Color Mining of Images Based on Clustering ⋆

The increasing size of multimedia databases and the ease of accessing them by a large number of users through the Internet carries a problem of efficient and semantically adequate querying of such content. A metadatabase may be used to shorten query resolution time by trying to limit the number of images being thoroughly analyzed to a smaller subset, having a high probability of finding the query image. In the article we propose a simple but fast and effective method of indexing such image metadatabases. The index is created by describing the images according to their color characteristics, with compact feature vectors, that represent typical color distributions. We present experiment results of typical search schemes by querying the metadatabase index created using a few different approaches.