The State of the Art in Pure Fractal Compression
Wednesday, January 17, 2001 - 1:30pm - 2:30pm
Dietmar Saupe (Universität Leipzig)
In pure fractal image compression - devised by Barnsley and Jacquin in 1989 - an image is partitioned into blocks each of which being similar to another typically larger block elsewhere in the image. Similarity is taken modulo intensity shift and scaling. The decoder recovers an image approximation from this self-referential description by iteration. A lot of effort has gone into getting the best performance out of this approach, and we will review the main streams of this work. Of particular interest are the optimization of the partition, reduction of the encoding complexity, and attractor coding. Some of our currently ongoing work addresses optimization of error protection for transmission of fractal codes in noisy channels and progressive fractal image coding.