双视场运动目标智能监控系统研究

发布时间：2018-12-21 11:28

【摘要】：一直以来,智能监控技术都是计算机视觉领域的研究热点。随着人们对生活品质要求的不断提高,传统的监控系统早已不能满足人们的应用需求。针对传统监控系统监控目标的清晰度有限和冗余数据量大的问题,本文提出一种多摄像机的双视场运动目标智能监控系统,其中宽视场主摄像机负责对大范围视野进行整体监控,并调度窄视场从摄像机对指定的运动目标进行主动跟踪,窄视场从摄像机负责获取高分辨率的目标图像。本文的主要工作及创新点概括如下:(1)不同于传统的多摄像机系统,本文对监控系统的摄像机进行了分工,以高分辨率的运动目标图像代替视频序列,在控制产生的冗余数据量的前提下,既实现了大范围的整体监控又得到了运动目标的细节信息。(2)对图像畸变校正算法、运动目标检测算法、运动目标跟踪算法和目标交接算法分别进行了研究和实现。重点研究了摄像机标定技术、背景差分法的不同背景模型、颜色特征匹配、不同颜色空间的特点、基于SURF(Speeded Up Robust features)算法和基于 ORB(Oriented FAST and Rotated BRIEF)算法的角点匹配技术。(3)结合信息熵的概念,对运动目标检测的背景差分法进行改进,改进的算法兼顾了计算速度和检测精度。(4)针对本文的应用,提出融合角点特征和颜色特征的运动目标交接算法,实现了双视场运动目标的实时准确交接。(5)在 VS2010(Microsoft Visual Studio2010)平台上使用 C++语言实现了系统的视频采集模块、图像校正模块、运动目标检测模块、目标交接模块、PTZ(Pan-Tilt-Zoom)摄像机云台控制模块和运动目标主动跟踪模块。
[Abstract]:Intelligent monitoring technology has always been a hot topic in the field of computer vision. With the improvement of the quality of life, the traditional monitoring system can not meet the needs of people. In order to solve the problems of limited clarity and large redundant data amount in traditional monitoring system, this paper presents an intelligent monitoring system for moving targets with two fields of view with multiple cameras. The wide field of view main camera is responsible for the overall monitoring of the wide field of view, and the narrow field of view is responsible for actively tracking the designated moving target from the camera, and the narrow field of view is responsible for obtaining the high resolution target image from the camera. The main work and innovations of this paper are summarized as follows: (1) different from the traditional multi-camera system, this paper divides the cameras of the surveillance system and replaces the video sequences with high-resolution moving target images. On the premise of controlling the amount of redundant data generated by the control system, both the global monitoring and the detailed information of the moving object are realized. (2) the image distortion correction algorithm, the moving target detection algorithm, the image distortion correction algorithm, the moving target detection algorithm, the image distortion correction algorithm, the moving target detection algorithm, The moving target tracking algorithm and the target handover algorithm are studied and implemented respectively. The techniques of camera calibration, background difference method, color feature matching, and different color space are studied in detail. Corner matching technology based on SURF (Speeded Up Robust features) algorithm and ORB (Oriented FAST and Rotated BRIEF) algorithm. (3) based on the concept of information entropy, the background difference method for moving target detection is improved. The improved algorithm takes into account the computation speed and the detection accuracy. (4) aiming at the application of this paper, a new algorithm of moving target handover is proposed, which combines corner feature and color feature. It realizes the real-time and accurate transfer of the moving object in the double field of view. (5) the video acquisition module, the image correction module, the moving target detection module and the target handover module are realized by using C language on the VS2010 (Microsoft Visual Studio2010 platform. PTZ (Pan-Tilt-Zoom) camera control module and moving target active tracking module.
【学位授予单位】：华东师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41

【参考文献】