摘要
文章通过分析武术视频的特点,提出了一种关于提取武术分段视频的关键帧的算法:图像颜色特征计算法与Ⅰ帧提取法相结合的方法,经仿真实验证明此算法对于武术分段视频的关键帧提取准确率高。在此算法的基础上,对关键帧的运动人体进行特征向量的概括。然后,对支持向量机进行了分析并利用得出的特征向量作为样本对分类器进行构造和训练。基于以上步骤设计的系统,可有效的实现对武术视频的分类检索。
It can be analysised wushu video＇s characteristic in this article,A segment on the extraction of wushu video keyframes algorithm：Image color characteristics calculation method and the Ⅰ-frame extraction method.After the simulation experiment proved that this algorithm is high regarding the wushu partition video frequency＇s essential frame extraction rate of accuracy.Based on this algorithm,the keyframes of human movement for eigenvector overview.On support vector machines for analysis and utilization of characteristic vectors as samples on the classifier for structure and training.Based on above step design system,effectively achieve the wushuvideo retrieval.
出处
《微计算机信息》
2011年第1期274-275,290,共3页
Control & Automation
关键词
关键帧
图像颜色特征
Ⅰ帧
特征向量
支持向量机
Key frame
Image color characteristics
Ⅰ-frame
characteristic vectors
SVM
