【作者】 邵晨；

【导师】 陈文； 王志杰；

【作者基本信息】 上海交通大学 ， Electronics and Communication Engineering， 2008， 硕士

【摘要】 数字图像压缩编码技术是各种数字图像应用的关键技术。在过去几十年之中,数字图像压缩编码技术取得了很大进步,由于小波变换在时频域的良好性能,小波变换编码成为当前编码的新的和重要的发展方向。随着数字图像压缩编码技术的广泛应用,人们对于压缩系统的要求越来越高,实时性作为一个重要的指标,正受到越来越多的关注。小波变换具有良好的局部特性和空间—频率特性,并且具有描述非平稳图像信号的能力和适应人眼视觉特性的良好性能,故在较高压缩比的图像编码领域中得到广泛研究和应用。基于小波变换的算法是国际标准JPEG 2000的核心技术之一。嵌入式零树小波编码(EZW)算法是小波图像编码研究的一个里程碑,对其研究具有较高的理论意义和实用价值。本文阐述了小波变换和图像编码的基础理论,简单介绍了多分辨率分析技术和Mallat算法,详细分析了小波变换的嵌入式零树图像编码算法的原理;在此基础上提出了一种改进的嵌入式小波零树图像编码方案,并通过仿真实验对比改进算法与原EZW算法,无论是在客观数据还是主观视觉上都有所改善,从而验证了该算法的有效性。更多还原

【Abstract】 Digital image compression is the key technology of different kinds of applications in digital image processing. In the past years, great processes have been made in digital image compression area. Digital image compression based on wavelet transform is becoming one of the most important trends of development in this field because of wavelet transform’s excellent performance both in time and frequency domain. With the development of technology, more and more algorithms emerge and their capability of real-time processing is attracting more focus than ever before.The wavelet transform have a better partial and space-frequency characteristics,an ability of describing the non-stationary image signal and a good function of adapting to the eye sense of visual characteristic. So it is extensively researched and applied in higher compression ratio of image coding field. The core of International Standard JPEG 2000 is based on the wavelet transform.EZW algorithm is the landmark in the research of wavelet image coding. It has been researched that EZW needs further improvement. In the thesis, after having introduced the theories about image coding and wavelet transform, then, the multiresolution analysis technology and the Mallat algorithm are introduced. And then the embedded wavelet zero-tree image coding (EZW) algorithm is discussed and analysed in detail. Based on the clear analysis, an improved algorithm is proposed in this thesis.The original and improved EZW algorithms simulated and their performances were compared. The improvement was proved to be efficient through the comparison between the data and the vision.更多还原