Multi-temporal Hyperspectral Images Unmixing and Classification Based on 3D Signature Model and Matching

International audience Land cover and land use types are challenged to access real-time and precise information of interest. The recent advent of sophisticated sensors permits to exploit independent observations of a phenomenon and to extract more detailed information and performs a decision level for scene interpretation. In this paper, we propose a new approach for multi-temporal hyperspectral images processing based on multi-temporal spectral signature representation. The 3D model characterizes all the pixels in a scene by considering their reflectance values as a function of time of imaging and spectral waveband. We showed the use of such modeling strategies in overcoming the dimensionality problem and improving both multi-temporal classification and unmixing problems associated with hyperspectral data. A case study was conducted on multi-temporal Hyperion series located in southern Tunisia. The obtained results showed good accuracies.