Algorithms for clustering data
Algorithms for clustering data
ACM Computing Surveys (CSUR)
An empirical comparison of four initialization methods for the K-Means algorithm
Pattern Recognition Letters
On clustering and retrieval of video shots
MULTIMEDIA '01 Proceedings of the ninth ACM international conference on Multimedia
X-means: Extending K-means with Efficient Estimation of the Number of Clusters
ICML '00 Proceedings of the Seventeenth International Conference on Machine Learning
Towards effective indexing for very large video sequence database
Proceedings of the 2005 ACM SIGMOD international conference on Management of data
Digital Image Processing (3rd Edition)
Digital Image Processing (3rd Edition)
Extracting story units from long programs for video browsing and navigation
ICMCS '96 Proceedings of the 1996 International Conference on Multimedia Computing and Systems
An integrated scheme for automated video abstraction based on unsupervised cluster-validity analysis
IEEE Transactions on Circuits and Systems for Video Technology
| Hi-index | 0.01 |
Video shot clustering is the basis of other high-level research of multimedia databases applications. This article proposes a novel and efficient shot clustering algorithm for videos by applying the multi-resolution analysis of Haar wavelets which is called MLHC(Multi-Level Hierarchical Clustering). Corresponding to the reconstruction procedures of Haar wavelets, MLHC is designed as a multi-level algorithm. When the algorithm runs to further levels, the clustering results are increasingly credible and precise. After the clustering results achieve a stable status, MLHC stops automatically. Thus it's an iterative incremental clustering algorithm. Each level of MLHC is an independent hierarchical clustering algorithm which resolves the dilemma of choosing proper initial cluster centers for most existing shot clustering algorithms. For each hierarchical level of MLHC, a novel stop criterion is designed to stop the iterative merging procedures and terminates MLHC on this level. By this stop criterion, the clustering results can be obtained automatically without any parameters and the number of clusters can also be estimated at the same time. The theoretical analysis and the extensive experiments witness the efficiency and effectiveness of our proposals.