Towards a new approach for an effective image retrieval based on Oracle DBMS

Facing the increasing need for modeling and implementing Content-Based Image Retrieval (CBIR) task, several search systems have been developed such as ImageFinder, Img(Anaktisi) and MedFMI-SiR. In this search area, the development of accurate and flexible CBIR applications is necessary to provide pertinent retrieval result with transparency to the end user. However, the relevance of the images search depends directly upon the choice of similarity criteria (texture, color, shape) that can give irrelevant results in case of non-trivial choice of these parameters. The goal of our approach is to provide an assistance tool for our image retrieval system that can guide the user to an appropriate choice of similarity criteria. We present an experimental evaluation that show that our CBIR system is a relevant and an optimized solution, being able to help the user to correctly model his query by selecting the appropriate search criteria over an Oracle database with 600 images.