云计算-面向多屏共享系统的H.264AVC运动估计算法改进.pdf

I

面向多屏共享系统的 H.264/AVC 运动估计算法改进

摘 要

随着科技的发展，智能手机、平板电脑、智能电视等越来越多的智能化家电

走进了千家万户。数字家庭一种以计算机技术和网络技术为基础，各种家电通过

各种互连方式进行通信及数据交换，从而实现家用电器之间的“互联互通”的技

术,它可以使人们足不出户就可以更加方便快捷地获取信息，从而极大程度地提

高人类居住的舒适性和娱乐性。多屏共享系统是在数字家庭的基础上发展出来的

缩比是 MPEG-2 的 2 倍以上，是 MPEG-4 和 H.263 的 1.5 到 2 倍，因此用在多屏

共享系统上具有节省带宽的优势。然而标准编码器同时还具有高运算复杂度的特

点，这使得用户在实际使用时硬件的运算负荷大，观看视频出现较大延时甚至卡

顿等现象，用户体验不理想。

在 H.264/AVC 编码器中，运动估计是耗时最高的环节。X264 是一种 H.264/AVC

的开源实现，因为其运行时简单高效的特性而被广泛的应用于各个领域。X264

中一共提供了 4 种运动估计算法，分别是变换全局搜索算法（TESA）、菱形搜索

算法（DS）、六边形搜索算法（HEX）和非对称十字形多层次六边形格点搜索算法

（UMH）。

在本文中，我们深入研究了 X264 提供的 4 种算法，然后对其中最适合多屏

II

An Improvement on Motion Estimation Algorithm of

H.264/AVC Encoder for Multi-screen Sharing System

With the development of technology, a variety of smart home appliances constantly

entered millions of households, and the concept of digital home has been widely

accepted by consumers. The digital home is based on computer and network

technology, it allows different kinds of smart appliances to communicate and

exchange data, and makes them connecting with each other. Using the digital home

technology people can obtain information more conveniently and quickly without

H.264/AVC has characteristics such like low bit-rate, high-quality images,

fault-tolerant capability and strong network adaptability. Currently, the multi-screen

sharing systems on home appliances in the market mostly are based on the standard

H.264/AVC encoder. Although the standard H.264/AVC encoder has a high data

compression ratio, using it on the multi-screen sharing system can save the bandwidth

of the network transmission, but the higher computational complexity of H.264/AVC

encoder contributed a lot to the delay time of multi-screen sharing system, users

always obtains poor experience.

Motion estimation is the most important and time-consuming part of H.264/AVC

encoder. As the most widely used open source implementation of H.264/AVC, X264

comparing the existing algorithms. Moreover, to test the effect of improved algorithm

on delay time of system, a multi-screen sharing system simulation platform has been

designed and implemented. Experimental results show that the improved method has

increased the encoding speed significantly while hardly affecting on picture quality,

and the simulation platform which using improved method has shorter delay time.

Key Words: Multi-screen sharing system; H.264/AVC; Motion estimation; Delay

time

目 录

1 绪论 ............................................................ 1

1.1 项目背景 ........................................................ 1

1.2 国内外研究现状 .................................................. 1

1.3 论文工作安排 .................................................... 3

2 H.264/AVC 编码器研究 .......................................... 5

3.1 运动估计算法原理 ............................................... 13

3.2 块匹配准则 ..................................................... 14

3.3 菱形搜索算法 ................................................... 15

3.4 六边形搜索算法 ................................................. 17

3.5 非对称十字形多层次六边形格点搜索算法 ........................... 19

3.6 六边形搜索算法的改进 ........................................... 21

3.7 本章小结 ....................................................... 23

4 实验平台设计及关键模块实现 .................................. 24

4.1 平台总体设计 ................................................... 24

4.2 软件开发环境搭建 ............................................... 25

V

4.4.1 FFmpeg 重要数据结构简介 ...................................... 30

4.4.2 编码器模块的实现............................................. 33

4.5 网络传输模块的原理和实现 ....................................... 35

4.5.1 RTP 协议简介 ................................................. 36

4.5.2 网络传输模块的实现 ........................................... 38

4.6 本章小结 ....................................................... 41

5 实验方法设计与结果分析 ...................................... 42

5.1 编码器性能的测试 ............................................... 42

5.1.1 实验方法设计 ................................................. 42

5.1.2 实验结果与分析 ............................................... 45

5.2 延迟时间的测试 ................................................. 46

6.1 总结 ........................................................... 51

6.2 展望 ........................................................... 51

个人简历 ......................................................... 57

发表的学术论文 .................................................. 58