Thermal Image Processing

The research area of ​​the professorship DST is not limited to the visible wavelength range (380 - 780 nm). Also the long infrared range between 7,5 - 13 μm wavelength is subject of the research. The thermal radiation is recorded with a thermal imager and converted into image data. This process is called thermography. Since the representation of objects and persons are clearly different from e.g. monochrome cameras or RGB-D cameras, the algorithms for image processing must also be redesigned or adapted.

Fig 1: Sensor system consisting of thermal imaging camera and RGB-D camera

Within the scope of research projects, RGB-D cameras were calibrated to thermal imaging cameras in order to determine a common coordinate system. Based on this, a pixel-to-pixel assignment of the two camera systems is possible.

Fig 2: Calibration target with circle pattern

Since the standard targets with chessboard patterns are only insufficiently usable in the infrared range, own calibration targets were designed and manufactured. In particular, cut-out circular structures have proven to be reliable due to the stable center point.

Fig 3: Determination of the temperature of body parts using the example of the upper arm

One focus of the research is the extraction of heat information on the surface of various body parts. For this purpose, a skeleton model is mapped onto the infrared image, which was obtained from depth data. This method is used in visual rehabilitation and training support to evaluate the effectiveness of trainings.

Fig 4: Skeleton extraction in thermal image

In addition, methods are being researched to extract skeletal models directly from the thermal image data.