节点文献

多摄像头视频融合系统的设计与实现

Design and Implementation of Multi Camera Video Fusion System

【作者】 吴琼

【导师】 卢光辉; 杨盛国;

【作者基本信息】 电子科技大学 , 软件工程(专业学位), 2013, 硕士

【摘要】 随着科技的发展,人们对电子产品的要求也越来越高。而在实际生活中,很多电子产品已经满足甚至超出了人们的需求,例如摄像头行业在近几年也有很快的发展,其中视频监控在日常生活,交通监管甚至军事领域得到广泛的应用。然而现有的摄像头还是无法满足人们全景捕捉的需求,摄像头较窄的视角给视频监控的发展带来了不小的阻碍。主要体现在以下两方面:(1)监控遗漏,由于摄像头视角较小,一般采用多摄像头同时监控,可是多摄像头得到的视频资源相对独立,视频信息分散,一般会造成监控死角,而且监控起来较为困难,最终造成监控遗漏。(2)冗余资源,在多摄像头的分散监控的情况下,为了减少监控死角,一般会增加摄像头的数量,这样必定会造成视频资源的冗余。本文首先主要对DirectShow技术进行了详细的介绍,对其架构和使用方法加以描述,为系统获取摄像头中的图像打下基础。然后对SIFT算法进行了详细的解释和分析。在同一时刻获取多个视频图像,对图像进行分析,获取图像SIFT特征点,然后对特征点进行匹配,通过点匹配图像融合的方式,获取图像融合参数,再对每一帧图像进行处理。同时本文对视频拍摄也进行了研究。在研究中发现视频拼接需要在严格的环境下才能比较好的实现。因为摄像头视频一般为近景拍摄,这种情况下会有大角度的三维变化,即从不同的角度,物体成像完全不同。对于拥有不同信息的图像,拼接也就没有意义了。本文基于这一特点,提出了固定摄像头的视频拼接的方法,文章最后给出了拼接算法的具体实现,将视频进行裁剪和拼接,并融入到本系统中,实现视频的融合,给出了系统的总体设计与实现。通过系统运行图,演示了整个系统的运行流程,并得出拼接结果。实验结果表明,本系统能够快速而有效的实现摄像头视频的融合。

【Abstract】 With the development of science and technology, people have more and morerequirement for electronic products.But in real life,many electronic products have tomeet and even exceed the needs of the people,for example, the camera industry in recentyears has developed rapidly, the video surveillance in daily life, is widely applied intraffic control and military field.However, the existing camera still can’t meet peoplepanoramic capture demand, camera’s narrow angle has brought not small obstacles tothe development of video monitoring.Mainly as(1). Monitoring missing, due to camera angle smaller, while the generalmulti-camera monitoring, but the multi-camera video resources are relativelyindependent, video information is scattered, have been such a video, usually caused bymonitoring dead, and monitor them More difficult, eventually leading to monitor theomission.(2). Redundant resources in distributed multi-camera monitoring the situation,monitor the dead in order to reduce the general will increase the number of cameras,video resources that will lead to redundancy.First, DirectShow technology are introduced in detail in this thesis,and thestructure and using methods are described,laid a foundation for obtaining the cameraimages for system.Then carried on the detailed explanation and analysis of SIFT algorithm.Acquiringa plurality of video image at the same time, analysising the image, geting image SIFTfeature points, and then the feature points are matched by point matching and imagefusion, getting image fusion parameters,and then processing each frame of image.At the same time, the video was also studied.Found in a study indicates that videostitching needs to better in strict environment. Because the video camera for shootingclose-up,3D change this situation will have a large angle, namely from the differentangle, the image of the object is completely different. For the images have differentinformation, mosaic also makes no sense. In this thesis, based on this characteristic, Iput forward the method of video mosaic fixed camera. Finally, the concrete realization of stitching algorithm is given, the video cuttingand splicing, and integrating into the system,and achieving the overall design of thesystem.By running the system diagram,operation process demonstrates the wholesystem, and draw the stitching results.Experimental results show that the system can achieve rapid and effectiveintegration of the video camera.

【关键词】 DirectShowSIFT裁切中线视频融合
【Key words】 DirectShowSIFTCutMidlineVideo Fusion
  • 【分类号】TP391.41
  • 【被引频次】5
  • 【下载频次】609
节点文献中: