基于Contourlet系数方向扫描格型量化的多描述图像编码算法  

Multiple description lattice quantization image coding algorithm based on coefficients direction scanning in Contourlet

作　　者：王相海[1] 郎丰华[1] 宋传鸣[1] 

机构地区：[1]辽宁师范大学计算机与信息技术学院,辽宁大连116081

出　　处：《辽宁师范大学学报（自然科学版）》2015年第2期184-190,共7页Journal of Liaoning Normal University:Natural Science Edition

基　　金：国家自然科学基金项目(41271422;61402214);高等学校博士学科点专项科研基金项目(20132136110002);辽宁省博士科研启动基金项目(20121076);辽宁省高等学校科学技术计划项目(L2011192;L2013405;L2013406);大连市科学技术基金计划项目(2012J21DW008)

摘　　要：提出了一种基于Contourlet变换方向的多描述格型矢量量化图像编码算法.利用Contourlet变换所具有的多尺度、多方向的特点解决小波变换在多描述格型矢量量化过程中对输入的图像方向信息不充分的局限性.对Contourlet变换后的系数按照子带中方向特征划分区域设计相应的扫描方式对系数进行重新排序;然后采用基于几何相似的六边格矢量量化方案的多描述编码方案对图像进行编码.实验结果表明,文中策略对单路信道解码图像的峰值信噪比较传统方案具有一定程度的提高,同时重构图像较好地保留了原图像的纹理、边缘等几何信息.M ultiple description lattice vector quantization image coding algorithm based on Contourlet transform direction is proposed .The characteristics of multi‐scale ,multi direction of Contourlet is used to solve the limitations of insufficiency of wavelet transform in the process of multiple descrip‐tion lattice vector quantization in directional information of input image .According to the direction characteristics of Contourlet coefficients in the sub‐bands dividing domain ,w hich design correspond‐ing scanning way to reorder coefficients ;As the multiple description coding of vector quantization scheme based on the geometric similarity six sides lattice for image coding .The experimental results show that the strategy of single channel decoding image peak signal‐to‐noise ratio has a certain degree of increase than traditional scheme ,at the same time ,reconstructed image retain the original image geometry information of texture and edge .

关 键 词：多描述编码 二维矢量量化 CONTOURLET变换 扫描策略 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]