A Scheme for Spatial Resolution Improvement of Thermal Images for Objects with Different Emissivity

We propose a method of acquiring spatial high-resolution thermal distribution from a low-resolution infrared camera while correcting the temperature by emissivity. Thermal images are dependent on the color and material of the measurement target. Therefore, the accurate temperature cannot be measured by infrared thermal camera as it is. This is a serious problem in the temperature measurement on electrical terminal with a large current for electrical equipment diagnosis. In this paper, we improve the temperature profile of targets by example-based super-resolution, that is, by a dictionary from a pair of low-resolution and high-resolution thermal images with corrected by emissivity. In some experimental results, we show that temperature profiles are improved, that is, the profile of temperature is close to the measurements by a thermocouple device, and those by a high-resolution infrared thermal camera.