The traditional method for post-weld quality inspection is visual testing by people. Major limitations of this method are the subjectivity of judgments and time required to perform the inspection. With the increasing use of laser welding in the automotive industry, the manufacturers are struggling with the challenge to automate the quality inspection process. This paper discussed the automatic post-weld quality inspection technology by computer vision sensing for mono-thickness tailored blanks of CO2 laser welding. A visual sensor was developed for acquiring the original image of the weld seam. The sensor consists of a PC based vision camera and a stripe-type laser diode. An image processing algorithm was presented to detect the geometrical defects of the weld seam. Some experiments are carried out and some applications are given.