Databases and Images Group

The Databases and Images Group researchs databases and image processing, developing complex data types in DBMS. Among the main recent results are:

• the development of one of the most efficient metric access methods: the Slim-tree, the DF-tree and the OMNI family;
• the use of the Fractal Theory for the development of precise techniques for the selectivity estimation and costs for query optmization in DBMS;
• the development of data mining techniques, information visualization techniques and query representation by similarity in image data sets;
• the development of image feature extraction techniques and similarity measures among them.

These activities turned possible the image support in commercial DBMS. A particular area of interest for these results is the support for similarity queries in medical image exams data sets. In cooperation with the Hospital das Clínicas de Ribeirão Preto-USP, the first experimental results obtained in prototypes running at the hospital are taking the research results directly to the community, in the form of a better support of activities like computer aided diagnostics.