摘要: 阐述一种新颖的新闻视频结构化浏览和标注系统。应用基于时空切片分析的新闻主播检测方法和基于颜色直方图的镜头分割方法实现新闻视频的结构化。通过自动语音识别技术和特定语义概念模型的建立实现了对主播场景的文本信息标注和对新闻故事镜头的语义概念标注。该系统有利于用户根据个人爱好进行新闻视频的浏览和编辑,有效实现新闻视频的索引和浏览。
关键词:
新闻,
时空切片,
语义概念,
自动语音识别
Abstract: This paper presents an innovative hierarchically browsing and annotating system for news video. With the anchorperson detection algorithm based on Spatio-Temporal Slice(STS) analysis and the shot boundary detection method based on color histogram, structuralization on video is implemented. With Automatic Speech Recognition(ASR) technique and semantic concepts models, the anchorperson scenes with text information are respectively annotated and the news story scenes with visual semantic concepts are labeled. Therefore, the system can realize the interest-oriented browse for viewers and facilitate video edit for editors by providing effective indexing and browsing method for news video.
Key words:
news,
Spatio-Temporal Slice(STS),
semantic concept,
Automatic Speech Recognition(ASR)
中图分类号:
刘安安;李锦涛;张勇东;唐 胜;杨兆选;吴佳鹏. 新闻视频结构化浏览与标注系统[J]. 计算机工程, 2009, 35(1): 33-35.
LIU An-an; LI Jin-tao; ZHANG Yong-dong; TANG Sheng; YANG Zhao-xuan; WU Jia-peng. Hierarchically Browse and Annotation System for News Video[J]. Computer Engineering, 2009, 35(1): 33-35.