Fractal coding of color images using earth mover's distance

Fractal theory gained attention by mathematicians and engineers during the last two decades as a promising technique to obtain compression of signals. Fractals also provides interesting interpolation features that can be used to zoom images or to achieve slow motions of video sequences. This paper proposes a fractal coding technique based on Iterated Function Systems (IFS) to encode color image signals. IFS aim to exploit the redundancy given by the self-similarity always contained in natural images. The novelty of this work is on the use of Earth Mover's Distance (EMD) measure to identify similarities within color images. The results obtained show that the proposed approach allows achieving better quality in the decoded images than a classic fractal coding based on Root Mean Square Error (RMS) measure.