Among the various types of sensors used in welding automation, are and vision sensors are widely used for automatic seam tracking and quality control of are welding processes. The development of a vision sensor which can provide 3D geometry information is needed especially for robot guidance in automatic welding of thin plates, because it is difficult to apply an are sensor to the welding of thin plates. Vision sensors using optical triangulation have been widely used in various ways for automatic welding systems. The separation angle between the camera and the laser axis has been mainly considered in previous studies of the vision sensor with structured light. Their measuring efficiency is, however, considerably influenced by the different arrangements of the weldment, the CCD camera, and a diode laser of a vision sensor in three-dimensional space. Therefore to enhance the effectiveness of a vision sensor for height-varying workpieces, other geometrical parameters such as diagonal and tilt angles should also be investigated. In the present study, the data deficiency in the vision sensor falls into two classifications: the shadow effect and the missing field of view (FOV), and a mathematical model is proposed to estimate the occurrence of data deficiency. (C) 1997 Elsevier Science Ltd.