带噪图像的多阈值零树编码方法  

Multi-threshold zero tree coding for noisy image

作　　者：杨波[1] 汪同庆[1] 叶俊勇[1] 

机构地区：[1]重庆大学光电技术及系统教育部重点实验室,重庆400044

出　　处：《光电工程》2004年第3期60-63,共4页Opto-Electronic Engineering

基　　金：国家技术创新基金资助项目(01C26215110972)

摘　　要：零树编码利用一固定量化阈值对系数进行扫描,而多阈值零树编码将小波分解后各子带的最大系数作为该子带的扫描阈值,将扫描阈值大于量化阈值的子带作为重要子带,反之作为次要子带。只在重要子带中判断其是否存在重要系数。针对含有高斯白噪声的图像,引入混合降噪模型对各子带的扫描阈值进行修正,有效地提高了编码图像质量和编码效率。The coefficient is scanned with a fixed quantized threshold in zero tree coding while the maximal coefficient for various sub-bands is used as scanning threshold of the sub-band after the wavelet is decomposed by multi-threshold zero tree coding. The sub-band with scanning threshold larger than its quantized threshold is used as an important sub-band, otherwise as secondary sub-band. Scanning will be performed in the important sub-band to determine if it has significant coefficient. Mixed noise-reducing model will be introduced for correcting scanning thresholds of various sub-bands in accordance with the image with Gaussian white noise, thus effectively improving of coding image quality and coding efficiency.

关 键 词：图像编码 图像压缩 噪声图像 小波变换 多阈值 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]