Image Mining

mage mining and information extraction from multi-source data

Objectives

Image mining is the process of searching and discovering valuable information and knowledge in large volumes of data. Image mining draws basic principles from concepts in databases, machine learning, statistics, pattern recognition and 'soft' computing. Using data mining techniques enables a more efficient use of data banks of earth observation data.
It is thus becoming an emerging research field in geosciences because of the increasing amount of data which lead to new promising applications. For example, the use of very high resolution satellite images now enables the observation of small objects, while the use of very high temporal resolution images enables monitoring of changes at high frequency.
However, actual data analysis techniques suffer from the huge amount of complex data to process. Indeed, earth observation data (acquired from optical, radar and hyperspectral sensors installed on terrestrial, airborne or spaceborne platforms) is often heterogeneous, multi-scale, incomplete, and composed of complex objects. Segmentation algorithms, unsupervised and supervised classification methods, descriptive and predictive spatial models and algorithms for large time series analysis will be presented to assist experts in their knowledge discovery.
The thematic school, dedicated to post-graduate students and young researchers, will present state-of-the-art image mining techniques by combining both theoretical lectures and practical exercises on specific datasets. Open source algorithms, codes and softwares will be presented.

Working language and documents

All lectures and practicals will be given in English. The documents given at the beginning of the lessons will also be written in English. The course will use Open Source image processing softwares (ORFEO/Monteverdi toolbox, ROI-PAC, Mustic).
Important: The participants are asked to bring with them their laptop on which all necessary softwares will be installed at the beginning of the course.