【作者】 杨波；

【导师】 管庆；

【作者基本信息】 电子科技大学 ， 电子与通信工程(专业学位)， 2015， 硕士

【摘要】 近年来,随着图像处理技术的发展,视频分析技术广泛应用于监控、物联网等行业。视频分析技术也提出了各方面的需求:高清化、智能化和网络化,而且高清化和智能化两种技术存在着越来越明显的依存关系。视频分析系统根据应用场景的不同,可分为前置分析和后置分析两种架构,各自拥有不同的应用领域,长期共存。就此,本文设计开发了一款基于TI DaVinci系列DSP芯片DM8127的前置分析高清视频分析平台。平台可使用图像传感器芯片获取进入镜头的高清图像。图像数据经过预处理以后,可用于后续分析,提取场景内的有用信息。同时,平台集成了网络视频传输功能,基于流媒体传输技术将视频画面呈现在用户IE浏览器中,实现网络交互。针对本课题设计视频分析平台,本文主要从系统硬件设计、开发环境搭建和软件设计三个方面依次进行了详细介绍。硬件设计方面主要包括:硬件平台的结构、选型和系统方案设计;硬件模块原理图设计、硬件调试及测试。软件方面主要包括:ARM端嵌入式Linux系统移植;TI McFW框架下图像传感器芯片驱动及图像捕获实现;TI McFW框架下人脸检测应用实现。开发环境方面主要介绍了平台操作系统、开发使用软件、交叉编译工具等内容。最后,本文分析了系统现有的不足之处,并根据产生这些不足的原因提出后续相应的整改与完善工作。更多还原

【Abstract】 In recent years, with the development of video processing technology, video analysis technology is widely used in surveillance, Internet of Things and other industries. Video analysis technology also puts forward various requirements, including high-definition, intelligence and networking.And the dependencies between high-definition technology and intelligenttechnology become more and more obvious. Video analysis technology has two forms based on different scenarios: front and rear. Both of them have separate applicability andcoexist for long time.In this regard, this paper has designed and developed a high definition front-facing video analysis platform based on TI’s DaVinci series DSP chip- DM8127. This platform can get high-definition digita image from the lens by CMOS sensor. The pretreated HD images can be used for subsequent analysis, and extracting useful information within the scene. Meanwhile, the platform has the capability for network video transmission which can presents the video to the user’s IE browser through streaming media technology. The interaction through the internet is realized by this way.This paper has made a detailed introduction from three aspects in turn: the design of hardware system, the build of development environment and the software design of system. The design of hardware system includes the structure of the hardware platform, chip selection, the design of system scheme, the design of schematic circuit diagram, debugging and testing of hardware platform. On the side of software design of system, the paper introduces the following aspects: the Linux system porting at the ARM side, the implementation of the driver of image sensor and Human face detectionapplications in TI’s McFW framework. On the bulid of the development environment side, the platform operating system, the software used in development, the cross-compiler and other content have been introduced. Finally, the paper analyzes the existing shortcomings of current system, and proposes subsequent work of corresponding rectification and improvement according to the causes of these deficiencies.更多还原