基于像面运动的事故车辆车速计算

发布时间：2018-03-26 14:30

本文选题：事故车辆　切入点：城市视频　出处：《长安大学》2014年硕士论文

【摘要】：事故车辆的车速，作为事故鉴定中重要的一环，不仅能描述发生事故的性质，还有利于对事故过程的重现。传统的车速鉴定手段，通过获取事故现场定位参数、事故痕迹、散落物等信息人工计算得出，不仅鉴定过程复杂，而且在一段时间内降低了道路通行能力。因此，利用直观、实时性好、对道路交通影响小的事故视频进行车速鉴定具有很强的运用价值。本文通过分析监控摄像头标定技术、运动目标检测与跟踪技术的计算模型与计算步骤，提出了新的标定方法与运动目标(车辆)检测、跟踪过程，并通过VC++程序得以实现，建立了基于VC++的车速自动化处理系统，最后通过实例完成系统实效性的检验。对监控摄像头标定方法进行研究。分析了监控摄像头标定的线性方法与非线性方法的计算过程、模型与参数，提出了根据聚点及投影透视关系的一种简洁、稳定的标定方法，通过具体环境实验进行精度校核。对运动目标(车辆)检测技术以及跟踪技术进行研究。根据计算模型与计算方法的不同，分析了运动目标(车辆)检测中常用的光流法、帧间差法以及背景减法，，比较了多种技术的优缺点，设计了运用背景帧差模型与三图像帧差模型相结合的车流体区域检测模型。对提取后的目标车辆运动中心进行微元惯性运动分析，设计了车辆流体区域跟踪模型。开发车速计算自动化处理系统。以VC++为平台，建立视频图像读入与提取模块，完成对视频流的单帧图像分离；依据实际路面参数，建立坐标点映射关系以及空间坐标系计算模块；建立车流体区域定位模块与跟踪模块，开发出“事故车辆车速计算系统”，并通过实例完成系统实效性的检验。
[Abstract]:As an important part of accident identification, the speed of vehicle is not only able to describe the nature of the accident, but also conducive to the recurrence of the accident process. It is found that the identification process is not only complicated, but also reduces the road capacity in a period of time. It is very valuable to identify the speed of road traffic accident video. In this paper, a new calibration method and the detection and tracking process of moving object (vehicle) are put forward by analyzing the calculation model and steps of camera calibration technology, moving target detection and tracking technology, and are realized by VC program. The automatic speed processing system based on VC is established. Finally, the effectiveness of the system is verified by an example. This paper studies the calibration method of surveillance camera, analyzes the calculation process, model and parameters of linear method and nonlinear method of camera calibration, and proposes a simple and stable calibration method based on the relationship between focus point and projection perspective. The precision is checked by specific environmental experiments. This paper studies the detection technology and tracking technology of moving target (vehicle). According to the difference of calculation model and calculation method, the optical flow method, inter-frame difference method and background subtraction method are analyzed, which are commonly used in moving target (vehicle) detection. After comparing the advantages and disadvantages of various technologies, a vehicle fluid region detection model combining background frame difference model and three-image frame difference model is designed. The extracted vehicle motion center is analyzed by micro-element inertial motion analysis. The vehicle fluid region tracking model is designed. The automatic processing system of speed calculation is developed. Based on VC platform, the module of video image reading and extracting is established, and the single frame image separation of video stream is completed. The mapping relation of coordinate points and the calculation module of spatial coordinate system are established, and the regional positioning module and tracking module of vehicle fluid are established, and the "vehicle speed calculation system" is developed, and the effectiveness of the system is verified by an example.
【学位授予单位】：长安大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U491.31

【参考文献】