基于小波系数结构特征的高倍率图像压缩方法  

Image compression with high-ratio based on structural characteristics of wavelet coefficients

作　　者：汪国有[1] 邹光宇[1] 王文涛[1] 

机构地区：[1]华中科技大学图像识别与人工智能研究所,湖北武汉430074

出　　处：《华中科技大学学报（自然科学版）》2005年第12期67-69,共3页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家自然科学基金资助项目(60372066)

摘　　要：针对图像小波系数在各子带内显著性分布的聚簇特征,提出了在SPIHT算法的输出环节引入游程编码.对游程编码算法的性能进行分析后,基于算法的最优编码条件,对SPIHT算法中LIS的初始化操作进行了优化,提出了一种改进的多分辨系数扫描方法,使得显著系数在LIS及LIP中的一维分布更趋集中,从而使游程编码获得了更高的压缩效率.实验结果表明,与SPIHT相比,在几乎不影响压缩时间的同时,在高倍率压缩应用中提高了图像的重建质量近1 dB.After the clustering characteristics of image wavelet coefficients＇ significance distribution in each sub-band was analyzed, run-length coding （RLC） was introduced into SPIHT＇s output stage. On the basis of the optimal coding conditions, the initialization of LIS in SPIHT was optimized by RLC＇s performance analysis. An improved scanning scheme for multi-resolution coefficients was proposed to make significant coefficients＇ 1-D distribution in LIS and LIP tend to be clustered, achieving better compression efficiency of RLC. The experimental results showed that in comparison with SPIHT, the reconstruction quality of the images was nearly increased by 1dB in compression applications with high-ratio and the compressing time least affected.

关 键 词：图像压缩 小波 结构相关性 游程编码 SPIHT 

分 类 号：TN919.81[电子电信—通信与信息系统]