Optimized H.264 compression of sign language video

Abstract

In TV broadcasting deaf people are not able to get information from the audio content. In public television, some programs may be accompanied by a sign language interpreter as a part of the broadcasted signal. As a supplementary service, it would enable more programs to be accessible with a sign language interpreter to assist in comprehension. To be able to transmit such data flow separately, we define the parameters of compression of sign language interpreter image to ensure intelligibility and quality while maintaining low bitrate. This paper deals with specific video compression of Czech sign language interpreter based on regions of interest implemented to the ×264 open source library. The results of this approach are verified in subjective tests with the deaf and hearing evaluators. The experiments examine the intelligibility of sign language expressions containing minimal pairs for different levels of image compression and also evaluate the subjective quality of the final image.