电子警察系统中车辆跟踪及夜间车辆检测算法研究

发布时间：2018-04-19 12:32

本文选题：电子警察 + 嵌入式　；参考：《西安电子科技大学》2014年硕士论文

【摘要】：随着交通工具的不断增长,交通管理变的越来越困难,因此,如何构建高效便捷的智能交通管理系统成为了人们关注的焦点。作为智能交通系统的核心模块之一,电子警察系统备受关注,其中,车辆检测与跟踪是其功能完善的基础。本文主要针对车辆跟踪及夜间车辆检测算法进行研究,并将算法移植到嵌入式电子警察系统中。主要研究工作如下:1.车辆跟踪算法研究。本文首先研究了Hash全局特征的车辆跟踪以及LBP特征和Hue特征相融合的车辆跟踪,给出了相应的实验效果并进行性能对比分析。然后,针对嵌入式电子警察系统的实时性要求,在传统的光流法上进行改进算法研究,取得了良好的跟踪效果。最后,在Haar特征和Adaboost分类器的基础上进行多目标跟踪算法研究,取得了较好的实验效果。2.夜间频闪灯的使用及EagleEye3智能相机的参数配置。由于夜间光线较暗,车辆的很多细节特征会被淹没掉,使得夜间车辆检测变得很棘手,因此,夜间频闪灯的使用及Eagle Eye3智能相机的夜间参数配置显得尤为重要。此处首先介绍了Eagle Eye3智能相机的基本框架及开发环境,然后针对频闪灯的使用以及Eagle Eye3智能相机的几个重要参数进行了详细地讨论,最后结合工程项目实践给出了最佳的组合参数配置。3.嵌入式电子警察系统中的夜间车辆检测算法实现。本文首先研究了基于车灯的夜间车辆检测算法,取得了较好的检测效果,但当交通场景中存在部分车灯损坏或者是其他非机动车干扰时,该算法无法正常检测。因此,接下来研究了基于Haar特征和Adaboost分类器的夜间车辆检测算法以及基于车牌的夜间车辆检测算法,并将后两种算法移植到EagleEye3智能相机中,通过实验测试以及数据统计进行比较分析。
[Abstract]:With the continuous growth of vehicles, traffic management becomes more and more difficult. Therefore, how to build an efficient and convenient intelligent traffic management system has become the focus of attention.As one of the core modules of intelligent transportation system, electronic police system is paid more attention to, and vehicle detection and tracking is the foundation of its perfect function.In this paper, vehicle tracking and night vehicle detection algorithms are studied, and the algorithm is transplanted to embedded electronic police system.The main research work is as follows: 1.Research on vehicle tracking algorithm.In this paper, the vehicle tracking based on global features of Hash and vehicle tracking based on LBP and Hue features are studied. The corresponding experimental results are given and the performance is compared and analyzed.Then, according to the real-time requirement of embedded electronic police system, the improved algorithm is studied on the traditional optical flow method, and good tracking effect is obtained.Finally, on the basis of Haar feature and Adaboost classifier, the multi-target tracking algorithm is studied, and good experimental results are obtained.Use of nightly stroboscopic lights and parameter configuration of EagleEye3 smart camera.Because of the dim light at night, many details of the vehicle will be drowned out, which makes the detection of vehicles at night difficult. Therefore, the use of night-time stroboscopic lights and the configuration of night parameters of Eagle Eye3 smart camera are particularly important.This paper first introduces the basic frame and development environment of Eagle Eye3 smart camera, then discusses the use of stroboscopic lamp and several important parameters of Eagle Eye3 smart camera in detail.Finally, the best combination parameter configuration. 3. 3 is given according to the project practice.Implementation of night vehicle detection algorithm in embedded electronic police system.In this paper, we first study the vehicle detection algorithm based on vehicle lights and obtain good detection results. However, when there are some vehicle lights damaged in traffic scene or other non-vehicle interference, the algorithm can not detect normally.Therefore, the night vehicle detection algorithm based on Haar feature and Adaboost classifier and the nocturnal vehicle detection algorithm based on license plate are studied, and the latter two algorithms are transplanted to EagleEye3 intelligent camera.Through the experiment test and the data statistics carries on the comparative analysis.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U495;TP391.41

【共引文献】