Investigation on performance of RGB point cloud and thermal information data fusion for 3D building thermal map modeling using aerial images under different experimental conditions

Three-dimensional thermal mapping provides many benefits for auditing building energy performance when compared with two-dimensional thermal images that provide only limited representation. However, the current thermal mapping approaches have accuracy and efficiency tradeoffs when modeling large areas using aerial images. Particularly, low-resolution thermal images make it harder to obtain a high-quality thermal mapping model. To avoid using low-definition thermal images to reconstruct building thermal models and improve the performance of large-area thermal mapping, we propose a thermal and RGB data fusion framework for thermal mapping. This paper aims to evaluate how different experimental conditions on the proposed fusion approach affect the results. The evaluated conditions include different camera altitudes (60 m and 35 m), distinct camera angles (45° and 30°), diverse flight path designs (mesh grid and Y path), and various building styles (university campus buildings and buildings in a central city area). The results demonstrate that different performances of conducting the proposed data fusion approach under different conditions are observed, and the study provides suggestions for using this approach in different conditions. In addition, this study provides strategies for improving the accuracy of the proposed framework.