摘要
提出了一种无乘法整型可逆时域交叠变换方法,并以此为核心变换技术设计了一种新的有损到无损渐进的图像压缩系统.利用所提出的压缩系统从一个单一的码流文件中既可以恢复出完全无损的图像,也可以在高压缩比下得到高质量的有损重构图像.算法通过前后向滤波器改进离散余弦变换性能,并在矩阵分解基础上通过多阶提升实现完全可逆整型变换.光学遥感图像的实验结果显示,该算法在绝大多数情况下可以达到优于图像压缩国际标准JPEGJ、PEG2000以及新一代压缩算法HD-Photo的率失真性能以及高质量的主观视觉效果.
A reversible non-multiplying integer time domain lapped transform was introduced,and a new progressive lossy-to-lossless compression system was designed based on this transform technique.Completely lossless compression images and high-quality lossy reconstructed images could be obtained from a single code-stream using the proposed compression system.RTDLT improved the transform efficiency of DCT by adding pre-and-post-filters,and it couldrealize reversible integer transform by multi-lifting based on matrix factorization.Simulation results show that RTDLT-based compression system obtains comparable or even higher compression ratio in lossless compression than that of JPEG2000 and HD-Photo,better rate-distortion performance in lossy compression in most cases,and also gratify subjective visual quality.
出处
《光子学报》
EI
CAS
CSCD
北大核心
2010年第12期2251-2256,共6页
Acta Photonica Sinica
基金
国家自然科学基金(60607010
60201029)
教育部长江学者和创新团队支持计划(IRT0645)资助
关键词
离散余弦变换
时域交叠变换
图像压缩
矩阵分解
整型可逆变换
Dicrete Cosine Transform（DCT）
Time Domain Lapped Transform（TDLT）
Image compression
Matrix factorization
Reversible integer transform
