| 摘要: |
| 提出一种播音员镜头自动检测算法.利用播音员镜头频繁出现的统计规律,基于自动聚类方法找出候选播音员镜头,然后根据播音员镜头出现的时空特征,用神经网络分类器对候选播音员镜头进行确认,从而实现对播音员镜头的检测.该算法是自动分析电视新闻节目内容的重要手段之一,对建立视频数据库索引是必不可少的.实验结果表明,该方法具有准确、快速的特点,可有效地应用于视频检索系统. |
| 关键词: 视频信息检索 多媒体数据库 视频内容分析 播音员镜头检测 视频数据库索引 |
| DOI： |
| 分类号: |
| 基金项目:国家自然科学基金资助项目(60003017,69935010);国家高技术研究发展计划资助项目(2001AA114120);上海市科技启明星计划资助项目(01QD14013);上海市科技发展基金项目(015115044) |
|
| A Method to Detect Anchorperson Shots for Digital TV News |
|
YANG Na,LUO Hang-zai,XUE Xiang-yang
|
| Abstract: |
| A new method is presented to detect anchorperson shots automatically for digital TV news programs. The algorithms are required for automatically parsing and indexing of news program. The identification process consists of two steps. The first step is that potential anchorperson shots are grouped by clustering based on the fact that anchorperson shots always present repeatedly within a news program. At the second step, a well-trained neural network classifier is used to identify the correct anchorperson shots according to temporal and spatial features of their occurrence law.Exprimental results show that the proposed method can detect the anchor shots quickl with very high precision. |
| Key words: video information retrieval multimedia database video content parsing anchorperson shot detection video database indexing |
