【作者】 韩芳芳；

【导师】 郑德忠；

【作者基本信息】 燕山大学 ， 测试计量技术及仪器， 2004， 硕士

【摘要】 随着现代通信技术的飞速发展和通信业务的不断拓展，图像信息传输已经成为现代通信系统所提供的基本服务。但是具有庞大数据量的数字图像难以传输和存储，数字图像通信与通信网容量的矛盾日益突出，极大地制约了图像通信的发展，已经成为图像通信领域中的“瓶颈”问题。为了解决数字图像数据量巨大的问题，必需对图像数据进行有效地压缩。因此，数字图像特别是数字视频图像的压缩编码，已经成为现代信息社会中一个研究的热点问题。在傅立叶分析基础上发展起来的小波分析，提供了一种自适应的时域和频域同时局部化的分析方法，通过伸缩和平移等运算功能进行多尺度细化分析，能够有效地从信号特别是非平稳信号中提取信息。因此，小波分析成为数字图像处理及压缩编码的有力工具，也是近年来研究的热点。本文首先对数字图像的特性及其进行压缩的必要性、可行性进行了分析，并对传统压缩编码方法进行了比较和综述。针对小波信号分析的特点，对小波理论在数字视频图像压缩中的应用进行了研究，并对小波基的选择问题和小波变换域内的运动估计补偿问题做了探讨。本文还提出了一种基于数学形态算法的运动估计补偿方法，这是本文的一个创新点，它跳出了传统的以“块”为单位的思想，使运动估计补偿基于对象进行。最后，本文给出了基于高速通用DSP芯片的实时视频压缩系统的设计方案。其中的算法方案主要是采用帧间预测、小波变换、量化编码和游程编码，并通过编写C语言程序在PC机和DSP实验箱上进行了算法的仿真实验，对可能影响压缩效果的几个主要问题做了比较和讨论。压缩系统设计以美国TI公司的TMS320VC5416 DSP芯片为主，它的高运算速度、特殊指令字以及代码流水线处理能力使得它能够很好地执行实时视频压缩。本文分析了TMS320VC5416的结构特点，对算法方案在DSP上的程序实现所应该考虑的几个问题做了探讨。在文章最后，对压缩系统的其它功能模块进行了芯片选型和设计，并给出了压缩系统构成的总框图。更多还原

【Abstract】 With the development of communication technology and communication business, image information transmission has become the base service of modern communication system. But the conflict between image transmission and communication network capacity is becoming more and more serious, especially for the digital images storage and transmission. The huge data for digital images is a knotty problem for image information communication. Therefore, the effective compression coding methods for digital images are necessary, and it has become a hot problem for the modern society to study. The wavelet transform, which evolved based on Fourier analysis, offers a self-adaptive localization method for a given signal in both time domain and frequency domain. Through extension and translation options and multi-scale analyzing, it can extract helpful signal information effectively. So wavelet analysis has become a powerful tool for digital image processing and compression. This paper analyzes the features of digital images and the necessary and feasibility for images compression at first, and then makes a comparison and summarize for the traditional compression methods. This paper studies the application of wavelet theory in digital video compression, and makes some discussions on the wavelet bases selection and motion estimation and compensation in wavelet frequency domain. This paper proposes a method for motion estimation and compensation by dilation and erosion algorithm, which is an innovation in this paper. This method is an object-oriented operation. It can avoid the block artifacts caused by traditional block-based methods. At last, this paper brings forward a design scheme for a real-time video compression system based on high-speed universal DSP. The arithmetic project is mainly including frames prediction, wavelet transform, quantization coding and run-length coding. This arithmetic project is experimented on personal computer by compiling C language program, and some problems that affect compression have also been compared and discussed. The video compression system is based on TMS320VC5416 DSP, which produced by American TI Company. Its features of high operation speed, special <WP=6>instruction word, and pipeline structure ensure it can execute real-time video compression well. This paper analyzes TMS320VC5416 DSP features and discusses some problems for the arithmetic program implementing on it. This paper also selects chips and design structure for the other function module of the compression system. At last, it gives the block diagram for the compression system.更多还原