摘要
提出了一种基于底层特征和基于高级语义特征的视频镜头分类方法,使用RBF核的支持向量机(SVM)作为分类器,分别将其应用于动漫/真人和足球比赛视频的镜头分类,前者的平均错误概率控制在了7.43%之内,而基于高级语义特征的足球比赛镜头分类的准确率达到了84%。
In this paper,two new methods for the classification of video shots based on low-level feature and high-level semantic feature has been proposed respectively.Using SVM with RBF kernel function as the classifier,the former method is applied to the classification of video shots of cartoon/real person and the latter is applied to the video shots of soccer matches.The former has achieved an average error rate less than 7.43%,while the classification of soccer matches video shots based on high-level semantic feature has a satisfied precision up to 84%.
出处
《计算机应用与软件》
CSCD
2010年第7期230-232,共3页
Computer Applications and Software
关键词
视频分类
底层特征提取
语义特征提取
Video classification Low-level feature extraction Semantic feature extraction
