View planning for inspection using CAD models

New view planning strategies for three dimensional objects are proposed. For object dimensional inspection, scaled image area and number of inspectable features are used to find the best view. With the required verification resolution, admissibility is obtained which is an index of the best view selection for dimensional inspection.; To perform view planning for three dimensional object reconstruction, surface area, number of surfaces and pixel resolution, have been selected as the criteria. Also introduced is a view admissibility concept for a view sequence algorithm. Two parameters (the grouped visible regions and the admissibility of the best viewpoint) are used to find an optimum view sequence for this task.; The view planning concept is expanded to Coordinate Measuring Machine (CMM) measurement. Taking measurement error from the CMM itself into consideration, view planning is carried out.; Regarding system operation, this dissertation also deals with laser-camera structured light systems. Three configurations of structured light systems are described and a comparison between systems is presented. Also described is visibility/occlusion of those system configurations needed to carry out the desired tasks. Because of the system configuration and properties of laser and camera, this sensing method also causes errors. As a result of measurement error analysis, some general guidelines have been suggested.; Depth estimation using a laser-based vision system requires calibration of both the laser and the camera. A new Least-Square Error (LSE) method is compared with a geometric method for laser/camera calibration. Using the calibrated laser/camera system, the depth map of a 3D curved surface is obtained, using striped laser lines as a structured lighting source. As verification, the 3D surface measured by the proposed active sensing is compared with the surface data measured by a Coordinate Measuring Machine (CMM).; Using real three dimensional objects (cuboid and coffee cup), the parametric information of view planning for dimensional inspection as well as object reconstruction tasks is extracted and demonstrated. The implemented three dimensional measurement system is used to demonstrate the upgraded 3D measurement system and suggest one easy way to the solve object/model correspondence problem for inspection.