摘要
提出一种基于时空变化信息的视频内容检索方法.此方法以自适应变间隔关键帧选择策略提取镜头在时间域上的变化内容,采用时空注意力模型提取空域显著内容;然后对每一显著区域按Mpeg-7标准抽取相似纹理描述子、可扩展颜色描述子和基于轮廓的形状描述子,联合三低层视觉特征进行显著区域间的匹配度计算;最后提出用于视频检索的两镜头相似度匹配算法.对比实验表明该方法能有效进行基于内容的视频检索.
A video content retrieval method based on spatio-temporal change information is presented.A variable interval key-frame selection strategy with self-adaptability is used to extract the temporal change content of the shot,and the spatial salient content is extracted with a spatio-temporal attention model.For each salient area in the key-frame,scalable color descriptor,homogenous texture descriptor and contour shape descriptor are extracted according to Mpeg-7 standard,and the match degree of two salient areas is calculated using the combined value of three low-level visual descriptors.Finally,an algorithm is proposed to calculate the match degree between two shots.Experiment is made to compare the proposed algorithm with other methods,and the results show that the presented algorithm can effectively retrieve the desired videos according to their contents.
出处
《信息与控制》
CSCD
北大核心
2007年第5期578-584,591,共8页
Information and Control
基金
国家自然科学基金资助项目(60273035)
江苏省科技攻关计划资助项目(BE2003064)
成都信息工程学院发展基金资助项目(KYTZ20060904)
四川省教育厅青年基金资助项目(2006B063)
关键词
时空注意力模型
基于内容的视频检索
关键帧选择策略
时空信息
spatio-temporal attention model
content-based video retrieval
key-frame selection strategy
spatio-temporal information