Correction des Effets de Relief en Spectro-imagerie Aéroportée

Abstract : We present a pipeline of algorithms applied to the images of an imaging static Fou- rier transform spectrometer (ISFTS), and more generally to windowing hyperspectral instruments. Those instruments measure hyperspectral cubes that can become flawed if the induced parallax error between the images is not corrected using a digital elevation model (DEM). The latter is generally not known. We surmise that the many images of the interferometric cube could be used in order to compute the DEM. What we aim to achieve in this thesis is therefore to compute the DEM of a given scene from the images of an ISFTS. We are faced with several difficulties. Firstly, the instrument provides images of the scene on which a fixed interference figure is superimposed. Secondly, the observed disparities between two consecutive images of the sequence are less than a hundredth of a pixel, which calls for sharp subpixel processing. We tackle the problem of interference fringes by ridding ourselves of them through image decomposi- tion. To this purpose we propose two models : an additive model and a multiplicative one. In order to take advantage of the redundancy of information on the images and handle subpixel disparities, we use the epipolar plane images to obtain a precise sparse disparity map. One important assumption to be able to use this method is that all camera centers should be aligned. This is not a given in airborne acquisitions, and so we study the influence of trajectory fluctuations on the disparity measure. Finally, in order to obtain a dense height map, we densify the sparse values by minimising a functional using graphcuts.
Complete list of metadatas

Cited literature [161 references]  Display  Hide  Download
Contributor : Clara Barbanson <>
Submitted on : Wednesday, November 7, 2018 - 9:34:42 PM
Last modification on : Friday, June 7, 2019 - 11:08:55 AM
Long-term archiving on : Friday, February 8, 2019 - 3:39:17 PM


Files produced by the author(s)


  • HAL Id : tel-01915693, version 1



Clara Barbanson-Graziussi. Correction des Effets de Relief en Spectro-imagerie Aéroportée. Traitement du signal et de l'image [eess.SP]. Telecom ParisTech, 2018. Français. ⟨tel-01915693⟩



Record views


Files downloads