Tomographie SAR pour la détermination des hauteurs des structures urbaines et forestières

الموضوعات

علوم تطبيقية وتكنولوجية

الملخص

La tomographie SAR est une technique basée sur l'acquisition de plusieurs images SAR de la même scène à des angles de vues différents. Elle est utilisée pour la cartographie des espaces urbains et forestiers en reconstruisant les diverses structures en 3D. Dans cette thèse, l'objectif que nous nous sommes fixés est de contribuer aux développements d'outils mathématiques nécessaires pour l'estimation de la hauteur des structures urbaines et forestières. Dans cette thèse, nous avons démontré que l’ algorithme développé basé sur la fusion DFT-LS présente une alternative intéressante par rapport aux autres techniques d'inversion tomographique, afin de trouver un compromis entre une résolution élevée, une complexité modérée et un temps de calcul acceptable.

روابط وملفات

التعريف والنوع

رقم الوثيقة

ede5428a-4565-440b-84c8-f42876def0c6

رقم العقد

0

نوع الوسائط

Crawler

نوع المحتوى

الرسائل العلمية

صيغة المصدر

رسائل دكتوراة

نوع الملف

word

أسماء الملفات

1372732_1.docx, 1372732_2.pdf

بيانات النشر

ألقاب المؤلفين

[{"name_ar":" Hichem Mahgoun","title_ar":"اعداد","title_en":"Preparation"},{"name_ar":" Mounira Ouarzeddine","title_ar":"اشراف","title_en":"Supervision"}]

اللغة

French

المصدر والدورية

اسم المصدر

Tomographie SAR pour la détermination des hauteurs des structures urbaines et forestières

المحتوى والصفحات

عدد الصفحات

0

ترجمة الملخص

SAR tomography is a technique based on the acquisition of several SAR images of the same scene at different viewing angles. Its objective is to improve the elevation resolution of the structures. It is used for mapping of urban and forest areas by reconstructing the various structures in 3D. For urban areas, tomography can have a direct impact on new 3D mapping services and land use planning. For forest areas, the ultimate perspective is the continuous monitoring of forest evolution and thus track carbon stocks and fluxes on a global scale.In this thesis, the objective we set ourselves is to contribute to the development of ma-the matical tools necessary for the estimation of the height of urban and forest structures using Tomo SAR. SAR tomography is a complex and rapidly expanding research area that opens up a multitude of approaches and techniques, based on parametric and non parame-tric spectral estimation algorithms. In this thesis, we first addressed the SAR polarimetry, by which we synthesized the volume component necessary to estimate the height of the forests.We have investigated the PolIn SAR by describing the various inversions existing algorithms.The purpose of these PolIn SAR inversions is to compare them with the various tomographic inversions. As for the various tomographic inversion algorithms and after the analysis of the pro-perties of the existing algorithms, we proposed improvements to the efficiency of inversiontechniques. The first contribution concerns the development of an hybrid inversion algo-rithm, dedicated to urban spaces with point backscattering contributions, by combining the non-parametric inversion DFT and the parametric inversion LS.This algorithm allows to reduce the computation time of the inversion and to increase the height resolution. The second contribution concerns the development of an algorithm dedicated to forest areas with volume backscattering contributions, based on the fusion of the non-parametric inversion DFT and the parametric inversion LS. This algorithm allows the estimation of the height of the volume and its position, to increase the height resolution and to optimize the calculation time. We then extended the principle of this algorithm to the MUSIC-LS fusion, and on the other hand we proposed a new inversion based on the optimal coherence exploited in the MUSIC inversion process.All the simulated and experimental results are grouped together in order to validate the proposed approaches. We first illustrated the results obtained for the simulation of urban structures with point backscattering contributions and we presented the performances of the inverse algorithms DFT, MUSIC, CAPON, combination DFT-LS and this in terms of reso-lution, bias, histogram and standard deviation of the distance separation over the elevationaxis .Secondly, we have illustrated the performances of the CAPON, MUSIC, MUSIC inver-sion algorithms with SKP decomposition, MUSIC with optimal coherence, RVoG (Random Volume over Ground), Fusion DFT-LS, Fusion MUSIC-LS, in terms of relative error, histo-xii grams, bias and standard deviation. Thanks to the experiments carried out, the difference inquality between the tomographic inversions and the RVoG of the PolIn SAR is highlighted,and we demonstrated that our algorithm based on the DFT-LS fusion presents an interesting alternative to the other tomographic inversion techniques, In order to find a compromise between high resolution, moderate complexity and an acceptable calculation time. Finally,we conclude with a synthesis and perspectives on all the research work carried out in this thesis

كلمات الباحثين

CAPON, DEM-DIFF, DFT, ESPRIT, Imagerie (technique), imagerie en trois dimensions (3-D), in-terf ́erom ́etrie, Interférométrie, interfero-metry, LS, MAP, MUSIC, Polarim ́etrie, Polarimetry, Radar (SAR), RVoG, SAR tomography, SINC, SVD, three-dimensional (3-D) imaging, Tomographie, tomographie-SAR, ارفوق, الانترفيرومتري, البولاريمتري, التمغرافي سار, دفتي, سفدي, سينك, صورة ثلاثية الابعاد, كابون, لس اسبريت, موزيك

إشراف وإعداد

الإشراف

Mounira Ouarzeddine

الإعداد

Hichem Mahgoun

الاقتباسات الببليوغرافية