CASIA OpenIR  > 毕业生  > 硕士学位论文
Alternative TitleTraffic Abnormal Event Detection Under a PTZ Camera
Thesis Advisor常红星
Degree Grantor中国科学院研究生院
Place of Conferral中国科学院自动化研究所
Degree Discipline计算机应用技术
Keyword交通异常事件 Ptz摄像机 背景拼接 运动目标检测 交通拥堵检测 Traffic Abnormal Event Ptz Camera Background Mosaic Moving Object Detection Traffic Congestion
Abstract进入21世纪以来,我国城市化发展不断加快,城市公路交通系统的压力不断加大,交通拥堵、事故等日益频繁,逐渐成为经济和社会发展中的全球性问题。交通事件自动检测系统(Automatic Incident Detection Systems, AIDS)正是解决该问题的主要途径之一。本文以动态Pan-Tilt-Zoom(PTZ)摄像机为监控设备,基于图像处理技术对常见交通异常事件—拥堵事件的检测方法进行了研究和探讨。主要工作包括三部分内容:PTZ摄像机监控场景拼接、运动目标检测和基 于交通参数的交通拥堵事件检测与分析。 在PTZ摄像机监控场景的多帧图像拼接中,采用SURF特征点匹配方法估计透视变换模型参数,并引入Levenberg-Marquard(LM)算法进行参数优化,得到了快速精确的全景拼接效果。 在运动目标检测中,本文提出了一种全景图像快速检测的方法,基于由粗到细的定位思想,首先用直方图统计的方法初步得到相关区域,再利用特征点匹配得到精确位置,从而将前景象素与背景分离开,最后经过一系列形态学处理提取出准确的运动目标,达到了实时检测的效果。 在交通拥堵事件检测与分析中,本文基于检测的方法提取道路车辆的交通参数,并对交通拥堵重新定义,将交通状态分为畅通、轻微拥堵和严重拥堵三类。算法首先提取图像中的路面作为感兴趣区域,然后通过实时检测路面车辆,统计车辆的排队长度和空间占有率两个交通参数,实时报告道路拥堵状况。实验结果表明该算法能够成功地识别出各种交通拥堵事件。
Other AbstractSince the 21st century, the pressure of traffic system on urban roads is increasing, with the accelerating development of China’s urbanization. More and more traffic congestion, accidents, etc., occur nowadays, which gradually become the global common concern in the economic and social development. Automatic Incident Detection System (AIDS) is one of the chief solutions to this problem. In this paper, the method for detecting traf- fic congestion—one of common traffic abnormal events—is studied on the basis of video images taken by a dynamic PTZ camera. Our research work consists of three parts, which are the PTZ camera monitoring scenarios mosaiking, moving objects detection, and traffic congestion detection and analysis based on traffic information parameters. In the first part, we use SURF feature-matching to estimate transformation model of multi-frames and mosaic a panorama image for the PTZ camera monitoring scenarios. In order to achieve a fast and precise mosaiking effect, we introduce Levenberg-Marquard (LM) algorithm for parameter optimization. In the second part, a method of “Rapid Detection of Pano-ramic Im- ages” is proposed, which is based on the thought of coarse-to-fine posi- tioning. Firstly, we get the corresponding region by histogram algorithm;then find the exact position by feature-matching method; thirdly, separate the foreground and background based on background subtraction; finally,extract the accurate objects through morphological processing. The algorithm achieves real-time and accurate detection effect. In the third part, traffic congestion detection and analysis, we propose a novel method about extracting traffic information parameters based on moving detection. A redefinition of traffic congestion is proposed, and the traffic status is classified into three categories: smooth, slight congestion, and serious congestion. The algorithm firstly extracts the road as a region of interest (ROI), and then calculates the two parameters that queue length and space occupancy through detecting road vehicles, at last reports the traffic status based on these parameters. Experiment results indicate that this method can detect traffic congestion successfully.
Other Identifier200928014628074
Document Type学位论文
Recommended Citation
GB/T 7714
张玉双. PTZ摄像机下的交通场景异常事件检测[D]. 中国科学院自动化研究所. 中国科学院研究生院,2012.
Files in This Item:
File Name/Size DocType Version Access License
CASIA_20092801462807(14114KB) 暂不开放CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[张玉双]'s Articles
Baidu academic
Similar articles in Baidu academic
[张玉双]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[张玉双]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.