Abstract:
This paper proposes video clips clustering method based on spectral of correlative graph. It constructs the correlative graph of the video clips, changes it to an adjacency matrix, and extracts the spectral characteristics of adjacency matrix which includes the leading eigenvalues, inter-mode adjacency matrices and inter-mode edge-distance. These vectors are embedded into the pattern space by using Principal Component Analysis(PCA) and Independent Component Analysis(ICA), and k-means method is used for clustering analysis. Experimental results show that the method is effective in distinguishing different types of video clips.
Key words:
spectral of correlative graph Principal Component Analysis(PCA),
Independent Component Analysis(ICA),
spectral characteristic,
pattern space,
k-means clustering
摘要: 提出一种基于关联图谱的视频片段聚类方法。构造视频片段的关联图并将其转换成邻接矩阵,提取邻接矩阵的主分量特征值、模间邻接矩阵和模间距离后,将三者分别嵌入主成分分析和独立成分分析模式空间中,利用k-means进行聚类分析。实验结果表明,该方法能有效区分不同类型的视频片段。
关键词:
关联图谱,
主成分分析,
独立成分分析,
谱特征,
模式空间,
k-means聚类
CLC Number:
TUN Yong-Long, FU Mao-Qing, LUO Bin. Video Clips Clustering Based on Spectral of Correlative Graph[J]. Computer Engineering, 2011, 37(18): 281-283.
吴永龙, 符茂胜, 罗斌. 基于关联图谱的视频片段聚类[J]. 计算机工程, 2011, 37(18): 281-283.