For welding process automation, the development of sensor system which can sense the position of weld seam and its shape informations such as joint size, gap size and surface mismatch is necessary to track the seam and to control welding process variables automatically. The vision sensor which is based on the optical triangulation theory with the laser as an auxiliary light source can sense not only seam position but the shape of seam. Therefore it is possible that the tracking weld seam and the variables control of welding process by using the vision sensor. In this study, the vision sensor using scanning beam was developed. To design the vision sensor which considers the reflectivity of the sensing object and satisfies desired resolution and measuring range, firstly, the equation of focused laser beam which has Gaussian irradiance profile was formulated. Secondly, the image forming sequence was formulated. Thirdly, the relation between the displacement in the measuring surface and the displacement in the camera plane was formulated. Therefore, the focused beam diameter in the measuring range could be determined, the influence of the relative location between laser and the camera plane could be estimated and the measuring range and the resolution of the vision sensor which is based on the Scheimpflug's condition could be calculated. From above results the vision sensor was manufactured, and a calibration technique of that was proposed. Image processing algorithm that can recognize the center of joint which is needed to track the seam and shape informations which are needed to control the welding process variables was proposed. Using developed vision sensor and image processing algorithm, the shape informations of vee-joint and butt-joint were extracted.