Détail de l'auteur
Auteur Faith Njoki Karanja |
Documents disponibles écrits par cet auteur (1)
Ajouter le résultat dans votre panier Affiner la recherche Interroger des sources externes
Use of knowledge based systems for the detection and monitoring of unplanned developments / Faith Njoki Karanja (2002)
Titre : Use of knowledge based systems for the detection and monitoring of unplanned developments Type de document : Thèse/HDR Auteurs : Faith Njoki Karanja, Auteur Editeur : Munich : Bayerische Akademie der Wissenschaften Année de publication : 2002 Collection : DGK - C Sous-collection : Dissertationen num. 558 Importance : 107 p. Format : 21 x 30 cm ISBN/ISSN/EAN : 978-3-7696-9597-7 Note générale : Bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Applications de télédétection
[Termes IGN] aide à la décision
[Termes IGN] base de connaissances
[Termes IGN] classification non dirigée
[Termes IGN] détection de changement
[Termes IGN] image satellite
[Termes IGN] interprétation automatique
[Termes IGN] modélisation
[Termes IGN] photo-interprétation
[Termes IGN] prospective
[Termes IGN] représentation graphique
[Termes IGN] simulation
[Termes IGN] surveillance de l'urbanisation
[Termes IGN] système à base de connaissances
[Termes IGN] système d'information géographique
[Termes IGN] traitement d'image
[Termes IGN] zone urbaineRésumé : (Auteur) Within the context of an urban land use, two general status can be distinguished namely developed land (residential, industrial area, construction area, etc), and reserved land (forest, water bodies, garden, parks, open area etc). However, not all developed land is legal or formal a phenomena prevalent in developing countries. For the planners it is important to have the capacity to detect, localise and predict the trend of this phenomenon in order to facilitate reaction planning. In this regard, the main objective of this study is to develop a methodology that exploits existing planning data and urban land use drivers in combination with remotely sensed imagery for the detection and quantification of unplanned developments and subsequently facilitate in monitoring their trends. Since planning is a continuous process, remotely sensed data lends itself to a good source of information pertaining to the extent of developed and reserved areas at any given epoch. The methodology has been subdivided into four modules, namely the interpretation, detection, trend prediction and the evaluation.
Knowledge based image interpretation, namely rule based system was employed in the extraction of developed and reserved areas from multispectral image data. As input into the interpretation process four images cues i.e. NDVI, Texture, Edge Density, and unsupervised classification have been tested. Experience has shown that data reduction and refinement prior to its incorporation in the knowledge base enables few rules to be established and thereby minimising rule correlation. However, knowledge representation is a challenge and especially when it involves rules association. Acceptable results have been obtained which imply that such a technique is promising. Additional information e.g. GIS data would nevertheless be useful as a guide in the extraction of objects like parking areas and some complex built up areas (e.g. buildings), roads, etc thus improving the results.
For the detection of unplanned developments, the constraints of planning data have been exploited. Specifically, planning data has been used in combination with remotely sensed data depicting As It Were situation to generate As It Should Be scenario with specific emphasis on the new legal/allowable developments. This formed a backdrop for the detection of unplanned developments, which essentially translates to the difference between the As It Should Be situation and As It IS within the planning and implementation time frame. Prototype experiments carried out show that this is a feasible technique and can easily be implemented for fast detection in comparison to the current ad hoc field techniques.
In the trend prediction of unplanned developments, land use drivers based on compatibility of land uses, transport network, and hydrography sources are used to establish their influence on the new developed areas. Results show that existing land uses influence highly new developed areas. A combination effect of these influences (land use drivers) when employed results in stratification of pressure zones into fuzzy blocks ranging from those which are likely to undergo extensions of unplanned developments to those which are unlikely. Such information could enable planners develop a program in preparation for such eventualities and priorities areas that require urgent reaction planning.
For the evaluation of the results, both qualitative and quantitative evaluation techniques have been tested namely visual comparison, polygonpixel count and kappa index. Comparable results have been obtained based on these three techniques for the interpretation and detection modules. The choice of which technique to apply depends on the application and the level of detail required.
In conclusion, this study has demonstrated that a 'complete package' that will enable planners to detect and predict the trend of unplanned developments is feasible. In order to ensure success, planning should be viewed as a collective responsibility, where all stakeholders are participants. Further, sound land law system that ensures land accessibility and security of tenure as well as promotes transparency in land allocation issues should be implemented and enforced.Numéro de notice : 15027 Affiliation des auteurs : non IGN Thématique : IMAGERIE Nature : Thèse étrangère En ligne : https://d-nb.info/966084659/34 Format de la ressource électronique : URL Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=55045 Réservation
Réserver ce documentExemplaires (2)
Code-barres Cote Support Localisation Section Disponibilité 15027-02 35.46 Livre Centre de documentation Télédétection Disponible 15027-01 35.46 Livre Centre de documentation Télédétection Disponible