摘要
针对音乐舞蹈视频动作复杂,冗余动作多的问题,提出了一种提取舞蹈视频关键帧的方法。对舞蹈视频中相邻的两个图像帧进行光流计算,计算每张光流图的熵并提取舞蹈视频中音乐的特征,将音乐特征与光流图的熵值序列进行融合,得到一个与音乐相匹配的熵值序列,通过阈值比较得到舞蹈视频的关键帧集合。实验结果表明:该算法能够有效地提取出冗余少且概括视频内容的关键帧,对舞蹈视频分析和舞蹈动作编排等方面起到了重要的辅助作用。
Aiming at the complicated and redundant action of music and dance video, a method of extracting dance video key frame is proposed. The optical flow of two adjacent image frames in the dance video is calculated. The entropy of each optical flow graph is calculated and the characteristics of the music in the dance video are extracted. The music characteristics are merged with the entropy sequence of the optical flow graph and a sequence of entropy matching with music is obtained. The key frame of dance video is achieved by threshold comparison. The experimental results show that the algorithm can effectively extract the key frames with less redundancy and summary of the video content, and plays an important role in the analysis of dance video and choreography.
作者
马楠
石祥滨
代钦
刘翠微
刘芳
Ma Nan;Shi Xiangbin;Dai Qin;Liu cuiwei;Liu Fang(College of Computer Shenyang Aerospace University, Shenyang 110136, China;College of Information Liaoning University, Shenyang 110036, China;College of Information, Shenyang Institute of Engineering, Shenyang 110136, China)
出处
《系统仿真学报》
CAS
CSCD
北大核心
2018年第7期2801-2807,共7页
Journal of System Simulation
基金
国家自然科学基金(61170185
61602320)
辽宁省博士启动基金(201601172)
辽宁省教育厅一般项目(L201607
L2014070)
关键词
关键帧
光流图
图像熵
音乐特征
特征融合
key frame
optical flow graph
image entropy
musical characteristics
feature fusion
