| Summary: | 碩士 === 朝陽科技大學 === 營建工程系碩士班 === 99 === Evaluating the restoration conditions for those areas affected by natural disasters is an important research issue in disaster prevention and protection programs. The issue needs to simultaneously evaluate the hardware (the restored conditions for the physical environment) as well as the software (the restored conditions for psychical development). This thesis focuses on evaluating the hardware restoration conditions using the thermal infrared remote sensing. In this thesis, the study area lies between the longitudes of - E and latitudes of - N, and is an area prone to typhoon attacks. The studied area suffered serious damage from Typhoon Morakot on 2009/08/08. Landsat-7 ETM+ SLC-Off scenes collected on 2009/2/24,2009/05/31,2009/09/04 and 2009/12/09, were used to generate land surface temperature (LST) images. By employing the multilayer level set approach to segment the given LST images, landscape patterns for the study on different dates were identified and compared, in order to evaluate the time required for the landscape patterns to return to their original states. The multilayer level set approach employs two implicit functions with pre-selected level values. Based on the fact that the segmented regions are homogeneous and presented as regional constants, the energy defined by the segmented regions and their corresponding regional boundaries is minimized such that the relationships between the defined energy and the implicit functions can be transformed into the relationships between the implicit functions and time. By implementing the algorithm in terms of finite difference, this method offers an efficient and stable approach to a numerical solution. By increasing iterations and preselected level values, the implicit functions evolve close to the regional boundaries based on the energy minimization. This thesis demonstrates that the generated LST scenes can provide the opportunity for better understanding the land surface characteristics and landscape patterns, especially for those areas visited with natural disasters.
|
|---|
