基于Zynq-7000高清视频采集处理系统Linux移植与应用程序设计
发布时间：2019-06-12 18:45
【摘要】：随着现代制造业技术的不断进步,机器视觉应用越来越广泛,同时,工业应用对视频分辨率、图像处理速度和处理精度要求不断提高。后PC时代,嵌入式机器视觉成为了机器视觉领域的研究热点,它集视频图像采集、处理、分析、传输等功能于一体,并且具有高性能、低功耗、小型化等特点,其在许多行业都有着广泛的应用前景。本论文重点研究了嵌入式机器视觉的核心——高清视频采集处理系统,探讨了基于Zynq-7000平台的高清视频采集处理系统Linux移植、视频处理IP核驱动设计、硬件加速字符识别处理等关键技术。首先深入分析Zynq-7000平台“FPGA+ARM”的基础特性,描述了高清视频采集处理的整体框架,实现了高清CMOS图像采集、FPGA上图像预处理、以及视频图像的缓存。然后在ARM上移植嵌入式系统,移植包括U-Boot移植、Linux内核移植、Ramdisk文件系统移植、OpenCV函数库移植和设备树设计。同时在深入理解Linux驱动程序与硬件IP核的关系、V4L2设备驱动框架及API使用的基础上设计了视频IP核的驱动程序。在应用层设计了基于V4L2框架的视频流采集传输控制程序,实现了字符识别软硬件协同处理。通过对高清视频采集处理系统进行实际测试,该系统采集高清视频图像流畅、清晰,能实现字符识别的实时处理,处理结果叠加在视频图像上并通过HDMI屏幕显示出来。此系统采用“FPGA+ARM”的软硬件协同方式构建高清视频采集处理系统运算速度快、图像识别算法可扩展集成度高,具有一定的工程应用价值。
[Abstract]:With the continuous progress of modern manufacturing technology, machine vision is more and more widely used. At the same time, industrial applications require more and more video resolution, image processing speed and processing accuracy. In the post-PC era, embedded machine vision has become a research focus in the field of machine vision. It integrates video image acquisition, processing, analysis, transmission and other functions, and has the characteristics of high performance, low power consumption, miniaturization and so on. It has a wide range of application prospects in many industries. In this paper, the core of embedded machine vision, HD video capture and processing system, is studied, and the key technologies such as Linux transplantation of HD video capture and processing system based on Zynq-7000 platform, IP core driver design of video processing, hardware accelerated character recognition and processing are discussed. Firstly, the basic characteristics of Zynq-7000 platform "FPGA ARM" are deeply analyzed, the overall framework of HD video capture and processing is described, and the high definition CMOS image acquisition, image preprocessing on FPGA and video image cache are realized. Then the embedded system is ported on ARM, including U-Boot migration, Linux kernel migration, Ramdisk file system migration, OpenCV function library migration and device tree design. At the same time, on the basis of deeply understanding the relationship between Linux driver and hardware IP core, V4L2 device driver framework and the use of API, the driver of video IP core is designed. A video stream acquisition and transmission control program based on V4L2 framework is designed in the application layer, and the hardware and software co-processing of character recognition is realized. Through the actual test of the high-definition video acquisition and processing system, the high-definition video image collected by the system is smooth and clear, and can realize the real-time processing of character recognition. the processing results are superimposed on the video image and displayed on the HDMI screen. This system adopts the hardware and software co-mode of "FPGA ARM" to construct HD video capture and processing system with fast operation speed and high extensible integration of image recognition algorithm, so it has certain engineering application value.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41
[Abstract]:With the continuous progress of modern manufacturing technology, machine vision is more and more widely used. At the same time, industrial applications require more and more video resolution, image processing speed and processing accuracy. In the post-PC era, embedded machine vision has become a research focus in the field of machine vision. It integrates video image acquisition, processing, analysis, transmission and other functions, and has the characteristics of high performance, low power consumption, miniaturization and so on. It has a wide range of application prospects in many industries. In this paper, the core of embedded machine vision, HD video capture and processing system, is studied, and the key technologies such as Linux transplantation of HD video capture and processing system based on Zynq-7000 platform, IP core driver design of video processing, hardware accelerated character recognition and processing are discussed. Firstly, the basic characteristics of Zynq-7000 platform "FPGA ARM" are deeply analyzed, the overall framework of HD video capture and processing is described, and the high definition CMOS image acquisition, image preprocessing on FPGA and video image cache are realized. Then the embedded system is ported on ARM, including U-Boot migration, Linux kernel migration, Ramdisk file system migration, OpenCV function library migration and device tree design. At the same time, on the basis of deeply understanding the relationship between Linux driver and hardware IP core, V4L2 device driver framework and the use of API, the driver of video IP core is designed. A video stream acquisition and transmission control program based on V4L2 framework is designed in the application layer, and the hardware and software co-processing of character recognition is realized. Through the actual test of the high-definition video acquisition and processing system, the high-definition video image collected by the system is smooth and clear, and can realize the real-time processing of character recognition. the processing results are superimposed on the video image and displayed on the HDMI screen. This system adopts the hardware and software co-mode of "FPGA ARM" to construct HD video capture and processing system with fast operation speed and high extensible integration of image recognition algorithm, so it has certain engineering application value.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41
【参考文献】
相关期刊论文 前5条
1 孟繁星;;基于ZYNQ-7000的视频图像处理系统设计[J];信息通信;2015年08期
2 叶h，
本文编号：2498223
本文链接：https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2498223.html
