| Abstract |
Optimal fractal coding consists of finding in a finite set of contractive
affine mappings one whose unique fixed point is closest to the
original image. Optimal fractal coding is an NP-hard combinatorial
optimization problem. Conventional coding is based on a
greedy suboptimal algorithm known as collage coding. In a previous
study, we proposed a local search algorithm that significantly
improves on collage coding. However, the algorithm, which requires
the computation of many fixed points, is computationally
expensive. In this paper, we provide techniques that drastically
reduce the time complexity of the algorithm. |