Abstract
Usual compression methods for limited color images do not use effectively two-dimensional context of images. In this paper, we propose a variation of finite context model to improve compression ratio. In this method, a context is made from the neighborhood of the current pixel to refer to two-dimensional information. We also propose a technique to reduce execution time using simply structured model. Instead of color indexes, a context is constructed from labels which represent the location of pixels with the same color in the neighborhood. We will also show the results of compression experiments.
