Download now Free registration required
Clustering is used in content-based multimedia retrieval to reduce the size of the search space and make user navigation through the database easier. In retrieval by example systems, clusters are selected and presented to the user before starting a search in order to give him an overview of the database contents. The discrete wavelet transform with specific mother functions is widely adopted to extract features to perform clustering of images in a given database. The main purpose of this paper is to investigate the incidence of mother wavelet and frequency channel features on the obtained clustering. The results show that clustering trees vary largely with mother wavelet function and sub-bands.
- Format: PDF
- Size: 273 KB