A method is introducedfor implementing a 3-D vision system considering manufacturing cost constraints stringent reliability requirements and limited processing power. It entails transducer characterization and includes a scanning algorithm and low-complexity processing algorithms providing position shape and orientation information about an unknown surface. A highly modular approach subdivides the problem into separable and sequential processing modules gradually modifying a " data matrix" obtained from a scanning algorithm. Some models considered for planar and curved surface analysis are described. Theoretical and experimental background validating the implemented algorithms is presented.