期刊名称：International Journal of Innovative Research in Science, Engineering and Technology
印刷版ISSN：2347-6710
电子版ISSN：2319-8753
出版年度：2014
期号：ICETS
页码：914
出版社：S&S Publications
摘要：The mixed raster content (MRC)standard specifies a framework for documentcompression which can dramatically improve thecompression/ quality tradeoff as compared totraditional lossy image compression algorithms. Thekey to MRC compression is the separation of thedocument into foreground and background layers,represented as a binary mask. Therefore, the resultingquality and compression ratio of a MRC documentencoder is highly dependent upon the segmentationalgorithm used to compute the binary mask. Theincorporated multi scale framework is used in order toimprove the segmentation accuracy of text withvarying size. In this paper, we propose a novel multiscale segmentation scheme for MRC documentencoding based on the sequential application of twoalgorithms. The first algorithm, cost optimizedsegmentation (COS), is a block wise segmentationalgorithm formulated in a global cost optimizationframework. The second algorithm, connectedcomponent classification (CCC), refines the initialsegmentation by classifying feature vectors ofconnected components using a Markov random field(MRF) model. The combined COS/CCCsegmentation algorithms are then incorporated into amulti scale framework in order to improve thesegmentation accuracy of text with varying size.
