摘要
本文首先提出了虚拟信源概念和基于虚拟信源构建数据压缩方法的思想 ,设立了 0与 1的字符长串的通用虚拟信源Y ,然后用神经网络建立了虚拟信源Y的模型 ,又用该模型和一个取整函数构造出了基于虚拟信源的无损数据压缩方法 .实验表明 ,一些情况下此压缩法的压缩比为 3∶1.该压缩方法体现了虚拟信源建模思想 ,它与已有的熵编码 (无损 )压缩方法考察数据压缩问题的角度完全不同 ,它能压缩一些已用熵编码压缩过的数据 .它与高保真的小波编码方法结合能获得一些高压缩比又高保真的数据压缩实例 .
An idea of virtual information source and the formation of data compression,based on a virtual information source is put forward and a general virtual information source Y of long character bunch of 0 and 1 is set up.Then the model of the virtual information source Y is established by neural network and the lossless data compression based on virtual information source is constructed with the model and an integer function.The experiments show that the compression ratio achieved maybe is 3∶1.The data compression embodies the idea of modeling of a virtual information source.It differs from the old entropy coding(lossless) and can compress some data which have been compressed by the old entropy coding.Some examples of data compression of high compression ratio and hifi can be obtained by combined wavelet coding with the compression method.
出处
《电子学报》
EI
CAS
CSCD
北大核心
2003年第5期728-731,共4页
Acta Electronica Sinica
基金
国家自然基金 (No 60 0 750 1 2 )
山东省自然基金 (No Y2 0 0 2G1 6)
关键词
数据压缩
虚拟信源
元损压缩
神经网络
BP算法
data compression
virtual information source
lossless compression
neural network
BP algorithm