|Clustering Methods for Automatic Summarization of Video Sequences
|Dipl.-Ing. Martin Haller
|DR.-Ing. Thomas Sikora
|The fundamental problem to assess the performance of clustering algorithms on visual similarity for high dimensional data set is discussed. The Main motivation of the thesis is not only to assess the performance of clustering algorithms on visual similarity but also to find the best distance measures used for clustering data. The clustering algorithms are K-means, spectral clustering and affinity propagation. In this thesis an effort has been made to assess the performance of these clustering algorithms.
|partition-clustering PC, partition-distance PD, distance-clustering DC, validity, clustering algorithms, matlab