基于鱼眼摄像头的网络监控系统研究

发布时间：2018-03-29 18:38

本文选题：鱼眼图像　切入点：畸变校正　出处：《宁波大学》2015年硕士论文

【摘要】：最近几年,伴随通信技术、计算机技术和多媒体编解码技术的进步,视频监控技术获得快速发展,并在人们生产生活的各个方面得到广泛应用,已经涉及到学校、政府部门、金融系统、交通、医疗系统、大型工厂等领域。随着其不断扩大的市场应用,新的需求随之产生,比如大范围无盲区实时监控。传统监控镜头因其视角范围较小,要实现大范围内的实时无盲区监控,需要机械地转动单个镜头或者对多个镜头捕捉的画面进行拼接。上述方案不可避免的导致摄像头数量的增加、多镜头曝光时间调节难度加大等问题。鱼眼镜头焦距短,达到甚至超过180°视角的优势,使其成为大视场无盲区实时监控系统中的最佳选择。本文在监控端选取鱼眼镜头作为监控摄像头,该鱼眼镜头吸顶安装后,可以获取水平方向360°以及垂直方向180°范围内的场景信息,搭建网络视频监控系统,在客户机上接收来自网络的流媒体数据并解码、畸变校正处理和显示。本文内容有以下部分:(1)首先介绍了本课题的研究背景、意义和现状;(2)论述网络监控系统的总体框架及方案;(3)通过分析鱼眼镜头成像原理,并对几种鱼眼畸变校正算法进行仿真和分析,从而提出建立一种新的校正模型的必要性;(4)提出本课题改进的校正算法,包括特征点的提取和伪特征点剔除,模型的建立,图像的插值,以及校正后图像的展开模式;(5)论述网络传输部分相关技术,包含网络传输协议的选择以及网络体系结构的布局;(6)基于Directshow技术的客户端设计,包括视频流的接收、解码、全景展开模式程序设计和小视角显示模式的程序设计;(7)实验结果的展示,工作总结和进一步的展望。实验结果表明,对分辨率为1600*1200的鱼眼图像,采用本课题的方法后校正效果较好,而且单帧图像处理时间小于35毫秒,使得视频能够流畅展示,达到了实时性。同时通过人机交互,实现了场景漫游,并使校正后的图像以不同的展开模式呈现出来,达到了预期结果。
[Abstract]:In recent years, with the development of communication technology, computer technology and multimedia coding and decoding technology, video surveillance technology has been rapidly developed, and has been widely used in all aspects of people's production and life, has been involved in schools, government departments, Financial systems, transportation, medical systems, large factories, and so on. With its expanding market applications, new needs arise, such as real-time surveillance in a wide range of blind areas. In order to realize real-time blind area monitoring on a large scale, it is necessary to rotate a single lens mechanically or to splice the images captured by multiple lenses. The above scheme inevitably leads to an increase in the number of cameras. It is difficult to adjust multi-lens exposure time and so on. The focal length of fish-eye lens is short, reaching or even exceeding 180 掳angle of view. In this paper, the fish-eye lens is selected as the surveillance camera at the monitoring end, and the fish-eye lens is installed after absorbing the top of the camera. It can obtain scene information in the range of 360 掳horizontal and 180 掳vertical direction, set up a network video surveillance system, receive streaming media data from the network on the client and decode it. Distortion Correction processing and display. The content of this paper is as follows: 1) first of all, the research background, significance and status quo of this subject are introduced. (2) the overall framework and scheme of network monitoring system are discussed. (3) the principle of fish-eye lens imaging is analyzed. Several algorithms for correction of fish-eye distortion are simulated and analyzed, and the necessity of establishing a new correction model is put forward. (4) the improved correction algorithm is proposed, including feature point extraction and pseudo-feature point elimination, and the establishment of model. The interpolation of image and the unwrapping mode of corrected image are discussed in this paper, including the selection of network transmission protocol and the layout of network architecture. The client design based on Directshow technology, including the receiving of video stream, is discussed. Decoding, panoramic development mode programming and program design for small view display mode.) display of experimental results, summary of work and further prospects. The experimental results show that, for fish-eye images with a resolution of 1600 ~ 1200, The correction effect of this method is good, and the processing time of single frame image is less than 35 milliseconds, which makes the video display smoothly and achieves real time. At the same time, the scene roaming is realized through man-machine interaction. The corrected images are presented in different expansion modes, and the expected results are achieved.
【学位授予单位】：宁波大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP391.41;TN948.6

【参考文献】