Fast fractal image compression with triangular multiresolution block matching

This paper presents an approach to fractal image compression whereby we reduce the computation to the one dimensional case using a triangular Sierpinski scan path algorithm which results in a binary triangulation tree structure. The scanned image is naturally partitioned into range blocks of contiguous intensities where each block maps to a triangle in the two dimensional image. A preliminary compression is achieved by pruning the binary triangulation tree by combining similar adjacent triangles with a compatibility constraint to ensure that the tree levels of the triangles do not differ by more than one. A wavelet based fractal image compression algorithm is applied to the pruned tree and the parameters are saved for the construction of iterative function systems (IFS) whose fixed point will approximate the pruned image