University of Twente Student Theses


Homogeneous Urban patches defined by texture and pattern for the detection of slums

Yilma, Hiwot Abera (2012) Homogeneous Urban patches defined by texture and pattern for the detection of slums.

[img] PDF
Abstract:The growing number of slum population has become a challenge to urban planners and decision makers. Different approaches have been applied to identify and monitor the slum at different stages. Most techniques relay on time taking analysis like through the use of census population data and field based identification methods. The availability of very high resolution images provides a new approach in remote sensing, allowing efficient analysis in the urban environment. The objective of this study is to develop a semi- automatic method using Object Oriented Analysis (OOA) that will enable to identify Homogeneous Urban patches (HUPs) and to characterize them for further texture and pattern based slum detection. In the previous studies HUPs have been delineated manually. Using very high resolution (VHR) pansharpened IKONOS image of New Delhi, the approach of this study is based on (1) delineating HUP boundaries through OOA approach and finding the spatial and spectral characteristics of HUPs through spatial metrics and Gray Level Co-occurrence Matrix (GLCM), (2) analyzing HUPs across different scale images, multi-scale analysis, (3) testing the transferability of the method on another area, and (4) HUP based slum identification. The result shows that HUPs were delineated successfully through a semi-automatic approach. The accuracy of the delineated HUPs is assessed by comparing the manually and automatically delineated HUPs and results an accuracy value of 85% and 71% in the two test areas respectively. Texture and pattern based analysis through spatial metrics and GLCM measures are found to be an important measure in characterizing HUPs. Spatial metrics were providing a higher amount of information for classification of HUPs in to different land use categories. The changes ‘within’ the delineated HUPs and ‘on the boundary’ were clearly recognized from the spatial metrics and GLCM results on different scales. The method can be transferred to another area depending on the nature of the site and the rules that have to be set at the beginning of the analysis. The HUPs show a major advantage in identification and characterization of slums in terms of texture and pattern. The results demonstrate the practicability of the approach in different areas. Keywords: Homogeneous Urban Patches, Object Oriented Analysis, Spatial metrics, GLCM, Multi-scale analysis, Slum identification
Item Type:Essay (Master)
Faculty:ITC: Faculty of Geo-information Science and Earth Observation
Programme:Geoinformation Science and Earth Observation MSc (75014)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page