数字图像离散小波变换的原理与硬件实现分析  

Principles and architectures of digital image coding with discrete wavelet transform

作　　者：孟军[1] 魏同立[1] 吴金[1] 常昌远[1] 

机构地区：[1]东南大学微电子中心,南京210096

出　　处：《东南大学学报（自然科学版）》2002年第6期842-847,共6页Journal of Southeast University：Natural Science Edition

基　　金：复旦大学专用集成电路与系统图像重点实验室开放课题资助项目

摘　　要：针对日益进步的图像变换编码技术 ,对目前已经纳入MPEG 4和JPEG2 0 0 0编码标准的采用离散小波变换进行数字图像编码的原理与硬件实现进行了综述介绍 .在分析小波变换快速算法的基础上 ,重点讨论了近 1 0年来所提出的各种离散小波变换硬件实现的典型结构 ,在硬件资源与处理速度两个方面进行了比较 .对于变换后的系数量化 ,总结了几种基于嵌入式零树小波编码的算法 ,比较其峰值信噪比和编码时间 .相较于离散余弦变换进行图像编码 ,采用离散小波变换在压缩效率。The principles and architectures of digital image coding with discrete wavelet transform (DWT), the advanced image coding method, which is the standard method of image coding in MPEG-4 and JPEG2000, are discussed. After the introduction of DWT and fast DWT algorithm, typical architectures used to realize DWT are given concerning their hardware cost and speed. To achieve the coefficient quantization, some algorithms based on embedded zero-tree wavelet (EZW) coding are listed. Their peak signal noise rate (PSNR) and coding cycles are compared. Finally prospect for image coding is made. Compared with the discrete cosine transform (DCT), the DWT can offer more advantage in coding efficiency and decoding quality.

关 键 词：原理 硬件实现 图像编码 离散小波变换 量化 数字图像处理 

分 类 号：TN919.81[电子电信—通信与信息系统] TN9[电子电信—信息与通信工程]