A New Content-Based Image Indexing and Retrieval System of Mosaic Images

Development and practice of content-based image retrieval (CBIR) systems becomes frequent during last decade, in favour of surfers and specialists. However, no recognised system have been dedicated to archaeological purposes, where archeologists try to identify mosaic images with similar content, or need to recognize one or more objects in those images. In this contribution, we propose a new CBIR of mosaic images system, using four description methods. To allow partial queries, we added an object-extracting subsystem from images, using a statistical segmentation combined with some morphological operators. These objects constitute a new images´ database, and define a new representation space. Retrieval is perform using two similarity metrics induced on the first and the second spaces