The Scalable Video Coding (SVC) amendment ofH.264/AVC provides benefits for a variety of video applications. One particular interesting application area is the backward-compatible format enhancement in video broadcast. The coding efficiency in broadcast applications is usually limited due to a frequent insertion of random access points. In this paper, we show that the coding efficiency for SVC broadcast can be improved by increasing the interval between enhancement layer random access points. With the introduction of adequate constraints for random access points and a minor adjustment of the decoding process at a channel change, this improvement can be achieved without any impact on the channel change delay. In our experiments, the coding efficiency for spatial scalable coding became virtually identical to that of single-layer coding while providing the same channel change delay characteristics.
展开▼