Abstract
ABSTRACT In recent years, due to the availability of affordable 3D sensors and the increased computing power, various methods for the generation of 3D thermograms have been developed. 3D thermal imaging describes the fusion of geometry and temperature data. A well-established approach is the fusion of data from depth and long-wave infrared (LWIR) cameras. However, these models generated in real-time have the limitation that the model size is limited due to inefficient data storage approach. Newer algorithms from Computer Vision promise to overcome this limitation by more efficient data handling and storage. Within this work, three state of the art 3D reconstruction algorithms from the computer vision community are compared and one of these is extended by overlaying thermal data, which allows the creation of large-scale 3D thermograms with a portable 3D measurement system. For this purpose, a geometric calibration is required, the data structure is adapted, and the handling of cyclic non-uniformity corrections required for uncooled LWIR cameras is described. The results will show exemplary 3D thermograms and the advantages compared to current existing systems.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call