基于FPGA远程网络监视系统

【作者】 郭赞；

【导师】 李玉惠；

【作者基本信息】 昆明理工大学 ， 检测技术及自动化装置， 2009， 硕士

【副题名】视频采集、处理及网络传输

【摘要】 近年来随着多媒体技术的迅速发展和计算机性能的不断提高,数字视频监视系统已逐渐取代传统的模拟式监视系统而广泛应用于银行、小区、宾馆等公共场所,并在公共安全领域发挥着重要的作用。在未来的发展中,智能化监视系统将有广阔的应用前景。本论文研究开发了一种基于运动目标识别技术的数字视频远传网络监视系统。本系统采用QuartusⅡ和NiosⅡ嵌入式开发软件,将图像处理、网络传输、SOPC(System on a Programmable Chip可编程片上系统)技术紧密连接在一起,并采用目标提取算法减少视频大数据量传输,提高视频传输的实时性和显示质量,为远程视频监视提供了一种新的解决方法。本设计通过摄像头采集图像、网络传输背景及运动目标,对现场有无人员入侵进行判断,实现无人值守的视频监视。论文探讨了一种基于SOPC技术的视频监视系统的设计方案——CPU (Central Processing Unit中央处理器)结合用户自定义逻辑设计的方案。此方案的设计思想是以下载到FPGA(Field Programmable Gate Array片上可编程逻辑门阵列)的NiosⅡ5.0嵌入式软核CPU为系统控制模块,实现数据流的接收、处理、控制和UDP (User Datagram Protocol用户数据报协议)协议网络传输,并运用FPGA逻辑单元实现视频图像采集和处理。本方案中视频采集和处理程序采用Verilog语言来进行设计,在很大程度上提高了系统速度。在软硬件协同设计方面,采用摄像头自定义组件模块挂接Avalon总线的方法,通过DMA (Direct Memory Access直接存储器访问)将硬件视频数据传输至SDRAM (Synchronous Dynamic RAM同步动态存储器)。在图像处理实现方面,采用相邻两帧图片进行差分的算法实现运动目标的提取,在网络传输视频数据时对背景和运动目标分别进行传输,降低了数据传输量,提高了传输速度。本监视系统核心技术是视频采集、转换、运动目标检测、提取及通信技术。本文主要研究内容包括视频采集、颜色模式转换、分辨率转换、图像数据的传输和运动目标检测等问题。在硬件系统集成过程中,系统采用专业的130万像素摄像头,并配以自定义组件模块实现其视频采集、预处理功能,并在SOPC Builder软件中配置硬件集成系统。在软件编程中采用C语言,编写了运动目标提取、网络有线传输等相关函数,并在此基础上构建软件系统。最后本文通过时序仿真判断硬件连接的正确性和用EtherPeek软件捕获UDP协议网络数据包来判断网络连接、视频传输和C语言编写函数的正确性。更多还原

【Abstract】 In recent years, with the rapid development of multimedia technology and the increasing enhancement of computer performance, the digital video surveillance system has gradually substituted for traditional analog video surveillance system, which is widely applied in public places, such as banks, intelligent communities and hotels. It plays an important role in the public security field. Therefore, the intelligent video surveillance system will own a bright application prospect in the future.A digital video remote network surveillance system based on the concept of moving object recognition was designed in this paper. This system adopted the embedded softwares QuartusII and NiosⅡ, and integrated video processing, network transmitting and SOPC embedded technology. Meanwhile, it applied the algorithm of object distillation to reduce a great number of transmitting data and improve the performance of real time and display quality, which presented a new solution available to the system. By video capturing, background transmitting and moving objects transmitting via network, the design can judge whether invading objects exist and realize surveillance system without attendant. A video surveillance system design plan based on SOPC technology,——the plan of CPU integrated with user’s self-defining logic, was discussed in the thesis. CPU NiosⅡ5.0 embedded IP core, which has been downloaded to FPGA, acts as the control module of the system. It can receive、process、control data stream, and network transmitting based on UDP protocol. FPGA’s logic units are used to implement the video image manipulation and processing. To a large extent, this plan has improved the speed, of the system, because the system’s hardware program was written in Verilog HDL language, in order to capture and process video data. In the way of collaborating design of software and hardware, the system adopt the solutions that video user’s self-defining module linked with Avalon bus and that DMA module transferred hardware video data to SDRAM memory. In the way of realizing image manipulation, the system realized moving object recognition through adopting the difference formula of two neighboring pictures. When the system transmits video data through network, the background pictures and moving object pictures are transmitted separately. So the solution reduces the great number of transmitting data and improves transmitting rate and display quality.The key technology of video surveillance system is video capture, video transmitting, moving-target detection, recognition and communication. The paper has realized a type of remote network surveillance system based on FPGA, and has solved several problems, including video capture, color mode transition, resolution change, video data transition and moving-target detection. In the hardware system integration part, this system adopted specialized mega-pixel camera、configured user logic component module、realized the function of gathering、processing video, and built hardware integrated system based on SOPC, as well as compiled NiosⅡC/C++ software system through the function of moving-target recognition and network communication. Finial, the paper confirmed the correctness of hardware communication through timing simulator and the correctness of network communication and C program.更多还原