Low bit-rate video coding using fractal compression of wavelet subtrees

Fractal compression is a relatively recent compression technique; built on local self-similarities, it is well known for its high compression ratio. However, in image compression, fractal coding techniques have not produced performance comparable to modern wavelet coders. The discovery of the link between fractal compression and wavelets has improved the performance of these coders. Due to the self-similarity property, fractal methods are most suited to video coding since there is high spatial and temporal correlation between video frames. Each frame is decomposed using the pyramidal multiresolution wavelet transform. Then, a motion detection operation is used in which the subtrees are partitioned into motion and non-motion subtrees. The non-motion subtrees are easily coded by binary decision, whereas the moving ones are coded using the fractal variable tree size coding scheme. Subtrees that are found to contain many motion structures are coded with the SPIHT algorithm. The domain subtrees for fractal coding are constructed from the previous frame instead of the current frame. This then removes the iterative decoding process and provides relatively fast decoding. All intra-frame compression is performed using the wavelet SPIHT coder. Simulation results show that the proposed scheme has comparable performance to the H.263 compression standard, making it suitable for low bit rate video