Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.590965
Title: Change detection of land cover using visual texture measures
Author: Yang, Fen
Awarding Body: University of Aberdeen
Current Institution: University of Aberdeen
Date of Award: 2007
Availability of Full Text:
Access through EThOS:
Full text unavailable from EThOS. Please try the link below.
Access through Institution:
Abstract:
It is important to detect land cover changes from remotely sensed data for monitoring environment. Although there are a few applications of visual textures to land cover classification, they are limited to a few major land cover categories with the application of one textural measure. For a semi-natural environment, such as Scottish land cover, there are many land cover categories, which differ from each other by subtle differences. For this situation it is not enough to apply one textural measure to characterise land cover categories. Moreover because land cover change cannot happen between any two land cover categories, it is possible to detect land cover change by discriminating between pairs of land cover categories which could change from one to the other. This thesis, therefore, attempts to present descriptions of various texture measures for land cover categories and apply them to detect land cover changes by pairwise discrimination. Three different bandpass filter banks: Haar masks, Laws masks and Gabor filters are applied to extract land cover textures. In order to learn the difference between the three filter banks, we analyse the properties and the spatial frequency for each filter bank. Their abilities to discriminate different land cover categories are evaluated by discriminant analysis. Because the three filter banks divide the spatial frequency domain differently, their abilities are different. In order to learn which filter bank is the best for the discrimination of a specified pair of land cover categories, we compare the performance of the three filter banks and obtain a ranking table which could be used as a textural knowledge for a system for monitoring land cover. Experiments on the Elgin area of Scotland show that texture measures are promising for discriminating land cover categories. And for each permissible land cover change, the best filter bank is shown.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.590965  DOI: Not available
Share: