In the past few years, urban areas have considerably grown and the major part of the population is now living in big cities all around the world. Management and control of urban areas requires appropriate monitoring tools. These tools should be able to give global and accurate measures on urban areas. SAR (Synthetic Aperture Radar) imagery has many advantages compared to other sensors. It provides information on larger areas and at a lower cost compared to LIDAR data, and is independent of lighting conditions, on the contrary of optic sensors. Besides, the use of polarimetric or multi-channel data provides information on different properties of the observed areas. Nevertheless, the usual SAR data are of limited use in complex urban areas because of geometric distorsions and speckle noise. This project focuses on SAR tomography, which is an original and new approach to analyze urban areas. By combining multiple images, this technique allows to discriminate different scatterers inside a vertical cell. This project aims at developing new tomographic methods dedicated to urban areas. It is built with innovation at all steps of the processing chain: data acquisition, tomographic processing, information extraction.
This project is decomposed in three main tasks. The first one is the constitution of a tomographic database, both with aerial high resolution (HR) data acquired by SETHI of ONERA, and satellite TerraSAR-X images. The data processing of aerial images is particularly complex for HR data and will lead to innovative SAR processing methods. This database will be completed with TerraSAR-X data to obtain a unique data set with different spatial and polarimetric resolutions. The second task of the project is dedicated to the development of new tomographic methods dealing with the specific problems of data in urban areas: separation between man-made objects and vegetation, use of data acquired with different polarization modes and temporal lags, introduction of structural information to improve the reliability and resolution of estimators. The third task is the development of new applications based on the obtained measures for urban planning and change detection or activity tracking. It will be based on new regularization models dedicated to tomographic data, on the fusion with optic image, and on simulated images derived from 3D tomographic measures. Eventually, the extension of these techniques to data with degraded resolution will be useful to assess performances of future satellite systems and will thus be evaluated on TerraSAR-X images.
This project is built on the complementarity of the three teams which control the whole chain of competences, from the data acquisition step to the tomographic processing and the exploitation of results to derive new applications of urban area management and activity tracking. The quality, experience and complementarity of the consortium should lead to the development of very competitive methods for SAR tomography.