Menu:

PhD Thesis in Signal and Image Processing

research grant awarded by the CNES (French Spatial Agency)

University of Toulouse - Institut National Polytechnique de Toulouse (INPT) - ENSEEIHT
Research laboratory : Institut de Recherche en Informatique de Toulouse (IRIT)
Team: Signal and Communication (SC) - Theme 1: Information Analysis and Synthesis
Advisors: Marie Chabert (Associate Professor) marie.chabert@enseeiht.fr and Jean-Yves Tourneret (Professor) Jean-Yves.Tourneret@enseeiht.fr

TITLE Analysis of remote sensing multi-sensor heterogeneous images

DESCRIPTION

This PhD thesis aims at evaluating the interest of using multivariate distributions for the analysis of heterogeneous images. The considered heterogeneous data are composed of images acquired by different sensors (including optical, radar and hyperspectral sensors) and possibly of an object database (containing roads, building, …). The applications considered in this thesis are mainly image registration, change detection and database updating. All these applications require to define a similarity measure between the different images or between features estimated from these images (called modalities). 

Various multivariate statistical models can be used to describe the statistical properties of the image modalities and to derive appropriate similarity measures. These models can be parametric (e.g., see the references [C07], [CTFI07], [CT11], [DV02], [CTPI10]) or non parametric ([MMS08], [BLLP10], [BP05]). However, it is not very clear to determine whether these statistical models provide more relevant similarity measures than  empirical methods derived without using a priori knowledge on the measured data (e.g., based on empirical mutual information or empirical Kullback-Leibler divergence). In particular, after selecting an appropriate multivariate statistical model, statistical methods have to be investigated for model selection and parameter estimation. These methods require reliable a priori knowledge and a sufficient spatial homogeneity of the data. By “a priori knowledge”, we mean the data extracted from the image and the resulting statistical model (there is a vast literature about multivariate distributions such as gamma, generalized Gaussian, Fisher, Pearson that have nice properties for remote sensing images).  Regarding “spatial Homogeneity”, it is well known that any estimation method requires a sufficient number of samples extracted from a homogeneous area to produce good estimates and (as a result) relevant similarity measures.

The main objective of this PhD thesis is to put in balance the complexity increase and the performance improvement of statistical methods compared to empirical ones, taking into account practical constraints (spatial homogeneity and required number of samples). The performance will be evaluated for various decision tasks (image registration, change detection, database updating) in the case of heterogeneous data acquired from different sensors (optical and/or radar) or from decision systems (database). 

The Signal and Communication team of IRIT has been studying related problems during the last years. Statistical models for remote sensing images have been investigated in three different theses:

·         Thesis of Florent Chatelain devoted to the study of multivariate gamma distributions for SAR images

·         Thesis of Vincent Poulain devoted to the problem of database updating

·         Thesis of Reza Shirvany devoted to the analysis of polarimetric images

The candidates are invited to consult the homepages of Jean-Yves Tourneret (http://www.enseeiht.fr/~tourneret) and Marie Chabert (http://www.enseeiht.fr/~chabert) for more informations including a detailed list of publications.

Required skills:  signal and image processing, probability and statistics 

Bibliographie

[OQ98] C. Oliver and S. Quegan, Understanding Aperture Radar Images, Artech House, 1998.
[I03] J. Inglada, “Change detection on SAR images by using a parametric estimation of the Kullback-Leibler divergence,” In Proc. IEEE Int. Conf. Geosci. and Remote Sensing (IGARSS), pages 4104–4106, Toulouse, France, July 2003.
[GNSRTR09] F. Galland, J.-M. Nicolas, H. Sportouche, M. Roche, F. Tupin, and P. Réfrégier, “Unsupervised Synthetic Aperture Radar Image Segmentation Using Fisher Distributions,” IEEE Trans. on Geosci. and Remote Sensing, vol. 47, no. 8, August 2009.
[T10] F. Tupin  “Fusion of optical and SAR observations,” dans Multivariate Image Processing, C. Collet, J. Chanussot, and K. Chehdi (Eds), Wiley, New York, 2010.
[C07] F. Chatelain, Lois Gamma multivariées pour le traitement d’images radar, Thèse de l’INPT, n°ordre 2525, 25 Octobre 2007.
[CTFI07] F. Chatelain, J-Y. Tourneret, A. Ferrari, and J. Inglada, “Bivariate gamma distributions for images registration and change detection,” IEEE Trans. on Image Processing, 16(7):1796–1806, July 2007.
[DGH97] Y. Delignon, R. Garello, and A. Hillion, Statistical modeling of ocean SAR images, IEE Proc. on Radar, Sonar and Navig., 44(66):348–354, 1997.
[CT11] M. Chabert and J.-Y. Tourneret, “Bivariate Pearson Distributions for remote sensing images,” in Proc. IEEE Int. Conf. Geosci. and Remote Sensing (IGARSS), Vancouver, July 2011.
[CB05] D. Cho and T. D. Bui, “Multivariate statistical modeling for image denoising using wavelet transforms Signal processing,”  Image communication 2005, vol. 20, no1, pp. 77-89, Elsevier.
[DV02] M. Do and M. Vetterli, “Wavelet-based texture retrieval using generalized gaussian density and kullback-leibler distance,” IEEE Transactions on Image Processing, vol. 11, no. 2, pp. 146–158, February 2002.
[MMS08] G. Mercier, G. Moser and S. B. Serpico, “Conditional Copulas for Change Detection in
Heterogeneous Remote Sensing Images,” IEEE Trans. on Geosci. and Remote Sensing, vol. 46, no. 5, May 2008.
[BLLP10] N. Brunel, J. Lapuyade-Lahorgue, and W. Pieczynski, « Modeling and unsupervised classification of multivariate hidden Markov chains with copulas », IEEE Trans. on Automatic Control, Vol. 55, No. 2, pp. 338-349, February 2010.
[BP05] N. Brunel and W. Pieczynski, Unsupervised signal restoration using hidden Markov chains with copulas, Signal Processing, Vol. 85, No. 12, pp. 2304-2315, 2005.
[TPCI09] J.-Y. Tourneret, V. Poulain, M. Chabert, and J. Inglada, “Similarity measure between vector data bases and optical images for change detection,” In Proc. IEEE Int. Conf. Geosci. and Remote Sensing (IGARSS), pages 992–995, Cape Town, South Africa, July 2009.
[CTPI10] M. Chabert, J.-Y. Tourneret, V. Poulain and J. Inglada, “Logistic regression for detecting changes between databases and remote sensing images,” In Proc. IEEE Int. Conf. Geosci. and Remote Sensing (IGARSS), Hawaï, July 2010.