基于视频点播的广告嵌入及检索系统的设计与实现

发布时间：2018-03-06 18:24

本文选题：视频点播　切入点：图像匹配　出处：《天津大学》2014年硕士论文　论文类型：学位论文

【摘要】：在互联网飞速发展的时代,人们观看视频的方式也在改变,现在的网民大都喜欢在网络上观看视频,海量的网络视频资源不仅满足了用户追求热播视频的需求,同时也给广告商提供了平台。时下最为常见的网络视频广告形式是贴片广告,这种加贴在视频头尾的与视频内容毫无关联的广告不仅无法取消强制观看,且时长也在不断增加。据统计,这种强制且缺乏针对性的广告模式让用户难以接受,那么如何在网络视频里推广自己的广告又不至于让用户反感就成为广告商们当前急需解决的问题,本系统的总体目标就是实现一个既能让用户不再饱受贴片广告困扰又能让广告商投放广告的视频点播系统。该系统根据时下视频点播的特点和用户需求分析,在功能上将广告嵌入到视频中,同时吸收国内外成熟图像分析算法的研究成果,让用户主动触发广告,呈现一种新的广告投放模式,补充乃至最终替换传统视频广告的展现形式。该系统根据颜色直方图和SIFT(Scale Invariant Feature Transform)算法的不同特点取长补短,采用将两种算法相结合的优化算法进行图像匹配,并结合三层架构、文件流的序列化、反序列化和视频流的写入技术,实现了将广告内容与视频内容紧密结合在一起,广告和视频合二为一的目标,增加广告粘度,且让用户主动触发广告的模式更容易被用户所接纳,深化了广告的传播效果,同时提高了广告主投放广告的积极性,形成一个良性循环。经过测试和试用,该系统取得良好的效果。
[Abstract]:In the era of rapid development of the Internet, the way people watch video is also changing. Nowadays, most Internet users like to watch video on the Internet. The vast amount of online video resources not only meet the needs of users in pursuit of popular video. It also provides a platform for advertisers. The most common form of online video advertising today is patch advertising, which has no connection to the content of the video, and is not only unable to cancel mandatory viewing. And the length of time is on the increase. According to statistics, this kind of compulsory and lack of targeted advertising model makes it difficult for users to accept, Well, how to promote their own advertisements in the network video without causing users to resent them has become a problem that advertisers urgently need to solve at present. The overall goal of this system is to realize a video-on-demand system, which can not only make users not suffer from patch ads, but also allow advertisers to place advertisements. The system is based on the characteristics of VOD and the needs of users. At the same time, it absorbs the research results of mature image analysis algorithms at home and abroad, so that the user can trigger the advertisement actively and present a new advertising mode. According to the different characteristics of the color histogram and SIFT(Scale Invariant Feature transform algorithm, the system uses the optimized algorithm which combines the two algorithms to match the image, and combines the three-layer structure, according to the different features of the color histogram and the SIFT(Scale Invariant Feature transform algorithm. File stream serialization, deserialization and video stream writing technology achieve the goal of combining advertising content with video content, advertising and video into one, increase advertising viscosity, And let the user trigger the model of advertising more easily accepted by users, deepen the effect of advertising dissemination, at the same time improve the advertising initiative of advertisers, form a virtuous circle. After testing and trial, The system has achieved good results.
【学位授予单位】：天津大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN948.64

【相似文献】