Building Change Detection Using High Resolution Remotely Sensed Data and GIS
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
In recent years, natural disasters have had an increasing impact, involving immense economical and human losses. Remote sensing technologies are being more frequently used for the rapid registration and visualization of changes in the affected areas, providing essential information for damage elimination, as well as the planning and coordination of recovery activities. Numerous methods of image processing have been proposed to automate a detection of changes on the Earth's surface, most of
... focus on the comparison of remotely sensed images of the same area acquired at different dates. However, atmospheric influences (e.g. clouds covering the objects of interest) often render the observations ineffective in the optical domain. In addition, the accuracy of the change detection analysis decreases if the images are acquired with different acquisition angles. These situations can be common in the case of sudden catastrophes (e.g. earthquakes, landslides or military actions), when there is no time to wait for the perfect conditions to acquire the data. This study presents a GIS-based approach for the detection of destroyed buildings. The methodology is based on the integrated analysis of vector data containing information about the original urban layout and remotely sensed image obtained after a catastrophic event. The integrated data processing enables minimizing the influence of the atmosphere and illumination effects, thus improving damage detection. Moreover, the object-oriented GIS technology makes it possible to concentrate on the investigation of specified objects, thereby reducing false alarms due to natural changes that occur around the investigated objects (e.g. seasonal changes of vegetation). Additionally, GIS based change detection analysis produces a tangible end product, namely damage maps. A new feature 'Detected Part of Contour' (DPC) was developed to identify a building's condition. The basic idea behind the proposed feature is the assessment of building contour integrity. The feature defines a part of the building contour that can be detected in the remotely sensed image, reaching a maximum value (100%) if the investigated building is intact. Furthermore, several features based upon the analysis of textural information are analyzed. Finally, a binary classification of building state concludes the change detection analysis. The experiments performed during this research indicate that employing a GIS-based analysis for change detection can essentially improve the potential for remotely sensed data interpretation and can be considered a powerful tool for the detection of destroyed building. The proposed methodology has been developed solely within the Open Source software environment (GRASS GIS, Python, Orange), whose use implies an innovative, flexible and cost-effective solution for change detection analyses. iv Acknowledgements First and foremost, I would like to thank my supervisor Prof. Dr. Eng. Manfred Ehlers for his unwavering support and valuable guidance during the research period. I am also very grateful to him for providing a workplace fully equipped with hard-and software, which enabled me to develop and implement all my ideas presented in this thesis. During this research, he supported me in various ways, including a financial base that allowed me to participate in international conferences to exchange experience with other researchers. Special thanks go to Prof. Dr. Eng. Peter Reinartz, the second supervisor, for sincere and objective support and criticism.