Personal tools
Home Pós-Graduação Informações para Alunos e Docentes do Programa Seminários no IC-UNICAMP Seminários de Pesquisa do IC Palestra Extraordinária: Content-based image retrieval.
Navigation
IC 40 anos
 
Document Actions

Palestra Extraordinária: Content-based image retrieval.

Prof. Sylvie Philipp-Foliguet, ETIS, Cergy, France, na Série de Seminários 2009 da Pós-Graduação, dia 13/04/2009, às 14:00 h, Auditório do IC , Sala 85 - IC 2.

What Palestra
When 13/04/2009
from 14:00 to 15:00
Where Auditório do IC - Sala 85 - IC 2
Add event to calendar vCal
iCal

The first stage of image retrieval is to find an appropriate description of the visual content. An image can be represented by a global signature (colour and texture histogram) or by local descriptors around characteristic points or regions of the image. In the later case, spatial positions or relationships have also to be stored.

The second stage is to define similarity measures adapted to the kind of representation. I will present kernel functions adapted to vectors, bags of features or graphs and k- nearest neighbour search.

The third stage of a retrieval system is the search engine, which is usually based on an interactive learning scheme, using classifiers such as Support Vector Machine.

Applications concern retrieval of images containing a specific object or of image categories, copy detection and 3D artwork object retrieval.

=============================================

Sylvie Philipp-Foliguet is Professor at the National School of Electronics (ENSEA) of Cergy-Pontoise, France, since 1988. She manages the MIDI (Multimedia Indexing and Data Integration) team of laboratory ETIS (Information Processing and Systems).

Her research domains are image segmentation and interpretation. Concerning the first topic she developed a fuzzy segmentation method. For the second topic she proposed methods using fuzzy regions, inexact graph matching and statistical learning. Applications concern indexing and retrieval of documents from databases, these documents are images, videos or 3D objects.


Instituto de Computação :: Universidade Estadual de Campinas
Av. Albert Einstein, 1251 - Cidade Universitária • CEP 13083-852 • Campinas/SP - Brasil • Fone: [19] 3521-5838