Computer Engineering and Applications ›› 2012, Vol. 48 ›› Issue (35): 139-142.

Previous Articles     Next Articles

Characteristics extraction and classification based on periodically moving object

WEI Wei, GAO Weina, LIU Ming   

  1. College of Computer Science and Software, Hebei University of Technology, Tianjin 300401, China
  • Online:2012-12-11 Published:2012-12-21

周期性运动物体的特征提取及其分类

魏  玮,高维娜,刘  命   

  1. 河北工业大学 计算机科学与软件学院,天津 300401

Abstract: Two kinds of characterisitics of the periodically moving object are proposed in video. The first one is the frequency by Fourier transform of the periodic motion video. The second is a similarity measure from a walking template-based as human motion characteristics. Through the support vector machine, six behavior characteristics of laboratory tests are classified. The results show that these characteristics can be fully expressed in the video acts, and have a higher classification, this method is also simple and fast.

Key words: external rectangular aspect ratio, keyframe, frequency, similarity

摘要: 提出视频中周期性运动物体两种特征。第一种是通过傅里叶变换得到视频中周期性运动的频率。第二种是采用了一种基于行走模版的相似性度量作为人体运动的特征。通过支持向量机对六种行为特征进行分类实验检测,结果表明上述特征能够充分地表示视频中的行为,得到了较高的分类效果,而且该方法、简单快速。

关键词: 外接矩形长宽比, 关键帧, 频率, 相似度