Integrated color-spatial approach to content-based image retrieval
Citations
Altmetric:
Alternative Title
Abstract
The use of color information for image retrieval has been used widely in many content-based retrieval system with some success. However, histogram-based color retrieval techniques suffer from a lack of important spatial knowledge. We discuss a technique of integrating color information with spatial knowledge to obtain an overall impression of the image. The technique involves three steps: the selection of a set of representative colors, the analysis of spatial information of the selected colors, and the retrieval process based on the integrated color-spatial information. Two color histograms are used to aid in the process of color selection. After deriving the set of representative colors, spatial knowledge of the selected colors is obtained using a maximum entropy discretization with event covering method. A retrieval process is formulated to make use of the spatial knowledge to retrieve relevant images. A prototype image retrieval system has been implemented on the Unix system. It is tested on an image database consisting of 260 images. The result shows substantial improvement over the histogram-based color retrieval methods.
Keywords
Source Title
Proceedings of the ACM International Multimedia Conference & Exhibition
Publisher
Series/Report No.
Collections
Rights
Date
1995
DOI
Type
Conference Paper