Prof. Dr. Eric Paquet

Biography:
Eric Paquet is a senior research officer at the Visual Information Technology Group of the National Research Council of Canada. He received his Ph.D. in Computer Vision from Laval University and the National Research Council in 1994. After finishing his Ph.D., he worked on optical information processing in Spain, on laser microscopy at the Technion-Israel Institute of Technology and on 3D hand held scanners in England. He is currently pursuing research on content-based indexing and retrieval of 2D images and 3D objects as well as on visualisation at the National Research Council of Canada. His current research interests include content-based description of multimedia and multi-dimensional objects, anthrometric databases, structural bioinformatics and cultural heritage applications. He is the author of more than 100 publications, member of MPEG, WEAR, CAESAR, ISPRS, SCC, CODATA, on the programme committee of several international conferences and owns many patents. He has received many international distinctions including the PKDD Innovative Application Award as well as two Recognition Awards from SAE.

Talk Title: Indexing and Retrieval of Content for Cultural Heritage, Anthropometry and Bioinformatics

Abstract: We present a content-based approach for information indexing and retrieval of both three-dimensional objects and bidimensional images based on their three-dimensional shape and on their composition respectively. Such a visual approach to information retrieval is intuitive and follows a semantic typical of art and architecture without any invasive interface since the object under consideration acts as its own interface. We report progress in three different fields: cultural heritage, anthropometry and bioinformatics. Furthermore, we survey one of the first attempts to standardize content-based retrieval: MPEG-7. In the case of cultural heritage, we focus our attention on museum databases and reconstruction of broken frescoes. Most of the results come from the C2RMF database (Centre of Research and Restoration of the Museums of France at the Louvre Museum) which contains the digital archives of all museums in France (more than 120.000 paintings) including museums such as the Louvre, Orsay and Cluny just to mention a few. We address the issue of finding similar or related painting by style, period and technique which is of interest for both scholars and the general public. Finally, we address the reconstruction of broken frescoes made out of 100.000 fragments or more. In the case of anthropometry, we analyze the CAESAR database which contains three-dimensional scans, anthropometric and demographic data of thousands of individuals in the USA, Canada, Netherlands and Italy. We present Cleopatra (two SAE awards): a content-based search engine that has been designed specifically for this database. We focus our attention on the interplay in between content-based retrieval, anthropometric and demographic data. Two applications are considered: calibration of virtual mannequins and design clothes that fit (PKKK Application Award). Both applications make use of data mining techniques. Finally we address the bioinformatics issue of structural protein retrieval which is of primary importance for the pharmaceutical industry given that their structure is related to their functionality. We present results obtains with the Protein Data Bank (more than 25.000 proteins) The emphasis of the presentation will be on practical issues of real world problems.