Development and evaluation of an image processing algorithm for monitoring fiber laser fusion cutting by a high-speed camera

Abstract

The authors report on a monitoring system for fiber laser fusion cutting based on a high-speed camera with spectral sensitivity in the visible spectral range. Without external illumination, the monitoring system is applied to a conventional cutting head with the optical detection path being coaxially aligned to the incident laser. It thus allows for the direct, spatially, and temporally resolved detection of the melt pool area in the cut kerf from the top view. The dependencies of the melt pool geometry on laser machining parameters can hence be evaluated. With different laser powers, feed rates, and assist gas pressures, incomplete cuts are forced in mild steel with sheet thicknesses ranging from 1 to 10 mm. The measured signal characteristics of the images captured from the melt pool are examined in the visible spectral range of the emitted, secondary thermal radiation from the process zone. Based on a specifically developed algorithm applied to the captured images, the authors are able to differentiate between complete and incomplete cuts with a distinguished detection probability of 94%.