摘要
以大图像块或整个图像为处理单元的图像编码算法需要大量的内存来缓存图像,且编码过程中也会消耗大量内存,这种直接分块算法往往带来方块效应,影响图像的恢复质量。提出了以重叠块为单位的提升小波变换的方法,重叠分块可减小编码器对大块内存的需求,同时还可去除分块引入的方块效应。在变换中提出了多级并行分解方法,提高了分解效率。在对重叠块提升小波变换后的子带进行了统计分析,采用了DPCM与SPIHT相结合的方法。对直接分块、重叠分块、不分块算法进行了对比实验。结果表明,经重叠分块算法压缩的遥感图像具有较高的恢复质量。
For the most of remote sensing image compression methods,a large image block or the whole image is usually needs as a smallest processing unit,which surely needed large memory sizes.Although the method based on block is proposed,the blocking artefacts are introduced.A novel method of lifting wavelet transform based on block overlap is provided.Coding by the method requires small memory sizes and can get rid of blocking artefacts.A method that multi-level bands are decomposed parallel is proposed to improve the efficiency.After analyzing the wavelet coefficients,DPCM+SPIHT method is taken that is fit for block overlap.In experiment,the methods based on block,block overlap and whole image are taken individually,and the result of experiment show that the quality of the reconstructed image compressed by the method based on block overlap is best.
出处
《光学技术》
CAS
CSCD
北大核心
2006年第z1期468-470,共3页
Optical Technique
关键词
图像压缩
遥感图像
小波变换
重叠块
image compression
remote sensing image
wavelet transform
block overlap