Lap welding is employed to join thin plate materials in car assemblies. The quality of welding depends on its conditions and seam tracking in the welding line. An arc sensor was employed by oscillating the welding torch in a thick weld. Since the welding torch was not oscillated in the welding of thin plate materials, the arc sensor could not be applied to lap welding. To trace the welding line, the weld pool in pulsed metal active gas (MAG) welding under the shielding gas of Ar 80% and CO 2 20% was pictured using a CMOS camera. The weld pool was observed in the rear to the welding direction. The features of the weld pool were investigated to recognize the welding line. Since the arc affected the brightness of the weld pool, the timing of the shutter of the CMOS camera was synchronized with the current waveform to take clear weld pool images. The welding line was detected using an image processing method. In this paper, a banalization method and a pattern matching method was investigated. The pattern matching method is useful than the binarization method. A digital controller was designed to trace the welding line. The validity of the proposed method was verified by carrying out a tracking experiment.