基于DSP和FPGA的智能相机关键技术研究
文献类型：学位论文
| 作者 | 张琦 |
| 学位类别 | 工学硕士 |
| 答辩日期 | 2013-05-23 |
| 授予单位 | 中国科学院大学 |
| 授予地点 | 中国科学院自动化研究所 |
| 导师 | 徐德 |
| 关键词 | 机器视觉 智能相机 FPGA DSP 图像传感器 硬件设计 Machine vision Smart camera FPGA DSP Image sensor Hardware design |
| 其他题名 | The Key Technology of Smart Camera Based on DSP and FPGA |
| 学位专业 | 控制理论与控制工程 |
| 中文摘要 | 随着计算机视觉、图像处理技术的发展进步，利用机器视觉代替人工完成工业生产线上产品的检测极大的提高了生产效率。智能相机作为机器视觉的一个重要发展方向，有着比传统的PC式机器视觉系统体积小、功耗低、实时性好、自动化程度高等优势。智能相机在工业检测、视频监控、智能交通、移动机器人等领域有着非常广阔的应用前景。因此，设计智能相机系统有着重要的研究意义和实用价值。本文是在“863” 项目的子项“快速视觉单元研制”的支持下，以智能相机在乒乓球机器人的应用为背景，研制了一种基于DSP和FPGA的智能相机系统。重点针对智能相机的硬件系统设计和软件算法实现进行了深入的研究。 本文完成的主要工作如下： (1)分析了智能相机实现的理论基础和软硬件技术条件。详细论述了常用的图像传感器原理、结构，图像存储单元、图像处理单元的硬件设计技术。 (2)成功构建了一个基于DSP和FPGA的嵌入式视觉单元的硬件平台。在充分考虑了DSP和FPGA在图像处理所各自具有的特点的基础上，提出了基于DSP和FPGA的智能相机设计方案。其中DSP为核心处理器，采用的是TMS320DM642芯片，FPGA为协处理器，采用的是EP3C16U484C6N芯片，它们协同完成图像的采集、处理与传输工作。 (3)在协处理器FPGA上实现图像的采集与预处理工作。在FPGA内部搭建了系统框架，实现了SCCB协议的读写，完成了OV5620的图像采集，并在外部存储器上实现了乒乓操作，设计了总线占用轮询机制，实现了多模块对外部存储器的优先级访问。 (4)在核心处理器DSP上对初始化配置进行了优化，完成了DSP对FPGA的异步数据读取，实现了从RAWRGB到RGB转换的图像处理算法，与上位了建立了基于UDP的网络通讯，实现了智能相机的远程操作。 (5)在上位机上对采集到的图像实现了目标识别算法，成功识别出乒乓球并计算出了乒乓球的中心坐标。 |
| 英文摘要 | With the development of computer vision and image processing technology, machine vision can improve production efficiency greatly instead of human beings in detection of industrial production line. Smart camera is one of the most important developing directions of machine vision. It has advantages of small size, low power consumption, real-time, high automatic degree over traditional PC-based machine vision system. Smart Camera has a very bright application prospects in the field of industrial inspection, video surveillance, intelligent transportation, mobile robot. Therefore，design of smart camera has a great significance in modern production. In this paper, supported by National 863 Projects, based on the application of ping-pang robot system, a smart camera system based on DSP and FPGA is designed. The hardware structure and software realization are the key point. They are researched in depth. The main work of this thesis includes the following issues: Firstly, the theoretical basis, hardware and software technology of smart camera are analyzed. The principle and structure of common image sensor, the image storage module, and image processor are described in detail. Secondly, the hardware platform of embedded vision system is built based on DSP and FPGA. Taking full consideration of the different features in image processing of DSP and FPGA, we raise a solution of smart camera based on DSP and FPGA. DSP is the core processor, using TMS320DM642. FPGA is the coprocessor using EP3C16U484C6N. They work collaborative to complete the acquisition, processing and transmission of image data. Thirdly, FPGA as coprocessor is in charge of image capturing and co-processing. The architecture of internal system framework of FPGA and the SCCB agreement help to complete image acquisition of OV5620 and the store of ping-pong; designing the bus occupancy polling mechanism gives a multi-module prior access to external memory. Fourthly, the initial configuration of the DSP is optimized. We have achieved the function for DSP to read data from FPGA asynchronously and the image processing algorithm for the image format of RAWRGB to RGB conversion; a UDP-based network communication is built and then parts of the upper computer software are developed, achieving remote operation of the smart camera. Finally, the image processing algorithms are achieved on computer. We calculate the ping-pang on computer successfully, as well as the central coordinate of the ping-pang. |
| 语种 | 中文 |
| 其他标识符 | 201028014628022 |
| 源URL | [http://ir.ia.ac.cn/handle/173211/7693]  |
| 专题 | 毕业生_硕士学位论文 |
| 推荐引用方式 GB/T 7714 | 张琦. 基于DSP和FPGA的智能相机关键技术研究[D]. 中国科学院自动化研究所. 中国科学院大学. 2013. |
入库方式： OAI收割
来源：自动化研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。