (determined by Eye Movement Tracking) and the characteristics of BSL video image content. The methods implement an accurate model of HVS foveation, applied in the spatial and temporal domains, at the pre-processing stage of a current standard-based system (H.264). Comparison of the performance of the developed and standard coding systems, using methods of video quality evaluation developed for this thesis, demonstrates improved perceived quality at low bit rates. BSL users, broadcasters and service providers benefit from the perception of high quality video over a range of available transmission bandwidths. The research community benefits from a new approach to video coding optimisation and better understanding of the communication needs of deaf people