Adaptable Similarity Search in Large Databases
Similarity search is a highly application dependent and even subjective task. Similarity models are desired to be adaptable to application specific requirements or individual user preferences. In our work, we focus on two aspects of adaptable similarity search:
(1) Adaptable Similarity Models. Examples include pixel-based shape similarity, 2D and 3D shape histograms, and slice-oriented similarity models. These models were applied to multimedia, biomolecular, and medical image databases.
(2) Efficient Similarity Query Processing. Similarity models based on quadratic forms result in ellipsoid queries on high-dimensional data spaces. We present algorithms to efficiently process ellipsoid queries on index structures, and to improve the performance by introducing several approximation techniques while guaranteeing no false dismissals for similarity range queries and k-nearest neighbor queries.
|Authors:||Seidl T., Kriegel H.-P.|
|Published in:||In: Burkhardt H., Kriegel H.-P., Veltkamp R. (eds.): Content-Based Image and Video Retrieval. Dagstuhl Seminar Report 261|
|Forschungsgebiet:||Exploration of Multimedia Databases|