A data compression algorithm for color images based on run-length coding and fractal geometry

An approach to data compression for NTSC composite video is introduced that uses run-length coding and concepts of fractal geometry adapted for color image compression. Compression is achieved by sending only the run lengths with a sign bit. A feedback mechanism is constituted by defining the encoder and decoder as finite-state machines. The edges are preserved by sending the edge information using a separate edge quantizer. The proposed system takes the characteristics of the human visual model into account. The low computational complexity of the algorithm makes it conducive to an easy real-time implementation. Good-quality reconstructed color images are obtained at 1.25 b/pixel.<>