| Data acquisition and pre-processing is the first step in reverse engineering. The reliability, usability and integraligy of measurement data can affect the accuracy, efficiency and effectiveness of model reconstruction directly. Therefore, it is meaningful for CAD model reconstruction to research how to acquire topographic data of the object accurately, quickly, completely and rationally in reverse engineering. On the basis of coordinate measuring machine, thorough study on multiple-sensor integration for free-form surface digitization and its data pre-processing was carried out in this dissertation. The research-work is supported by the Key Program of Zhejiang Provincial Natural Science Foundation of China (Grant No. Z1090590), "Research on new method for CAD model reconstruction in reverse engineering based on the theory of semantic data information", and the Major Program of Electromechanical Accoutrement of Science and Technology Committee in Zhejiang Provinence of China (Grant No.2006C11256), "Research on key technique for multiple-sensor integrated CMM and its industrialization".For multiple-sensor integration metrology, a vision system was applied to acquire the edge features of free-form surface to be measured first, with the data acquired by vision system, measurement regions could be planed for further contact or non-contact measurement; then contact or non-contact sensors were used to implement digitizing tasks under the guide of measurement regions.In order to realize high-speed acquisition of free-form surface's edge features, a new 3D measurement approach based on axial stereo vision was presented. The method was characterized by integrating two CCD cameras which arranged one-font, one-back and shared a common optical axial. The depth information of measuring points could be derived from 2D images obtained by the two cameras. The special epipolar geometry within axial stereo vision method that the epipolar lines are parallel to each other and pass through their corresponding main point in image panel was researched. Furthermore, the speciality of epipolar geometry was well applied to simplify image matching procedure in axial stereo vision. Then, based on coordinate measuring machine, one camera axial stereo vision was successfully realized. Finally, the axial stereo vision method was applied to acquire edge features of free-form surface successfully.In view of contact measurement for free-form surface, a new adaptive contact sampling strategy based on front path detecting was presented. The method was characterized by integrating one touch probe and two point laser sensors. The two point laser sensors were located on both sides of the touch probe. During the sampling time, at the measuring direction, the point laser sensors were applied to detect the measuring path information in front of the touch probe. The data acquired by the point laser sensors was processed instantaneously, and fitted into B-spline curves. Then, the curvature feature of the fitted B-spline curves were calculated to generate adaptive sampling path for touch probe. The obtained path information was subsequently used to guide the touch probe to achieve adaptive sampling. While at the cross-section intake direction, curvature analysis algorithm was applied to subdivide the cross-section if it was necessary. Consequently, the sampling points generated by the touch probe were laid out rationally according to the curvature of the surfaces, therefore the sampling efficiency and accuracy can be improved significantly.In view of non-contact measurement for free-form surface, a new approach based on combing two point laser sensors was proposed. In this approach, two point laser sensors were fixed onto a self-designed hinge structure. The two sensors were fixed in an upright position first, and scanned the surface applying front path detecting strategy. At the same time, the derived data was processed immediately to find out inaccuracy regions whose inclination value exceeded the preset range. After the first time measurement, the two point laser sensors were swung to left and right separately, and a new calibration method named "Differences with equal step" was presented to calibrate the laser beam-direction of two point laser sensors respectively. Then the two tilted sensors were used to implement a second measurement to the inaccuracy regions. Finally, the accuracy data generated by different measuring sequence was merged into a whole data body as the final result. Through the proposed approach, free-form surfaces whose angle value between its exterior normal and the positive direction of Z-axis are within [-90°, 90°] can be digitized accurately and completely.For pre-processing of multiple-sensor data, analogizing from semantic web, a novel method of semantic measurement data constructing and measurement data pre-processing based on semantics was proposed. Among the measuring process, by constructing semantic measurement data model, both the 3D coordinate values and the "implicit information" which may useful for data pre-processing of the measuring points could be merged into the final result as semantic measurement data. Among the data pre-processing, semantics of the measurement data would be analyzed first to get the incidental "implicit information". And then, the system would make good use of the "implicit information" to simplify the data pre-processing process. Therefore, the automation, correction and stabilization of the data pre-processing can be improved effectively.Simulation and experiment have been done and the results indicated that with the comprehensive application of the proposed multiple-sensor integration metrology and its data pre-processing approach during the free-form surface digitizing process, the topographic data of the measuring object could be obtained accurately, quickly, rationally and completely. As a result, effective sampling data, excellent approach and fundamental apparatus can be offered for CAD model reconstruction in reverse engineering.
|
PDF Full Text Request