纹理表达在图像压缩中的应用  

Using Texture Representation in Image Compression

作　　者：孙可[1] 杨盈昀[1] 张勤[1] 

机构地区：[1]中国传媒大学信息工程学院,北京100024

出　　处：《中国传媒大学学报（自然科学版）》2008年第1期12-19,共8页Journal of Communication University of China：Science and Technology

摘　　要：本文针时静止纹理图像展开研究,详细阐述了基于高斯马尔科夫随机场参数和基于样图这两种纹理分析合成技术的原理及其实现,综合二维自相关纹理统计分析方法形成了一套针对不同类型的纹理图像自适应地选择表达合成方法的方案。实验表明,所提方案能够压缩图像数据量,是对新一代图像编码的一种探索。In this paper, we studied still texture images and presented in detail theories and realizations of two texture analysis and synthesis technique respectively based on GMRF parameters and sample pictures. Combining with a 2D-autocorrelation statistic analysis method, we formed a framework which automatically adopted corresponding texture representation and reproduction method in terms of texture types. Experiment results indicated the presented framework can reduce image data quantity, and explored a novel path to actualize new generation image coding.

关 键 词：纹理分析与综合 二维自相关 高斯马尔可夫随机场 图像缝合 

分 类 号：TN919.81[电子电信—通信与信息系统]