Fractal coding of displaced frame difference signals
Nino Ferrario · Library and Archives Canada (Government of Canada) · 1997
Much of the work on fractal video coding has focused on 3-D fractal block coding and coding of original frames using previously reconstructed frames and variations thereof. This thesis investigates the possibility of using fractal image coding techniques for direct coding of displaced frame difference signals. Both standard affine transform and orthogonal basis iterated function system (OBIFS) coders are considered. These coders are generalized for video coding by introduction of alternative domain pool sources. It is shown that the use of these alternative domain pool sources provides much better performance than classical fractal coding of the displaced frame difference signals alone. Extensive simulations are also performed to quantify the effects that various transformation parameters have on coding performance. Finally, the objective and subjective coding performance of the best affine transform and orthogonal basis IFS coders are compared to that of a reference discrete cosine transform (DCT) coder for two standard video sequences. The simulation results indicate that affine-transform-based fractal coders are feasible for direct fractal coding of displaced frame difference signals. The peak signalto-noise ratio (PSNR) performance of these coders was as good or better than the reference DCT-based coder for the two sequences tested. The OBIFS coders, however, are not feasible for direct fractal coding of displaced frame difference signals. The reference DCT-based coder provided much better performance for all sequences tested.