Abstract

Infrared thermography (IR) is a modern, non-destructive evaluation technology for monitoring and assessing built environments. It mainly relies on measuring surface temperature to identify any potential defects or damages. Currently, IR has been introduced widely in applications such as facility condition assessment and energy performance analysis of existing buildings. However, most of the current practices in IR rely only on 2D thermal images which are time-consuming and labor-intensive. On the other hand, the rapid improvement of high-defined IR cameras has become a powerful tool in infrared sensing. Accordingly, this has facilitated its implementation in 3D thermal modeling techniques to replace the current 2D approach in thermal inspection and building energy efficiency. Yet, further studies need to be performed to overcome 3D thermal modeling limitations such as the high cost, slow process, and the need of highly trained professionals.
The main objectives of this research are to (a) test the potentiality of using 2D visible and thermal images which were collected separately through digital and infrared cameras respectively, for the 3D thermal modeling of built environments, and (b) investigate the efficiency of the proposed methodology by comparing it to a developed experimental design in terms of evaluating density, time, and cost. In specific, the visible images were used in modeling 3D point clouds by applying the structure from motion (sfm) approach. In parallel, the overlapping thermal images were stitched to form a thermal panoramic image that covers a large surface area with an accurate temperature representation. The stitched thermal images were then mapped to the reconstructed 3D point cloud in order to generate both thermal and metric measurements of built environments. Correspondingly, the output was compared to another 3D thermal point clouds which were developed by a laser scanner and an infrared camera. The comparison was conducted by means of evaluating density, time, and cost. Finally, the comparison results of three different built environments in the city of Montreal, Canada; demonstrate that 3D thermal modeling using separate 2D thermal and visible images was able to generate a dense geometric and thermal information of built environments. Also, this approach is affordable in terms of cost and time.