嵌入式智能交通车流量监控系统的实现

发布时间：2019-06-20 04:09

【摘要】：道路智能车流量检测技术是智能交通系统的关键技术之一,其可以检测高速公路、一般公路、城市道路上的过往机动车辆的车型、号牌、行驶速度、交通量、道路占有率等交通参数,根据这些参数来判断道路的交通状况,作出决策并优化交通中的薄弱环节。目前交通信号灯的切换由人工一次设定后固定不变,而实际的反应交通状况的各参数均为动态量值,不能按实际车流量的变化控制红绿灯的变换,这必然会照成交通资源利用率的低下。为了避免这种情况,减少车辆的停等时间并提高车辆通行率,就要根据交通流量动态地控制红灯和绿灯的切换时间,再者,智能交通系统是一个综合性很高的系统,需要将每个路口的车流状况适时的返回到交通控制中心,通过中心的综合分析,为车辆的导航提供准确的数据保证。针对这种在实际应用中存在的问题,需要研究一种依据机器视觉的智能型交通信号控制系统。该系统是一种嵌入式机器视觉应用装置,能通过对实时采集的视频图像进行处理和分析后获取道路车流量数据,然后智能地控制交通信号装置,同时为系统的高级应用提供数据源。另外,本次研究的另一目标是实现一个软硬件系统低耦合度、模块化的机器视觉应用的嵌入式平台。将机器视觉算法与硬件的耦合度降到最低,这样做的好处是:第一,可以最大化利用现有开发资源,减低开发的难度和复杂度加快产品的成形;第二,由于丰富了可利用软硬件资源,降低了开发难度和复杂度,必然可以缩短产品的研发周期降低开发成本;第三,系统软硬件采用模块化开发,可根据实际应用需求灵活地进行集成,增强了系统的可扩展性;第四,采用本次研究的方式可以不用修改机器视觉算法就可以快速移植到手机等移动终端上,可以开辟更多的应用领域和市场。本次研究有以下创新点:(1)探索了一种和现有机器视觉算法在嵌入式应用不同的实现思路和方法,采用软硬件模块化的思想实现研究目标;(2)使用最新型ARM处理器,很好的实现了机器视觉图形图像处理的应用,这样的实现方法具有很好的灵活性、移植性和可扩展性,研发成本和生产成本较低,可以很好的应用于民用系统,尤其是移动终端领域;(3)成功实现了著名的自适应通信环境库ACE、图形图像算法库OpenCV的嵌入式系统移植和应用,使得以模块化、软硬件系统低耦合度思想研发机器视觉应用成为可能;(4)成功构建了一套嵌入式机器视觉应用的嵌入式系统平台,可以非常便利的移植其他机器视觉算法模块,快速的形成嵌入式的机器视觉应用产品。(5)成功实现了在嵌入式系统中基于OpenCV的机器视觉车流量检测算法,其中包括采用概率统计的方法实现道路背景的重构,采用背景差法和最大类间方差法实现了车辆目标的提取,采用虚拟线圈法实现了通过车辆的计数。
[Abstract]:Road intelligent traffic flow detection technology is one of the key technologies of intelligent transportation system. It can detect the vehicle type, license plate, driving speed, traffic volume, road share and other traffic parameters of highway, general highway, urban road and so on. According to these parameters, we can judge the traffic condition of the road, make decisions and optimize the weak links in traffic. At present, the switching of traffic lights is fixed after manual setting, and the actual parameters reflecting traffic conditions are dynamic values, which can not be controlled by the change of actual traffic flow, which will inevitably lead to the low utilization rate of traffic resources. In order to avoid this situation, reduce the waiting time of vehicles and improve the vehicle traffic rate, it is necessary to control the switching time of red light and green light dynamically according to the traffic flow. Moreover, the intelligent transportation system is a very comprehensive system, which needs to return the traffic flow condition of each intersection to the traffic control center in time, and provide accurate data guarantee for the navigation of the vehicle through the comprehensive analysis of the center. In order to solve this problem in practical application, it is necessary to study an intelligent traffic signal control system based on machine vision. The system is an embedded machine vision application device, which can obtain the road traffic flow data after processing and analyzing the real-time video images, and then intelligently control the traffic signal device, and at the same time provide the data source for the advanced application of the system. In addition, another goal of this study is to realize an embedded platform for low coupling and modular machine vision applications of software and hardware systems. The coupling degree between machine vision algorithm and hardware is minimized. The advantages of this method are as follows: first, it can maximize the use of existing development resources, reduce the difficulty and complexity of development to speed up the formation of products; second, because it enriches the available software and hardware resources and reduces the difficulty and complexity of development, it can inevitably shorten the research and development cycle of the product and reduce the development cost. Thirdly, the software and hardware of the system are developed by modularization, which can be integrated flexibly according to the practical application requirements, which enhances the scalability of the system. Fourth, the method of this research can be quickly ported to mobile terminals such as mobile phones without modifying the machine vision algorithm, and can open up more application fields and markets. The innovations of this study are as follows: (1) A different realization idea and method from the existing machine vision algorithm in embedded application is explored, and the idea of modularization of software and hardware is used to realize the research goal; (2) using the latest ARM processor, the application of machine vision graphics and image processing is well realized. This method has good flexibility, portability and expansibility, low R & D cost and production cost, and can be well used in civil systems, especially in the field of mobile terminals. (3) the embedded system transplantation and application of the famous adaptive communication environment library ACE, graphic and image algorithm library OpenCV are successfully realized, which makes it possible to develop the machine vision application with the idea of modularization and low coupling degree of software and hardware systems. (4) A set of embedded system platform for embedded machine vision application is successfully constructed, which can conveniently transplant other machine vision algorithm modules and quickly form embedded machine vision application products. (5) the machine vision traffic flow detection algorithm based on OpenCV in embedded system is successfully realized, including the reconstruction of road background by probability and statistics method. The background difference method and the maximum inter-class variance method are used to realize the extraction of vehicle targets, and the virtual coil method is used to realize the counting of vehicles.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U495

【相似文献】