Surface Reconstruction from Cross-Sectional Images using the Shrink-Wrapping Algorithm

This paper addresses a new surface reconstruction scheme for approximating the isosurface from a set of tomographic cross sectional images. Differently from the novel Marching cube algorithm, our method does not extract iso-density surface(isosurface) directly from the voxels but calculates the iso-density point(isopoint) first. After building the relatively coarse initial mesh by the Cell-boundary algorithm approximating the isosurface, it produces the final isosurface by iteratively shrinking and smoothing the initial mesh. Comparing with the Marching Cube algorithm, our method is robust and does not make any crack in resulting surface model. Furthermore, the proposed method surmounts the O(1)-adjacency limitation of MC in defining the isopoints by permitting the O(2) and O(3)-adjacent isopoints in surface reconstruction, and can produce more accurate isosurface. According to experiments, it is proved to be very robust and efficient for isosurface reconstruction from cross sectional images.