Generating High Fidelity 3-D Thermograms With a Handheld Real-Time Thermal Imaging System

Infrared thermography is a widely used technique to measure and portray the surface temperature of an object in form of thermal images. Two-dimensional images, however, have some inherent limitations with regard to the fidelity with which they can depict the surface temperature of a three dimensional object. In the past two decades, there have been several works describing different techniques to generate 3-D models textured with thermal information using various combinations of sensors in order to address some of these limitations. Most of these approaches generate 3-D thermograms of an object from a single perspective with bulky measurement systems and therefore do not address problems that arise when scanning objects in a continuous manner from multiple perspectives. But reductions in cost, size, and weight of infrared and depth-sensing cameras as well as a significant increase in computational power of personal computers have enabled the development of low cost, handheld, real-time 3-D thermal imaging systems. This paper elaborates through a series of experiments on the main factors that affect the real-time generation of 3-D thermograms with such a system and demonstrates how taking these factors into consideration significantly improves the appearance and fidelity of the generated 3-D thermogram. Most of the insight gained in this paper can be transferred to 3-D thermal imaging systems based on other combination of sensors.