基于SURF算法的视频贴片广告图标的检测方法  

A Video Advertisement Logo Searching Approach Based on SURF

作　　者：李炎[1] 杜秀华[1] 曹俊 

机构地区：[1]上海交通大学自动化系系统控制与信息处理教育部重点实验室,上海200240 [2]上海东方娱乐传媒集团广告经营中心,上海200041

出　　处：《实验室研究与探索》2013年第6期77-80,共4页Research and Exploration In Laboratory

摘　　要：视频媒体中的广告元素随着商业化步伐日趋增多,媒体市场的监测部门常常需要检测视频中某个广告图标出现的时间位置和存在的时间长度。针对这一需求,一种基于SURF算法的视频贴片广告图标的检测方法被提出。首先详细介绍SURF算法,通过对匹配比的计算等改进,以定量表示图像间的匹配程度,并将改进后的方法引入视频贴片广告的检测工作中,提出了一种检测视频中出现的贴片广告图标的方法,最后使用实际视频作为用例对方法进行验证,效果良好,结果准确。The advertising elements are increasing in the video media with the pace of commercialization. Video monitori,lg sections oflen need to detect the time position and time length of one certain advertisement logo in the video. To meet this requirement, a method based on SURF （Speeded Up Robust Features） approach is introduced in this paper. First, SURF is described in detail. Some improvements, such as the calculation of the matching ratio, are also introduced, to be expressed as the matching extent between images, and to be used in video detecting to detect advertisemenl Iogos. Then, an approach for logo detection is given. And finally, an example is taken to test this method, which shows that the performance is good and accurate.

关 键 词：视频处理 软广告 SURF算法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]