Linking Thermal Images with 3D Models for FFF Printing

The thermal profile plays a major role in additive manufacturing. Thermal cameras are thus increasingly used for quality monitoring. So far, either full thermal images or metrics extracted from them are used to monitor the manufacturing quality or detect defects. To additionally allow the detection of local anomalies, it is necessary to link the thermal image to the 3D object geometry. We propose a framework that includes steps for filtering object points that are visible from the camera perspective, projecting 3D points onto thermal images and removing pixels that represent the printhead. Our framework can be used for process monitoring and subsequent on-line defect detection which are necessary components for production automation and Industry 4.0 applications. In a validation experiment, we show that the temperature extracted from thermal images and assigned to 1mm × 1mm × 1mm voxels is highly correlated to the temperature measured with type K thermocouples.