Nowadays long multi-temporal series of SAR images taken under similar conditions are available. When no change has occured between the different acquisitions, the temporal distribution of a pixel provides useful information on the statistical models that can be used for SAR images.
The objective of this project is to study and model these temporal distributions. In particular the usual models (Gamma, Fisher, K, Rice, …) could be analyzed in the Mellin framework [1]. Then the comparison between these different distributions (for instance using Kullback Leibler divergence, etc.) could be exploited to define a classification approach for SAR scenes. Another axis of the project will be change detection in the temporal serie by the analysis of the temporal distribution [2].
Programming
Matlab or C/C++
Supervision
F. Tupin and JM Nicolas
References
[1] Nicolas, J.-M. and S.N. Anfinsen: Introduction to second kind statistics: Application of log-moments and log-cumulants to analysis of radar images
[2] F. Bujor, E. Trouvé, J. M. Nicolas et J.-P. Rudant, Application of Log-cumulants to the detection of spatitemporal discontinuities in multitemporal SAR images, IEEE Trans on Geoscience and Remote Sensing, Octobre 2004, vol. 42, n° 10, pp. 2073-2094 [DOI 10.1109/TGRS.2004.835304].