一种基于边缘检测的小波包非对称分解在图像编码中的应用  

An Application Based on Edge Detection and Wavelet Package Asymmetry Composition in Image Coding

作　　者：陈志高[1] 叶吉祥[1] 

机构地区：[1]长沙理工大学计算机通信工程学院,长沙410076

出　　处：《世界科技研究与发展》2005年第2期53-57,共5页World Sci-Tech R&D

摘　　要：图像的像素点的值分布并不是均匀的,而是集中地分布在一些区间之内。小波包分解是将频带等分为2 j个新的子带,但很多子带里只有少数无用的孤立点;而在像素点集中的子带里,如果要进行更细的分解,将使子带数量呈指数级数增长。本文提出一种方法先对图像进行边缘检测,得到边缘与非边缘部分。对于像素点集中的几个部分,再分别对这几个部分定义不同的细度并采用小波包非对称分解再进行进一步细分,对于少数无关紧要的孤立点则舍弃。这样既保持了图像的有效信息,又能极大地减小图像编码的冗余度。The value of the point in image is not distributed evenly, but concentrated within a few ranges. The division of the wavelet package divides the frequency band evenly to 2j new subordinate bands; but there are only few useless and isolated points in them. While in the bands with concentrated points, if the more delicate division of the frequency band is to be made, the number of the subordinate bands will increase greatly. This paper proposes a new way to detect the edge of image andgets two parts of edge and non-edge. In the parts with concentrated points, the division these parts is in wavelet package asymmetry composition with different degree of execration again, and give up some isolated points. In this way, the useful information of the image can be saved and the redundancy of the image coding can be decreased to a large scale.

关 键 词：图像编码 对称分解 边缘检测 应用 小波包分解 像素点 指数级数 孤立点 子带 值分布 集中 冗余度 频带 

分 类 号：TN919.81[电子电信—通信与信息系统] TN911.6[电子电信—信息与通信工程]