Hilfe beim Zugang
Information Extraction and Modeling from Remote Sensing Images : Application to the Enhancement of Digital Elevation Models
The analysis of remote sensing images aggregates numerous disciplines which were, before the advent of Earth observation (EO) techniques, distinct fields. The synergies between Cartography, Photogrammetry, Image processing and analysis, Computer vision, etc. allow to observe, monitor and study our e...
Ausführliche Beschreibung
The analysis of remote sensing images aggregates numerous disciplines which were, before the advent of Earth observation (EO) techniques, distinct fields. The synergies between Cartography, Photogrammetry, Image processing and analysis, Computer vision, etc. allow to observe, monitor and study our environment. In this dissertation, several methods to analyse, interpret and enhance satellite images are considered. Those different axes are motivated by a crucial need of efficient and novel methods for the extraction and the modeling of relevant information from standard remote sensing images in order to fully exploit those data presenting metric resolution over large areas. During this research work, an innovative, fast and robust image processing system has been fully developed, tested and validated. It consists in, (1) a robust region-based segmentation approach, using the Mumford and Shah formalism to generate meaningful partitions from large remote sensing images, (2) a novel dynamic algorithm to extract and express the regions with the complex cellular model. A tree structure is introduced to store the regions while reflecting their topological (adjacency, inclusion) and geometrical properties. (3) information learning methods to analyse and interpret the extracted and modeled primitives. The interest of such a processing line is emphasized on various EO data ranking from decametric to metric resolution (e.g. multispectral SPOT 5, Ikonos or QuickBird, panchromatic Cartosat) presenting various coverage types (e.g. agricultural fields, communication networks, man-made structures, urban areas) ... Ausführliche Beschreibung