
78 Spectral Transform
(h) k = 3. (g) k = 5. (f) k = 10. (e) k = 50.
(d) k = 100. (c) k = 200. (b) k = 300. (a) Original.
Figure 5.12: Shape reconstruction via spectral analysis using a mesh
Laplacian operator, where k is the number of eigenvectors or spectral
coefficients used. The original model has 7,502 vertices and 15,000 faces.
Figure 5.13: A few visual examples of mesh filtering with plots of the
applied filters shown on the side.
Just as in the 1D classical case, the oscillatory behavior of the
Laplacian eigenvectors and the implied spectral transform of ge-
ometric mesh signals allow us to apply Fourier-like analysis to
mesh models. In addition to ...