Estimation of Weld Quality in High-Frequency Electric Resistance Welding with Image Processing An image analysis algorithm can estimate the weld quality by using the weld image data during high-frequency electric resistance welding

In high-frequency electric resistance welding (HF-ERW), the weld quality is determined by the welding phe- nomenon in the welding spot proximity. Methods to eliminate or minimize defects, therefore, include real-time monitoring of the weld quality and problem solving as problems arise. It is possible to estimate weld quality qualitatively by using the weld image data during HF-ERW. A method to predict the weld quality using image pro- cessing in the heating process of the HF- ERW point is proposed. An algorithm, which predicts the weld quality by using a vision sensor to obtain the image of the welding spot proximity thus best express- ing the welding phenomenon, is devel- oped. The heated area, which shows the image data with the highest correlation to the weld quality and the smallest amount of noise, is consequently calculated.