Local MAP estimation for quality improvement of compressed color images

概要

論文の詳細を見る
This paper concerns color image restoration aiming at objective quality improvementof compressed color images in general rather than merely artifactreduction. In compressed color images, colors are usually represented byluminance and chrominance components. Considering characteristics of humanvision system, chrominance components are generally represented morecoarsely than luminance component. To recover such chrominance components,we previously proposed a model-based chrominance restoration algorithmwhere color images are modeled by a Markov random field. This paperpresents a color image restoration algorithm derived by the MAP estimationwhere all components are totally estimated. Experimental results show thatthe proposed restoration algorithm is more effective than the previous one.