抄録
To effectively use scanned meshes obtained by 3D laser scanner or X-ray CT scanning systems in engineering applications, such as inspection and CAD model reconstruction, we need to segment meshes and extract desirable regions and their approximating surfaces. Surfaces of engineering objects are commonly represented by a set of analytic surfaces, such as planes, cylinders, spheres, cones. Therefore, the mesh surface of them needs to be approximated by a set of analytic surfaces. In this paper we propose an efficient mesh segmentation algorithm based on region growing and region merging. Our algorithm first robustly and accurately estimates mesh principal curvatures using the two-pass estimation method. Then it fast extracts regions and their approximating analytic surfaces based on region growing even if regions are smoothly blended. Finally it efficiently merges initial set of regions of large and complex scanned meshes under the user specified tolerances using region merging. We demonstrate the efficiency and accuracy of our algorithm with scanned meshes acquired from real engineering objects by X-ray CT scanning systems.
