摘要
为克服互联网中视频信息传统检索方法的不足,保证检索的准确率和高效性,针对镜头代表性强弱的问题,提出镜头鉴别力概念以及计算鉴别力大小的方法,并基于镜头鉴别力设计一种新的视频检索方法。该方法提取镜头静态特征和动态特征作为计算鉴别力大小的依据。实验结果证明,使用镜头鉴别力分析后的检索效果好于未使用镜头鉴别力分析的检索效果,在实验环境中达到94%的准确率。
With the development of video resources, traditional video retrieval systems cannot achieve satisfactory results. Therefore, Content-based Video Retrieval（CBVR） becomes a hot research field. Video shots have different representativeness, some shots can represent the content of videos effectively while others cannot. In order to overcome the disadvantage of traditional video retrieval methods and ensure the retrieval nccuracy and efficiency, this paper proposes the concept of shot discrimination and the calculation method based on shot representativeness. A new video retrieval method is designed based on shot discrimination analysis and this method extracts both shot static feature and motion feature to calculate shot discrimination. Experimental results demonstrate that this method outperforms other ones and can achieve the accuracy of 94%.
出处
《计算机工程》
CAS
CSCD
2012年第14期223-226,共4页
Computer Engineering
基金
上海市科学技术委员会基金资助项目(09511501402)
关键词
视频检索
鉴别力
静态特征
运动特征
最短距离权重
软性判决
video retrieval
discrimination
static feature
motion feature
shortest distance weight
soft judgment
作者简介
殷杰(1987-),男,硕士,主研方向:图像视频分析与处理,信息安全,E-mail:superyin87531@hotmail.com;王士林,副教授;李建华,教授、博士生导师
