首页 | 本学科首页   官方微博 | 高级检索  
     

基于直方图差分的视频分类方法
引用本文:孟丽,徐法升,李金屏. 基于直方图差分的视频分类方法[J]. 济南大学学报(自然科学版), 2007, 21(2): 100-103
作者姓名:孟丽  徐法升  李金屏
作者单位:济南大学,信息科学与工程学院,山东,济南,250022;济南大学,信息科学与工程学院,山东,济南,250022;济南大学,信息科学与工程学院,山东,济南,250022
基金项目:山东省高校中青年学术骨干计划;济南大学校科研和教改项目
摘    要:针对当前视频自动分类算法复杂计算量大等问题,提出一种简单的自动视频分类方法,即基于直方图差分的统计方法。获得每个视频的直方图差分曲线后,调整去掉编辑特效造成的虚假峰,得到每个视频的每秒平均镜头切换次数,以此作为广告视频和非广告视频的分类依据,进行C-均值聚类。实验结果表明,该方法以较少的工作量获得了较好的分类结果。

关 键 词:自动视频分类  直方图差分  每秒平均镜头切换次数  C-均值聚类
文章编号:1671-3559(2007)02-0100-04
修稿时间:2006-09-01

Video Classification Method Based on Histogram Difference
MENG Li,XU Fa-sheng,LI Jin-ping. Video Classification Method Based on Histogram Difference[J]. Journal of Jinan University(Science & Technology), 2007, 21(2): 100-103
Authors:MENG Li  XU Fa-sheng  LI Jin-ping
Abstract:Seeing that most of the traditional automatic video classification algorithms are complex and always waste lots of time,we proposed a simple video classification approach of histogram difference statistical method.In this plan,the false peaks caused by gradual shot change in the histogram difference curves have been removed.It is very convenient to find out those advertising videos from others according to the shot change times per second by C-Means clustering method.It has been proved by our practice that this method gains great benefit while saves tremendous time.
Keywords:automatic video classification  histogram difference  shot change times per second  C-Means clustering
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号