Image Compression System Using Colorization and Meanshift Clustering Methods

摘要

We propose a color image compression system which uses the colorization method used in the computer graphic society for automatic colorization and the meanshift clustering method. The encoder makes use of the meanshift clustering algorithm in automatically selecting the representative pixels(RP) from the original image from which the colored image is reconstructed in the decoder. Using the basin of attraction of the modes obtained by the meanshift clustering as the RP, the compression rate becomes high and the reconstructed image has good visual quality. Experimental results show that the proposed colorization method can provide good visual quality and reduce the smearing artifact which often occurs in other colorization based compression algorithms.