【作者】 伍友龙；

【导师】 刘伟铭；

【作者基本信息】 长沙理工大学 ， 计算机应用技术， 2005， 硕士

【摘要】 数字图像处理技术是蕴涵巨大发展潜力的新兴科学研究领域之一,在交通行业中,其应用日益广泛。基于图像分析的事件自动检测系统是利用计算机视觉、神经网络、模糊逻辑和遗传算法等技术和方法进行交通事件检测、车辆识别和道路利用监视,其性能完全优于传统的基于交通流特征的事件检测算法。本文实现了一个由车辆检测、形态滤波、连通域分析、车辆跟踪、事件检测等部分组成的高速公路交通事件检测算法,并针对实现过程中遇到的问题及难点提出了相应的解决方法。本文在阴影检测中,利用颜色空间C1C2C3的颜色不变性质,提出了有效的阴影检测算法。由于在高速公路上的行驶车辆可以近似认为是直线匀速运动的,所以根据Kalman滤波理论建立了运动车辆状态预测模型,该模型能够同时进行多车辆跟踪,而且根据跟踪结果判断车辆是否遮挡、分裂,在车辆产生遮挡的区域运用金字塔搜索方法进行模板匹配,对遮挡车辆进行分割。在基于轨迹的事件检测中,通过直线拟和的方法,把复杂的车辆跟踪轨迹曲线简化为直线组合,以直线差角代替轨迹曲线曲率变化,以此分析车辆行驶方向的变化,最后判定事件。并且在本文中给出了车辆速度估计方法,能够检测车辆超速、慢速等事件;而且运用了隐马尔可夫模型(HMM)进行高速公路车辆碰撞事件的预测。同时,本文应用VC++开发工具与OPENCV图像开发包实现了算法,开发了系统原型,并对录制的高速公路视频监控录像做了实验,验证了算法的有效性。更多还原

【Abstract】 Digital image processing technology is a new science research field which has more development potential, and it is applied more and more in traffic industry. The system of freeway traffic accident detection based on image analysis makes use of the technology of computer vision, artificial neural network, fusion logic, and genetic algorithm, etc. And it has the better efficiency than the traditional traffic flow-based accident detection system. In This thesis, an algorithm of freeway incident automatic detection based on video is proposed. Vehicle detection, morphologic filter, connected region analysis, vehicle tacking and incident detection are included in the algorithm, and the solution for the problem and difficulties in the implementation are proposed. In the shadow detection, an algorithm based the invariability of color space C1C2C3 is proposed. Because the vehicles’running on the freeway can be looked as at a uniform speed in a straight line, a moving vehicle status prediction model based Kalman filter is building, and according to the result of Kalman filter tracking, overlapped vehicles and splited vehicles are determined, template matching based pyramid searching is implemented to overlapped vehicle segmentation; Through straight line fitting, complicated vehicle trajectory is simplified to the collection of lines, and analyzing the change of driving based the angle difference between lines, and the incident is determine. Furthermore, the algorithm based Hide Markov Model is applied to detect vehicles collision. At the same time, the incident detection algorithm is implemented based on VC++ and OPENCV, and it worked on the video recorded on freeway well and proved to be available.更多还原