Compressing color-indexed images by dynamically reordering their palettes

A new lossless compression scheme for coding color-indexed images is proposed. This scheme first turns the index map of a color-indexed image into a new index map each element of which serves as an index to a pixel-dependently reordered version of the palette. This process reduces the zero-order entropy and the energy of the index map significantly, and hence a matching coding scheme can then be exploited to encode the new index map effectively. Simulation results show that the proposed method is superior to other state-of-art lossless compression algorithms for coding color-indexed images in terms of compression performance.