一种基于多项式的图像压缩方法  

A METHOD OF PICTURE COMPRESSION BASED ON MULTINOMIAL

作　　者：孙宏志[1] 刘书生[2] 

机构地区：[1]辽宁省地震局,辽宁沈阳110031 [2]东北大学,辽宁沈阳110004

出　　处：《东北地震研究》2003年第2期54-61,共8页Seismological Research of Northeast China

摘　　要：介绍了数字图像压缩的一种新的思维方法,即在将给定的图像块根据纹理特征分为平滑块、斜面块、曲面块、边缘块后,在基于不同多项式对不同类图像块分别进行处理的过程中,对于包含大量高频信息的边缘块,通过提取边缘轮廓位置信息,减少图像块内的高频信息量,而后再对被边缘轮廓分割的新图像块进行分类、处理:由于把图像边缘轮廓的大部分高频信息分解于边沿信息中,因此提高压缩比时不会丢失太多高频细节,减少了物体的边界噪声和可见的图块边界,从而在相同压缩比的前提下,更加保证了图像的视觉效果。文章中给出了此种方法的理论公式、推导过程以及整个实现过程,并通过实例比较了此方法与二维离散余弦变换在图像压缩中的性能差异。This article introduces a new method of digital image compression,which means that after we have sorted the given image blocks into smooth one,inclined one,curved one and marginal one,in the course of dealing with the different image blocks separately based on the different multinomial,for the marginal one that comprises a great deal of high frequency information,by extracting the positional information of marginal figure,we can reduce the high frequency information in the block,and then assort and manage the new image block whose marginal figure has been intersected.Because the image high frequency information of the marginal figure has been mostly decomposed into the marginal information,most high frequency details will not be lost with the increasing of compression ratio,then the object's marginal noise and the visible fringe of the image block is cut down,which can ensure more visual effect with the premise of the same compression ratio.This article introduces the theoretic formula,deducing process and the whole course of realization of this method.It also emphatically discusses the difference between this method and the Discrete Cosine Transform(DCT)

关 键 词：图像压缩 多项式 像素 图像块 解码 

分 类 号：P315.63[天文地球—地震学]