Effective Temporal Scalability by Selective Update Operation for Wavelet Based Scalable Video Coding

The scalable video coder generates embedded bit stream that will satisfy all the end users irrespective of their capacities. Lifting implementation of wavelet transform is a benchmarked technology for temporal decomposition of video in a scalable video coding. The temporal filtering is done along the motion trajectory in Motion Compensated Temporal Filtering (MCTF) framework and the motion model failure in particular region results in ghosting artefacts in the low frames, hence temporal scalability will be at stake in such scenario. This paper proposes a novel idea which selectively applies the update step of lifting wavelet transformation during temporal analysis stage. The technique identifies the failure of motion model and update operation will not be applied for such block. The resulting technique reduces the ghosting artefacts in the low frequency frame and hence supports for efficient temporal scalability. There will be an additional cost of one bit per block to indicate the status of update step for a particular block.