Satellite Image Processing Facing Big Data Revolution
Satellite image processing is facing a big data revolution. The amount of data to be processed every day is indeed increasing tremendously. As a consequence, robust processing techniques are required. In IRT (Technological Research Institute) of Toulouse we are targeting applications involving large scale analysis. I will present processing techniques and tools regarding three topics of interest. The first topic is cloud cover annotation. We have a 200 000 satellite image database, where for each image clouds have been segmented by operators. We will discuss performances and scalability of various classifiers in this context. I will also present first results in automatic change detection, and in oil spit detection from radar images. My presentation will focus on technologies and tools amenable to large scale processing (eg. google cloud, Apache Spark, python, etc…) in our image processing context.
Mathias Ortner received his B.Sc. degree in 2001 in aeronautics and space sciences from the French School of Aeronautics and Space and his Master of the Science in applied mathematics the same year from the University of Toulouse, France. He prepared his Ph.D. at INRIA-SAM in Ariana research group from 2001 to 2003. He received the Ph.D. degree in 2004 in computer science and image processing from the University of Nice Sophia-Antipolis, France. From 2004 to 2006 he was a post-doctoral research visitor in Prof. Nehorai’s team, first in University of Illinois at Chicago, USA, then in Washington University in Saint Louis, USA. He is now working as a research engineer for Airbus Defence and Space, in Toulouse. His work interests are primarely stochastic methods, inverse problems and machine learning.