【作者】 陈阳；

【导师】 周明全；

【作者基本信息】 西北大学 ， 计算机软件与理论， 2005， 硕士

【摘要】 随着城市化速度的加快,机动车日益普及,人们在享受机动车所带来的巨大便利的同时,也面临着交通拥挤的困惑。然而直接地去修建更多的路桥却根本赶不上车辆的发展速度。在现有的条件下,提高交通控制和管理水平,合理使用现有交通设施,充分发挥其性能,是解决交通问题的有效方法之一。随着计算机硬件技术和计算机视觉技术的发展,基于计算机视觉的交通监控系统成为可能。基于视频车辆的实时检测和跟踪是智能交通监控系统的核心部分。目前存在的检测和跟踪技术在复杂场景下、大范围、多目标的情况下,运动目标的分割和跟踪的效果不是很理想,需要进一步改善。就此现状,本文主要做了如下研究工作: 采用了基于混合高斯分布的自适应背景模型区分前景和背景,从而对前景运动目标进行捕捉,但是此方法忽略了相邻像素的相关性,本文通过形态滤波的方法对其进行了改进。 在跟踪方面提出了以扩展卡尔曼滤波为基础的跟踪模型,对物体进行跟踪。通过采用扩展卡尔曼滤波模型,减小了搜索范围提高了搜索的精度和算法的效率。并使用彩色直方图来对前后帧之间对应的目标区域图像进行匹配。 在交通参数测量方面,通过将屏幕坐标转换为真实世界坐标,然后根据真实世界坐标,实现了对车速车型等数据进行测量。 最终实现的系统,适用于大面积、多目标的复杂场景,能排除干扰,统计车流量、车速和简单的车型分类,可应用于高速公路和城市交通的管理中。更多还原

【Abstract】 With the acceleration of urbanization, automobile became more and more popular and people are enjoying the convenience that offer, however we are trapped in the bewilderment of traffic congestion. However to directly construct more highway bridges could not to be able to catch up with the development speed of the vehicles. Under the existing condition, enhancing the level of transportation control and the management, the reasonable use existing traffic facility, and fully displaying its performance is the effective methods to solve transportation question. Along with the computer hardware technology and the computer vision technology development, a computer vision-based traffic monitoring system has become possible. Vehicle detection and tracking real-time system based on video is the key to traffic monitoring system. Many popular related technology didn’t meet vary requirements. Therefore, an efficient and robust detection and tracking system is needed eagerly. For these all, our research is mainly in following:This thesis used the auto-adapted background model based on mixture of the Gaussians to distinguish foreground object and the background, thus carried on the detection to the foreground movement object. At the same time we propose that with morphology filter we avoid the shortcoming of ignore the relationship between the pixels.In track aspect this thesis used track model based on expanded kalman filtering, carried on the track to the object. By expands the kalman filtering model, reduced the hunting zone to increase the search precision and the algorithm efficiency. And we use color histogram to match the image of object area between the continuous frames.In the transportation parameter survey aspect, we firstly transform the screen coordinates into the real world coordinates, then with the real world coordinates we could survey the speed and simply classify the vehicle.Finally, the algorithm of vehicle detection and tracking used in this paper can limit the noise of system and pedestrian factor, and used in large area, multiple objects and complex environment in traffic surveillance. And the implementation can be applied in the highway and in the municipal transportation management.更多还原