manufacturing requires inspection methods that are both reliable and compatible with real-time production environments. Conventional non-destructive testing (NDT) techniques are typically applied after welding, are costly, and do not provide continuous feedback on process stability or weld integrity. This thesis investigates thermographic image and video post-processing as a basis for automated inline monitoring of laser and TIG welding processes, with the objective of detecting defects directly from thermal data and supporting real-time quality assessment. A laboratory dataset of welding runs was acquired using a high-frame-rate long wave infrared thermal camera, capturing the full thermal evolution of the molten pool and surrounding heat-affected zone.