Summary: | 碩士 === 國立雲林科技大學 === 機械工程系 === 102 === Measuring point data on physical objects to construct CAD model is an important process in reverse engineering, the method of doing this is primarily completing it by 3D scanning technique. When we use 3D scanning technique to measure an object, we can get a large amount of point. However, there’s no way for us to measure the surface of an object completely just in one time, accordingly, we need to measure an object in different angle and then couple with point cloud registration to get complete information about the surface of an object. The main target of the research in this thesis is the method for point cloud registration.
As for the implements utilized in the thesis, ASUS Xtion PRO LIVE(RGB and Depth sensor) is used as hardware implement to measure point clouds from different angle, and PCL(Point Cloud Library)is used as software to process point cloud data. The measurement environment of the research is putting object on a rotary platform, measure the object with sensor in fixed position and get the information of the object from different angle.
The thesis introduces three methods of point cloud registration, which is composed of coordinate transformation, feature matching, ICP (Iterative Closest Point). The above mentioned methods will be implied to point cloud matching experiment and then the advantages and disadvantages for each method will be discussed.
|