基于视频的车辆违章判别研究及无线传输实现

发布时间：2018-12-25 18:27

【摘要】：面对日益严重的交通状况和交通事故率,智能交通领域的视频监控系统越来越受到国内外学者的关注。随着计算机视觉、图像处理、模式识别、人工智能等技术应用于智能视频监控系统,这些技术和方法也逐步成为研究的重点和热点。然而由于违章转弯判定条件比较复杂,国内现在对于违章转弯的判别仍然依靠人工手段。因此,违章转弯的智能化检测方法具有重要的研究意义。论文的研究是基于视频的车辆跟踪、车辆的异常行为判别、视频的无线传输研究及实现展开的。主要工作为： 1.车辆跟踪算法的研究。论文首先介绍了Meanshift算法与kalman算法在跟踪领域的应用,随后提出了将kalman预测与Meanshift相结合的算法。新的算法比Meanshift算法在跟踪方面有很大程度的改善,当前景与背景颜色相似时,新的算法依旧可以跟踪运动目标且不会丢失,或者在遇到复杂场景时也可以对运动目标进行跟踪。提出了多目标跟踪状态的分类,对新目标的出现、目标的消失、目标的跟踪保持进行了定义,定义后的算法在进行多目标跟踪时,复杂度有所降低。 2.车辆的异常行为判别方法研究。论文主要研究的是基于传统的轨迹判别方法。依据轨迹的坐标变化来判别,也就是对车辆坐标位置的变化来判断车辆的行为。对左转、右转、直行的坐标变化设定阈值范围,并对于实际偏差的影响,增加判别的连续帧数,再判定车辆的行为,最后结合车辆的所属车道和路口的红绿灯来判别车辆是否有违章转弯和闯红灯等行为。 3.基于无线技术的视频传输实现。主要介绍了UWB技术特点,利用NS模拟了UWB的传输以及无线网桥技术的特点。由于UWB传输距离和技术的局限性,传输的距离只能在10m左右,相应的产品技术发展还不成熟。而无线网桥不仅传输的带宽宽,而且传输距离能够达到公里级。最终选用无线网桥来解决路口的视频与路口交换机之间距离远,传输困难等问题。用无线网桥技术将远距离的视频传输到实验室。论文在室外模拟路口的环境下搭建了视频采集与传输系统,该系统能够将400m外的视频进行采集,并利用无线网桥传回实验室。之后在服务器端编写了基于VC++6.0平台利用OpenCV对视频图像序列进行车辆检测、跟踪和违章判别。
[Abstract]:In the face of the increasingly serious traffic situation and traffic accident rate, the video surveillance system in the field of intelligent transportation has attracted more and more attention of scholars at home and abroad. With the application of computer vision, image processing, pattern recognition, artificial intelligence and other technologies to intelligent video surveillance systems, these technologies and methods have gradually become the focus and focus of research. However, due to the complicated judgment conditions of illegal turn, the judgment of illegal turn still depends on manual means in our country. Therefore, the intelligent detection method of illegal turn has important significance. The research of this paper is based on video vehicle tracking, vehicle abnormal behavior discrimination, video wireless transmission research and implementation. The main work is as follows: 1. Research on vehicle tracking algorithm. This paper first introduces the application of Meanshift algorithm and kalman algorithm in the field of tracking, and then proposes an algorithm combining kalman prediction with Meanshift. The new algorithm is much better than the Meanshift algorithm in tracking. When the foreground is similar to the background color, the new algorithm can still track the moving target without losing, or can track the moving target in the complex scene. The classification of multi-target tracking states is proposed. The emergence of new targets, the disappearance of targets and the tracking retention of targets are defined. The complexity of the defined algorithm is reduced when the multi-target tracking is carried out. 2. Research on the method of distinguishing abnormal behavior of vehicle. The main research of this paper is based on the traditional locus discrimination method. The vehicle behavior is judged by the change of the coordinate of the vehicle, that is to say, the change of the coordinate position of the vehicle. Set a threshold range for the coordinate changes of the left, right, and straight lines, and increase the number of continuous frames to determine the behavior of the vehicle, as well as the effect of the actual deviation. Finally, the traffic lights of the vehicle lane and intersection are combined to judge whether the vehicle has illegal turn and run red light and so on. 3. Video transmission based on wireless technology. This paper mainly introduces the characteristics of UWB technology, simulates the transmission of UWB by NS and the characteristics of wireless bridge technology. Due to the limitation of UWB transmission distance and technology, the transmission distance can only be about 10m, the corresponding product technology development is not mature. The wireless bridge not only transmits wide bandwidth, but also the transmission distance can reach km level. Finally, wireless bridge is chosen to solve the problem of long distance and difficult transmission between video and switch. Remote video is transmitted to the laboratory using wireless bridge technology. In this paper, a video acquisition and transmission system is built in the environment of outdoor analog intersection. The system can collect the video from 400m away and transmit it back to the laboratory by wireless bridge. Then, based on the VC 6.0 platform, the vehicle detection, tracking and violation discrimination of video image sequences based on OpenCV are written on the server side.
【学位授予单位】：扬州大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP391.41;U495

【参考文献】