Temporal and Spatial Scalability with 3D Wavlet Transforms

In today’s streaming video content delivery infrastructure scalability is a highly desired feature. Not all the network enjoy the same bandwidth and as such the video servers should be able to provide different resolution and frame rate versions of the requested content without the need to re-encode the content to the desired bandwidth. Also video servers should be able to change the frame rate and/or resolution during a presentation to sustain any changes in Network’s behavior. Existing hybrid coding schemes like MPEG-1 and MPEG-2 which use block-based DCT+MC(Motion Compensation) are not optimal for temporal or spatial scalable coding. Although there have been provisions in MPEG-2 Standard [6] for both of these scalabilities but these techniques add extra complexity to both encoder and the decoder. In contrast the Temporal and Spatial Scalability comes almost for free with 3D Wavelet (or Sub-band) coding [3] [2]. In this Work we have implemented a 3D wavelet coder to measure the efficiency of a 3D wavelet coder for temporal and spatial scalability based on Lifting implementation of wavelet transform over the temporal axis and simple 2D wavelet Transform over Spatial domain. We also used this Coder to investigate the efficiency of a new coder that we proposed by introducing the concept of B frames in 3D Wavelet coder.