Computationally efficient fine grained scalability for low bit rate video coding

Real time transmission of video contents of optimal quality over a varying bandwidth network like Internet is a challenging task. A codec, in this scenario, should be able to cope with the packet loss and bandwidth variation for the transmitted stream. Fine Granular Scalability (FGS) has been adopted in MPEG-4 standard to provide optimal video quality over a range of bandwidths. FGS is well suited to coping with varying bandwidths but it has the drawbacks of involving expensive computations and the fact that the enhancement layer adds considerably to the overall size of the coded bit stream. This may lead to congestion in a low bit rate connection. This paper suggests structural improvements in FGS codec so that it can be efficiently used for low bit rate connections in real time applications. The proposed scheme is based on open loop and closed loop enhancement predictions that can provide up to 35 times reduction in coding time.