最小二乘拟合与形态学膨胀的图像数率可分级编码算法  

A Rate Scalable Coding Algorithm of Image Based on Least Square Fitting Method and Morphological Dilation

作　　者：刘男[1] 王相海[1,2] 

机构地区：[1]辽宁师范大学计算机与信息技术学院,辽宁大连116029 [2]南京大学计算机软件新技术国家重点实验室,江苏南京210093

出　　处：《电脑与电信》2008年第12期35-37,共3页Computer & Telecommunication

基　　金：辽宁省自然基金项目;项目编号:20072156;辽宁省教育厅科学技术研究项目;项目编号:20060486;南京邮电学院图像处理与图像通信江苏省重点实验室开放基金;项目编号:ZK207008

摘　　要：本文提出一种基于最小二乘的线性拟合和形态学膨胀操作的小波域图像数率可分级编码算法,该算法利用图像小波变换子带中的零树结构系数间的对应关系,利用最小二乘线性拟合方法通过对父亲系数和间接后代系数的线性拟合确定直接后代的重要系数,同时通过对重要系数进行数学形态学的膨胀操作,形成重要系数的聚簇,从而降低重要系数位置的同步信息,来提高图像的编码效率。实验结果表明,在低码率下,解码后的图像较EZW和MRWD具有更好的峰值信噪比。A rate scalable coding of wavelet image based on least square fitting method and morphological dilation is proposed in this paper. In this algorithm, not only the relationship of zerotree structure of wavelet coefficients in subbands of image wavelet transform is used, but also the least square fitting method for parent coefficients and their indirect child coefficients is adopted to determine the positions of direct child coefficients between them. At the same time, we form the clusters of significant coefficients by using the morphological dilation. Then, we can reduce the in-phase information of the significant coefficients＇ positions and improve the coding efficiency of image. At last, the experiment results show that, in low rate, the proposed method for decoded images is superior to the EZW and MRWD in PSNR.

关 键 词：最小二乘拟合 形态学膨胀 提升方案小波 零树 图像编码 

分 类 号：TN919.81[电子电信—通信与信息系统] TG86[电子电信—信息与通信工程]