【作者】 刘雄伟；

【导师】 成礼智；

【作者基本信息】 国防科学技术大学 ， 计算数学， 2005， 硕士

【摘要】 目前基于小波变换(DWT)的压缩方法是图像压缩的主流方法,但是小波变换作为一种全局的变换方法,存在着存储开销大、计算量高等特征,不能满足便携式设备以及星载压缩等环境下的应用要求。叠式变换(LT)兼有余弦变换(DCT)实现简单、存储开销小和DWT的优点,文中通过分析已有的时域重叠(TLT)和频域重叠(FLT)叠式变换,提出一种同时进行时域和频域重叠的叠式变换算法(TFLT),并结合上下文熵编码,讨论时频域重叠式变换算法。实验结果表明,本文方法与JPEG2000相比,在保证恢复图像质量相同的前提下,过程的复杂性与存储量都有所改善,是一种低存储,低复杂度的高效方法。由于多小波将光滑性、紧支性、对称性、正交性和更高阶的消失矩完美地结合在一起,它不但可以对信号提供一种更新的分析手段,而且对信号的逼近性质更好,重构信号在边界位置的性能也更完善。本文结合EBCOT的良好性能和多小波变换系数特性,提出一种结合滤波器和基于近似阶的预处理方法,并对多小波变换系数进行重新排列和量化处理,对EBCOT中位平面进行混合扫描方法相结合的压缩算法。实验结果表明该方法在一定程度上减少过程的复杂性,同时取得较好的压缩效果。更多还原

【Abstract】 Nowadays compression method based upon wavelet transform has become the mainstream in image compressing field. But as we all know, wavelet transform can’t satisfy the requirement of portable devices and Space-Based compression for its drawback in large storing cost and high calculating complexity. Fortunately, Lapped Transform(LT), a classical transform arithmetic, possesses the DCT’s feature of simple realization and low storing cost, while maintaining DWT’s advantage. By analyzing its Time Lapped Transform(TLT) and Frequency Lapped Transform (FLT), this article brings forward a arithmetic which carries the temporal laps and frequent laps at the same time and discusses this time-frequency lapped method(TFLT) combining with context-based entropy coding. Experimental results implicate that this arithmetic is a low storage-cost and complicated method, for its better performance over JPEG2000 in both realization and storage requirement.Owing to combination of smoothness, tight support, symmetry, orthogonality and higher order of vanish moments, Multi-wavelets supply us with a new signal analysis method , which can better approximate the signal and performance well on the edge in reconstruction. Combining the favorable capability of EBCOT and characteristic of Multi-wavelet’s coefficients, This article puts forward an integrated filter and pretreatment method based on approximated stairs. It processes the vector quantification of Multi-wavelet’s coefficients and carries out the mix-scan compression arithmetic towards bit planes of EBCOT. Experimental results show that this method reduces calculating complexity to some extent while obtaining favorable compressing effects.更多还原