A Novel Error-resilient Video Codec Framework Based on Scene-Context Reference Picture

Abstract

The transmission of video over wireless networks always confronts the risk of packet-losses and bit-errors, which result in the rapid propagation of errors in both time and space and the quick of degradation of the reconstructed video quality. In order to stop the error propagation, the traditional method is intra-frame refresh. This paper proposes a novel error-resilient video codec framework utilizing the predictive picture refresh based on scene-context reference picture. When the unreliable mobile IP network requires high intra-frame-refresh rates (In our experiment, we define that an I-frame is sent once every 20 sequence frames at a frame rate of 30 fps), compared with the traditional method, our method can achieve over 10% saving effect in frame bits while maintaining the similar quality incorporated in the H.264/AVC Joint Model 9.0