N.G. Zagoruiko, S.E. Gulyaevsky, V.V. Dubanov, and O.A. Kutnenko (Russia)
Images description, nearest analogue, fast retrieval, col lective decision rule
Description methods of black-and white micro-photos and fast ways of the nearest analogue retrieval to the given photo in the large scope data base are presented. Two leveled collective decision rule is used while making decision. Tests results for described methods on the model and real data are given. Collective decision rule provides high recognition reliability. "Locator" algorithm allows effective pretenders' number to the role of the nearest analogue reduction. For large databases this algo rithm retrieval time is lower then the full overrun search time on 2-3 orders.
Important Links:
Go Back
