Détail de l'auteur
Auteur Urs Frei |
Documents disponibles écrits par cet auteur (1)
Ajouter le résultat dans votre panier Affiner la recherche Interroger des sources externes
Compilation of cartographic and spaceborne remote sensing data for thematic/topographic mapping / Urs Frei (1993)
Titre : Compilation of cartographic and spaceborne remote sensing data for thematic/topographic mapping Type de document : Thèse/HDR Auteurs : Urs Frei, Auteur Editeur : Zurich : Université de Zurich Année de publication : 1993 Collection : Remote sensing series num. 22 Importance : 102 p. Format : 21 x 30 cm Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Traitement d'image optique
[Termes IGN] classification
[Termes IGN] correction géométrique
[Termes IGN] correction radiométrique
[Termes IGN] image Landsat-TM
[Termes IGN] intégration de données
[Termes IGN] modèle numérique de terrain
[Termes IGN] occupation du solIndex. décimale : 35.20 Traitement d'image Résumé : (auteur) A survey of the current topographic maps at the scale 1:100,000 from all member states of the European Space Agency in 1985 showed that the average age was 12 years, that the relief presentation in many cases was relatively poor and that they contained little information on land use. The Alban Hills, situated about 20 km south-east of Rome served as a test site to demonstrate how digital terrain data, spaceborne remote sensing data and conventional cartographic data can be integrated in order to extend the current map contents and to expedite intermediate map revision.
After a description of the Landsat system, the algorithm for both the geometric and radiometric preprocessing of the Thematic Mapper imagery is outlined. Evaluation of system corrected data revealed excellent accuracy with respect to the internal geometry. However, scene dependent, relief induced distortions had to be treated separately, so a simple algorithm for their removal was developed. The ground control point characteristics were investigated and the manner of their selection was found to be decisive. The accuracy of the geometric rectification was verified with independent test points; residuals were in the range of 0.5 pixels.
The aim of the radiometric preprocessing was to improve the conditions for the subsequent classification. This goal was achieved by applying an iterative edge preserving smoothing filter that generates more homogeneous areas.
After the preprocessing of two scenes that were acquired in March and August 1986, bands 3, 4 and 5 were selected for cluster analysis. This unsupervised classification approach led to 36 classes, subsequently merged into seven final land use categories. The overall classification accuracy was 84.5 %, with significant differences between the individual categories.
For the compilation of the new map, some information was derived from the existing map, namely the planimetric layer and the digital elevation data. The planimetric layer was digitized and reformatted in order to match the reference UTM coordinate grid. The digital elevation model was derived from the map contour lines. Its quality was sufficient for the geocoding, but an elevation model of higher resolution is necessary for a good representation of the relief.
The basic idea behind the new map concept is that both the relief (which determines the shape of the landscape) and the land use (which determines the colours of the landscape) should be displayed in a topographic map, in order to fully capture the landscape's character. Such a product can be generated by applying appropriate colour models. In the final map, the HLS (Hue, Lightness, Saturation) colour model was applied, with the land use defining the hue and saturation while the hill shading modulated the lightness.
This study shows one way to integrate remote sensing data with cartographic and digital terrain information. However, the data preprocessing and the classification algorithm must be improved before operational application will be feasible:
- A ground control point chip library is required for semi-automatic geocoding.
- Raw (rather than system-corrected) data should be used to avoid a second resampling.
- The geocoded data should have a standard pixel size of 25 x 25 m.
- A digital elevation model with a spatial resolution that is four to five times better than the resolution of the image should be used for slope/aspect corrections.
- Banding effects and forward/reverse scanning differences should be corrected.
- A library containing training areas for the classification must be built up to expedite the classification process.
- Better classification results could be achieved using a region-based classification approach.Note de contenu : 1- Introduction
2- Fundamentals
3- Preparation of Thematic Mapper data
4- Multitemporal land use classification
5- Preparation of map data
6- Integration of different data sets
7- Discussion of resultsNuméro de notice : 14345 Affiliation des auteurs : non IGN Thématique : IMAGERIE Nature : Thèse étrangère Note de thèse : PhD dissertation : Geography : Zurich : 1993 Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=88780 Exemplaires(1)
Code-barres Cote Support Localisation Section Disponibilité 14345-01 35.20 Livre Centre de documentation Thèses Disponible