基于Zynq的运动车辆检测与跟踪系统设计

发布时间：2018-05-11 15:39

本文选题：Zynq + 软硬件协同设计　；参考：《大连海事大学》2017年硕士论文

【摘要】：智能交通系统(Intelligent Transportation System,ITS)是目前交通运输方向正在研究和普遍关注的课题,其中运动车辆检测与跟踪系统是ITS的重要研究内容。因此,研究体积小、功耗低、速度快的运动车辆检测与跟踪系统具有重要的意义。本系统采用的是Xilinx公司Zynq-7000系列全面可编程片上系统平台,片内集成了双核ARM Cortex-A9处理器和Xilinx 7系列FPGA,采用软硬件协同设计方法,将图像采集、车辆检测与跟踪、显示集合到一个嵌入式系统中。车辆检测方面,利用Xilinx高级开发工具Vivado HLS(High-Level Synthesis)完成图像预处理算法,并利用此工具把改进的三帧差分法的运动目标检测算法成功移植到FPGA中。车辆跟踪算法方面,对运动车辆求取最小外接矩、质心、速度作为目标特征,建立Kalman滤波器运动模型,预测被跟踪车辆在下一帧中可能出现的位置,并确定搜索范围,再结合基于欧式距离的匹配算法,完成运动车辆的跟踪。本设计完成的主要工作包括:(1)用高层次综合工具Vivado HLS设计了改进的三帧差分法的运动目标检测算法的硬件加速IP核,包括直方图均衡化、高斯滤波、噪声去除、Sobel边缘检测、形态学处理以及帧差图像预处理算法。(2)利用Vivado工具,以图形界面化的方式,进行硬件工程设计,主要包括Zynq处理器、AXI互联、VDMA、HDMI以及硬件加速IP核等模块。(3)在ZedBoard平台上搭建嵌入式Linux系统,主要包括Linux系统移植、交叉编译环境的搭建以及FFmpeg+OpenCV库和Qt库的移植等。(4)在ARM部分实现系统软件设计,主要包括VDMA IP核和硬件加速IP核驱动程序设计,Kalman滤波器跟踪算法的实现,以及建立Qt软件执行操作界面的设计等。本系统设计采用软硬件协同设计的方法完成了基于Zynq的运动车辆检测与跟踪,具有实时性好、跟踪效果好和界面友好等特点,达到了很好的预期效果。
[Abstract]:Intelligent Transportation system (ITS) is an important research topic in the field of transportation, in which moving vehicle detection and tracking system is an important research content of ITS. Therefore, it is of great significance to study the moving vehicle detection and tracking system with small size, low power consumption and high speed. The system adopts Xilinx Zynq-7000 series comprehensive programmable system platform, which integrates dual-core ARM Cortex-A9 processor and Xilinx 7 series FPGA. it uses hardware and software co-design method to collect image, detect and track the vehicle. Display is assembled into an embedded system. In the aspect of vehicle detection, the image preprocessing algorithm is completed by Xilinx advanced development tool Vivado HLS(High-Level synthesis, and the improved three-frame differential algorithm of moving target detection is successfully transplanted into FPGA. In the aspect of vehicle tracking algorithm, the minimum external moment, centroid and velocity of moving vehicle are taken as the target features, the motion model of Kalman filter is established, the possible position of the tracked vehicle in the next frame is predicted, and the search range is determined. Combined with the Euclidean distance matching algorithm, the tracking of moving vehicles is completed. The main work of this design includes: (1) using high-level synthesis tool Vivado HLS to design the hardware accelerated IP core of the improved three-frame differential algorithm for moving target detection, including histogram equalization, Gao Si filtering, noise removal and edge detection. Morphological processing and frame difference image preprocessing algorithm. 2) the hardware engineering design is carried out by using Vivado tool and graphical interface. The embedded Linux system is built on the ZedBoard platform, including the Zynq processor, the VDMA-HDMI and the hardware accelerated IP core. It mainly includes the transplantation of the Linux system. The cross-compiling environment and the transplantation of FFmpeg OpenCV library and QT library are used to realize the software design of the system in ARM, including the design of VDMA IP core and hardware accelerated IP core driver, and the realization of Kalman filter tracking algorithm. And QT software implementation interface design. The system adopts the method of hardware and software co-design to complete the detection and tracking of moving vehicles based on Zynq, which has the characteristics of good real-time, good tracking effect and friendly interface, and achieves a good expected effect.
【学位授予单位】：大连海事大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U495;TP391.41

【相似文献】