Global optimization for mesh model reconstructed by industrial CT serial images

The quality of the mesh model is poor by direct-reconstruction from the industrial CT slice images. This paper proposed a global optimization methods based on implicit surface reconstruction without topology restrictions. In this method, 3D surface model was represented by an implicit function, which was calculated first using point cloud information provided by the model, then iso-surface extraction was employed. As for the large data processing problems, we used the fast Fourier transform to improve the efficiency of the algorithm. Experimental results showed that our algorithm could realize mesh denoising, simplifying, smoothing and hole-filling at one time. Compared with the feature-preserving uniform mesh smoothing algorithm, the presented algorithm could effectively remove noise from meshes within a short time. (authors)