Signal Images Physique
Team manager : Jocelyn CHANUSSOTCornel IOANA

Research Areas

This axis involves developing advanced methods for signal and image processing for data issued from remote sensing, optics and/or radar, aerial or satellite. The recent techniques for processing and analysis of very high resolution data (spatial and spectral) are of a particular interest for our team. An unifying theme concerns the multivariate character of the data and its correct handling. We will often mention the multi dimensionnal processing used in environment monitoring and comprehension using remote sensing.

The implications of our work lead to numerous applications:
- regular measurements of geophysical objects movements/deformations;
- soil type recognition using terrain zone segmentation;
- implementation of detectors adapted to zones with non-homogenous textures;
- characterization of the interaction between electromagnetic waves and retro scattering environment;
- surface/volume mapping of objects.

. Passive remote sensing

The team's activities in this field are concentrated in the following fields:

- hyperspectral imaging : The investigated subjects concern the supervised high volume data classification, dimension prediction, source separation (linear or non-linear), physical integration of mixtures,  super-resolution, developing of kernel based methods;
- high resolution optical imaging: the performance of recent high spatial resolution satellites (IKONOS, QUICKBIRD, PLEIADES) open the possibility for a very fine analysis provided that algorithms for processing the rich information are developed. A study for the seismic risk handling in urban areas integrating the optical remote sensing and numeric model data fusion is under proposal (ANR).
- data fusion: fusion of images with different resolutions, multimodal data fusion (optical and lidar, for example), decision fusion (local and global reliability estimation for different algorithms in order to improve decision).

. Active remote sensing

The research for SAR image processing are guided by two theoretical and applicative perspectives:
- The first axis targets the methodology aspects of the quantitative remote sensing in the context of polarimetric (POL) or interfrometric (In) SAR images.
- The second axis studies the technological potential and limits of the POL/In/POL-InSAR imaging for the tracking of complex geophysical objects such as glaciers on the test site of "Chamonix Mont-Blanc".

These actions initiated in 2004 (ACI-MEGATOR (2004-07) and ANR-EFIDIR (2008-12)) join 6 French complementary teams. This gathering of radar specialists, unique in Europe, provides France a privileged position.

The research efforts are directed also towards the SAR clutter, coupled with the study of radar waves diffusion on the scanned surfaces provided by the latest high spatial resolution radar satellites. The applications target the monitoring of levees, snow, névé and ice in order to extract the parameters of snowfall, thus improving the foreseeing of avalanche risks and also to detect the moving snow mass, predicting landslides associated with the thaw of permafrost.

GIPSA-lab, 11 rue des Mathématiques, Grenoble Campus BP46, F-38402 SAINT MARTIN D'HERES CEDEX - 33 (0)4 76 82 71 31