分形图像压缩技术在神经网络中的实现  

Implementation of Fractal Image Compression Technology on Neural Network

作　　者：丁姗[1] 刘希玉[2] 周晓一[1] 

机构地区：[1]山东师范大学信息科学与工程学院,山东济南250014 [2]山东师范大学管理与经济学院,山东济南250014

出　　处：《计算机技术与发展》2010年第11期55-58,共4页Computer Technology and Development

基　　金：国家自然科学基金重大项目(60873058;60743010);山东省自然科学基金重大项目(Z2007G03)

摘　　要：由于分形图像压缩技术具有解码分辨率无关性、快速编码及高压缩比和低损耗率等特点而被广泛应用,但基于迭代函数系统的分形图像编码方法却存在着计算量大的缺点,采用神经网络对分形图像进行压缩及解压缩目的在于解决压缩时间较长等问题。文中使用神经网络方法以并行方式完成对分形图像的压缩与解压缩。并通过实验,在实验中结合非线性网络和最速下降法实现对分形图像的压缩,在基本保证重建图像质量的前提下,减少了压缩时间,提高了压缩质量,进而说明神经网络技术应用于分形图像压缩中的可行性。In image compression technology,fractal image packing coding method has some characters such as irrelevance of decoding resolution ratio,fast encoding,high compression ratio and low rate of loss and so on.But the fractal image packing coding method based on iterative function system has some shortcomings,likes huge calculated amount.Neural networks is used in image compression and decompression,in order to resolve issues such as decompression time is too long.Gives the parallel method of neural network to finish the count of fractal image compression and decompression.And in the experiment,nonlinear nework and method of steepest descent are combined for fractal image compression.On the premise of quality assurance of the reconstruction image,times are shorten,qualities are improved.This shows the feasibility of neural network is used in fractal image compression and decompression.

关 键 词：分形图像 压缩 神经网络 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]