基于DSP的视频跟踪器的设计与研究

发布时间：2018-09-05 12:57

【摘要】：随着社会经济的发展和人民生活水平的提高,公共安全成为社会普遍关注的焦点。视频跟踪系统的主要功能就是对视频图像进行处理,并完成运动目标的识别和跟踪。视频图像处理主要可以分为视频采集、视频预处理、视频叠加以及视频目标跟踪四个部分。跟踪是建立在识别的基础上的,由于目标不尽相同,因此采用的跟踪手段也有很大差别,其中比较典型和常用的有相关跟踪,全视场跟踪以及边缘跟踪等。由于视频跟踪处理的信号为实时视频数字图像信号,因此它具有两个主要特点:一是海量数据,二是需要有很快的处理速度。随着数字信号处理器——DSP芯片集成度和运算速度的不断提高,图像的实时性传输和高速处理要求从技术上能够得以保证,这也促使DSP芯片被广泛应用于视频跟踪系统。本文针对视频跟踪系统海量数据、高处理速度的特点以及实时性跟踪的要求,同时结合已有技术和当前先进器件,完成基于DSP的视频跟踪系统的设计。系统选用TI公司的TMS320C6416作为核心器件实现了对视场中目标图像的识别和跟踪,并实时给出目标位置信息;选用ALTERA公司的EP1C6Q240C8作为辅助处理器,负责视频图像的采集、预处理以及视频叠加显示,以及DSP与FPGA之间的时序逻辑控制。论文提出了一种基于硬件的快速中值滤波算法,解决了传统中值滤波算法实时性差和资源占用率高的缺点,很好的满足了系统实时性的需求;利用FPGA内部M4k RAM资源构成高速FIFO,完成了DSP和FPGA之间的速率匹配,并利用系统DSP的EDMA传输机制,实现了视频数据从FPGA到DSP的缓冲加载。系统的调试结果表明,该系统具有一定的实时性和稳定性。
[Abstract]:With the development of social economy and the improvement of people's living standard, public safety has become the focus of the society. The main function of the video tracking system is to process the video image and to realize the recognition and tracking of moving objects. Video image processing can be divided into four parts: video capture, video preprocessing, video superposition and video target tracking. Tracking is based on recognition, because the target is different, so the tracking methods are very different, such as correlation tracking, field of view tracking, edge tracking and so on. Because the signal of video tracking processing is real time video digital image signal, it has two main characteristics: one is massive data, the other is that it needs fast processing speed. With the improvement of integration and operation speed of digital signal processor (DSP), real-time image transmission and high speed processing need to be guaranteed technically, which makes DSP chip widely used in video tracking system. Aiming at the characteristics of mass data, high processing speed and real-time tracking of video tracking system, this paper completes the design of video tracking system based on DSP combined with the existing technology and current advanced devices. The system chooses TMS320C6416 of TI Company as the core device to realize the recognition and tracking of the target image in the field of view, and gives the target position information in real time, and selects the EP1C6Q240C8 of the ALTERA Company as the auxiliary processor, which is responsible for the collection of the video image. Preprocessing and video overlay display, and timing logic control between DSP and FPGA. This paper proposes a fast median filter algorithm based on hardware, which solves the shortcomings of traditional median filtering algorithm, such as poor real-time performance and high resource occupancy rate, and meets the real-time needs of the system. Using M4k RAM resource in FPGA to construct high speed FIFO, the rate matching between DSP and FPGA is completed, and the buffer loading of video data from FPGA to DSP is realized by using EDMA transmission mechanism of system DSP. The debugging results show that the system has a certain real-time and stability.
【学位授予单位】：兰州交通大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP391.41

【参考文献】