| With a rapid development of world population, urbanization has become an important issue in the past century and generated great pressure on the environment. In order to assist urban decision-makers to present, analyze, and solve related issues, accurate and timely spatial information is needed in urban areas. This research focused on building information extraction and the relationship analysis of buildings and temperatures in the downtown Indianapolis, IN. Through object-oriented classification, the image derived from LiDAR data was segmented into objects, and spectral, spatial, and textual attributes of objects were measured and selected to extract buildings. Moreover, each building's surface temperature was produced by combining Land Surface Temperature data (derived from ETM+ Thermal band) and building footprints to represent 3D building temperatures in a 3D environment. 3D building temperatures were then visually and statistically analyzed to examine the relationships between 3D building geometries (such as area and height) and building surface temperature. Through object-oriented classification, the extracted building image got an accuracy of 87.5% (Detection Percentage) and 78.9% (Quality Percentage). Through correlation analysis and cluster analysis, certain relationships were shown between 3D building geometries and building temperatures. While they are not high enough for practical application, under the similar situation, height has much higher significance in the relationship with building temperature than area for business buildings with large area and/or big height. |
