平安城市视频监控系统中异常行为识别技术研究与实现

发布时间：2018-02-01 00:31

本文关键词： 平安城市安防视频监控异常行为识别运动目标检测　出处：《南京邮电大学》2014年硕士论文　论文类型：学位论文

【摘要】：随着科学技术的不断发展,城市安防视频监控越来越广泛地运用到平安城市的各种场合。城市安防视频监控系统中计算机进行视频信息的压缩、储存、分析、显示以及报警等自动化处理,从而实现无人值守；通过网络平台实现了远距离视频监控,即使是数千公里外也能达到亲临现场的效果；利用先进的软件系统不仅在几分钟内便可完成传统视频监控中大量的数据分析,提高了视频监控效率,且能获得更为逼真、清晰的数字化图像质量与更为便捷、实用的视频监控管理和维护。通过城市安防视频监控项目的建设,实现覆盖整个城市的城市安防视频监控,城市报警联动,车辆跟踪,智能卡口等功能,实现整个城市安全视频监控,为公安系统提供统一的视频监控指挥平台,通过先进的信息化技术,提高城市安全建设水平。城市中心作为一个高度复杂、人口密集的场所,城市安防视频监控的有效利用可以保障公共执法人员和人民群众的安全。对该类场所群众日常行为及特殊情况下的视频监控,是平安城市管理不可缺少的工作。在平安城市场景中,我们关心城市安防视频监控中的异常行为,识别视频监控画面中的异常情况,并能够以实时可靠的方式发出警报,有效的协助视频监控人员获取准确信息和处理突发事件以阻止更多危险活动的发生。因此如何将这些城市安防视频监控中出现的异常行为准确检测和识别并及时通知处理是文章研究的主要内容。文章研究平安城市视频监控系统中异常行为识别相关技术,主要对视频序列进行前景提取、目标行为识别等问题进行了研究,并实现了对一个及多个运动前景的实时异常行为的检测。文章首先对几种最新异常行为的识别方法进行了算法实现,解决了如打斗等这类剧烈运动特征复杂且提取图像速度慢的技术难题,接着提出了一种基于目标轮廓特征检测、目标中心点跟踪和区域光流特征检测相结合的异常行为识别新方法。文章使用VC技术和开源计算机视觉库进行城市安防视频监控异常行为算法的实现,并且通过现场实验验证了算法的有效性。
[Abstract]:With the continuous development of science and technology, urban security video surveillance is more and more widely used in various occasions of the city. The computer in the city security video surveillance system compresses, stores, and analyzes the video information. Display and alarm automatic processing, so as to achieve unattended; Through the network platform to achieve remote video surveillance, even thousands of kilometers away can also achieve the effect of in-person presence; Using advanced software system can not only complete a large amount of data analysis in traditional video surveillance in a few minutes, but also improve the efficiency of video surveillance, and can obtain more realistic, clear digital image quality and more convenient. Practical video surveillance management and maintenance. Through the construction of the city security video surveillance project, the implementation of the city security video surveillance covering the entire city, the city alarm linkage, vehicle tracking, intelligent bayonet and other functions. Realize the whole city security video surveillance, provide a unified video surveillance command platform for the public security system, through advanced information technology, improve the level of urban security construction. The city center as a highly complex. In densely populated places, the effective use of urban security video surveillance can ensure the safety of public law enforcement personnel and people. In the safe city scene, we care about the abnormal behavior in the city security video surveillance, and identify the abnormal situation in the video surveillance screen. And can give an alarm in a real-time and reliable manner. Effectively assist video surveillance personnel to obtain accurate information and deal with emergencies in order to prevent more dangerous activities. Therefore, how to accurately detect and identify the abnormal behavior in the security video surveillance of these cities and timely. Notification processing is the main content of this paper. This paper studies the detection technology of abnormal behavior in the video surveillance system of Pingan city. This paper mainly studies the foreground extraction and target behavior recognition of video sequences. And the detection of real-time abnormal behavior of one or more motion prospects is realized. Firstly, several new methods of abnormal behavior recognition are implemented. This paper solves the technical problem of such violent motion features as fight and so on, which is complex and slow to extract images. Then, a new method based on target contour feature detection is proposed. This paper uses VC technology and open source computer vision library to realize the abnormal behavior algorithm of urban security video surveillance. The effectiveness of the algorithm is verified by field experiments.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP391.41;TN948.6

【相似文献】