Algorithm Development And Software Design For Three Dimensional Measurements In Industrial CT

Three dimensional(3D) measurement technology is a research direction which combining computer graphics and computational mathematics, and it has broad prospects in applications of computer vision, machine learning and pattern recognition, reverse engineering and computer-assisted areas. There are mainly two categories about this technology including contact type and non-contact type. Traditional contact type such as coordinate measuring machining has high precision. But the internal specifications cannot be available without damaging the work-piece. With the development of computer technology, there emerged non-contact type base on optics, digital imaging and computer vision technology, which can be used to obtain the internal specifications fast. Industrial 3D CT measurement technology mainly obtained the three dimensional information of work-piece with the digital image reconstructed by(Computed Tomography) CT, hence nondestructive measurement can be implemented.Feature extraction and human-computer interaction are two major challenges in 3D measurement technology. In this paper, 3D measurement algorithms based on industrial 3D CT data are studied. With the combination of C++ and open source libraries of open GL and open CV, 3D measurement software is developed, which can behave well in the following four modules: data processing, 3D registration, 3D measurement and true 3D display. In the data processing module, fuzzy region competition method and marching cube(MC) method are utilized to develop a new feature extraction method, which can obtain sub-pixel features of the edges in 3D industrial CT data. In the 3D registration module, initial registration are implemented with the refined data by computer-aided design(CAD) model and sub-pixel features of industrial CT data, then SVD-ICP is used for precise registration between 3D CT data and CAD data, finally the global errors of manufacturing can be calculated. In the 3D measurement module, dot, line and plan based interactive mode is utilized to design the enclosed box interactive tool, which can extract the region of interest(ROI) about the features of CT data. Then further quantitative measurements about ROI such as wall thickness, diameter of holes, cylindricity and the total volume of work-piece cavity can be achieved. In the true 3D display module, parallax images are obtained with human eyes simulation based method to observe sub-pixel features of 3D CT data from different viewing angles. Then 3D video is generated with these images, and the defects and related features of work-piece can be displayed intuitively.