Lossless coding techniques for color graphical images

Summary form only given. Two types of images are considered: line-art images and computer generated graphics. The line art format specifies the standard representation format for digital images used in electronic pre-press systems. Colors are defined in a palette table that specifies the values of color components for each entry in the palette. The proposed techniques have been extended to full color graphics. The input data is transformed into color spaces which tend to concentrate a significant amount of the signal modulation into a single channel, thereby reducing the entropy in the other two channels. Each component channel is then processed by an independent spatial coder. A lower bound on the color entropy for the sample set of images is being derived by means of the optimal linear KL transformation to obtain the principal color components. An alternative method of compressing color graphics is described which utilizes a spatial DPCM codec operating directly on the composite color signal. Various approaches for determining the optimal values for the color correlation matrix are explored and comparisons made on the example image set