zl1 A sharpness dependent filter for mesh smoothing Chun-Yen Chen Kuo-Young Cheng available in CAGD Vol.22. 5(2005)
zl2 Outline Introduction about authors Introduction about works Sharpness dependent filter Sharpness dependent weighting function Comparison Conclusion
zl3 Introduction about authors Chun-Yen Chen, Kuo-Young Cheng Institute of Information Science, Academia Sinica, Nankang, Taipei Department of Computer Science and Information Engineering, National Taiwan University Computer Graphics, Chinese Processing
zl4 Outline Introduction about authors Introduction about works Sharpness dependent filter Sharpness dependent weighting function Comparison Conclusion
zl5 Introduction about works Mesh straight-line graph embedded in R³ a pair (K, V), where K is a simplicial complex representing the connectivity of vertices, edges, and faces and V=( ) describes the geo- metric positions of the vertices in R³ … …
zl6 Introduction about works Mesh
zl7 Introduction about works Mesh smoothing problem arising creating high-fidelity computer graphics objects using imperfectly-measured data from the real world
zl8 Introduction about works Mesh smoothing main task adjusting the position of mesh vertex the to remove undesirable noise and uneven edges while retaining desirable geometric features
zl9 Introduction about works Mesh smoothing regarded as a filter design problem, to remove high-frequency tiny part on surfaces filter : a function or a procedure which remove unwanted parts of a signal Taubin, G., A signal processing approach to fair surface design. Siggraph ’ 95. Taubin, G., Optimal surface smoothing as filter design. Research Report RC IBM Thomas J.Watson Research Center.
zl10 Introduction about works Mesh smoothing dilemma how can one get rid of the noise by smoothing the surface, while preserving sharp edge to keep the underlying geometry intact or feature?
zl11 Introduction about works Related works Notations mesh S={V, F}, where V and F are the sets of vertices and faces, respectively vertex element, face element collection of neighboring vertices of vertex Laplacian operator
zl12 Introduction about works Laplacian smoothing (Taubin, 1995, 2000) adjust vertex for smoothingCompensate shrinkage
zl13 Introduction about works Laplacian smoothing (Taubin, 1995, 2000) iterative process anti-shrinkage good overall smoothing, bad feature preserving 200 smoothing steps100 smoothing steps
zl14 Introduction about works MCF (Mean Curvature Flow) isotropic filter design (Desbrun et al., 1999) mean curvature, discrete mean curvature operator
zl15 Introduction about works MCF (Mean Curvature Flow) isotropic filter design (Desbrun et al., 1999) new vertex position
zl16 Introduction about works MCF (Mean Curvature Flow) anisotropic filter design (Meyer et al., 2002)
zl17 Introduction about works MCF (Mean Curvature Flow) isotropic filter feature non-preserving anisotropic filter feature preserving
zl18 Introduction about works Bilateral Filter (Fleishman, et al., 2003; Jones et al., 2003)
zl19 Introduction about works Mean-filter design (Ohtake et al., 2001) surface normal based compute weighted average normal
zl20 Introduction about works Mean-filter design (Ohtake et al., 2001) surface normal based update each vertex
zl21 Introduction about works Mean-filter design (Ohtake et al., 2001) feature non-preserving
zl22 Introduction about works Median-filter design (Yagou et al., 2002) surface normal based compute weighted average normal
zl23 Introduction about works Median-filter design (Yagou et al., 2002) surface normal based adjust normal choose as media angle in N(T) replace m(T) by m( )
zl24 Introduction about works Median-filter design (Yagou et al., 2002) surface normal based update each vertex
zl25 Introduction about works Median-filter design (Yagou et al., 2002) feature preserving
zl26 Introduction about works Remark mean-filter flat region median-filteredge min-filtercorner (Gonzalez, Woods, 2002) to smooth mesh appropriately, combine filters above together
zl27 Introduction about works This paper propose a sharpness dependent filter design based on the fairing of surface normal, selecting a mean-filter for flat region and a min-filter for sharp region automatically
zl28 Introduction about works This paper
zl29 Outline Introduction about authors Introduction about works Sharpness dependent filter Sharpness dependent weighting function Comparison Conclusion
zl30 Sharpness dependent filter Basic concepts sharpness a measure of the distribution of the included angles between polygon face normals
zl31 Sharpness dependent filter Basic concepts sharpness dependent weighting function defined as the distribution of sharpness cutoff value of sharp criteria for sharp and non-sharp, derived by Bayesian classification (Chen, et al., 2004)
zl32 Sharpness dependent filter Algorithm 1. compute mean normal for each polygon face is the No. of neighboring faces of
zl33 Sharpness dependent filter Algorithm 2. determine the closet face normal,, for each as follows calculate the angle between normals normalized in a range [0,1]
zl34 Sharpness dependent filter Algorithm 2. determine the closet face normal,, for each as follows find the minimum value of
zl35 Sharpness dependent filter Algorithm 3. Calculate the local sharpness
zl36 Sharpness dependent filter Algorithm 4. compute a new face normal user-defined sharpness dependent weighting functon
zl37 Sharpness dependent filter Algorithm 5. update each vertex position area weight contributed by
zl38 Sharpness dependent filter Algorithm 6. proceed to next iteration step until a steady state, i.e., is a preset tolerance
zl39 Sharpness dependent filter Remark we ’ ve got a filter design for mesh smoothing based on the weighting function defined by sharpness mean-filter min-filer
zl40 Sharpness dependent filter Remark how to select weighing function ?
zl41 Outline Introduction about authors Introduction about works Sharpness dependent filter Sharpness dependent weighting function Comparison Conclusion
zl42 Sharpness dependent weighting function Selection principle experiment to compare sharpness distribution of most noisy models Fandisk
zl43 Sharpness dependent weighting function Selection principle experiment to compare sharpness distribution of most noisy models Two-hole structure
zl44 Sharpness dependent weighting function Selection principle experiment to compare sharpness distribution of most noisy models Golf driver head
zl45 Sharpness dependent weighting function Selection principle monotonic decreasing function, vanishing beyond the cutoff of sharpness Gaussian function Laplacian function El Fallah Ford function
zl46 Sharpness dependent weighting function Selection principle monotonic decreasing function, vanishing beyond the cutoff of sharpness
zl47 Sharpness dependent weighting function selection user-defined, chosen such that for large sharpness and for small sharpness Remember cutoff value of sharpness for sharp and non-sharp obtained by applying Bayesian classification?
zl48 Sharpness dependent weighting function selection user-defined, chosen such that for large sharpness and for small sharpness obtain the best cutoff,, should be small when Gaussian weighting function
zl49 Sharpness dependent weighting function selection sharpness factor to control degree of sharpness for feature preserving
zl50 Sharpness dependent weighting function Remark sharpness factor controls the degress of sharpness for feature preserving, non-feature preserving the larger, the stronger the feature preserving
zl51 Outline Introduction about authors Introduction about works Sharpness dependent filter Sharpness dependent weighting function Comparison Conclusion
zl52 Comparison Different sharpness factor sf=0, 10, 15, 20
zl53 Comparison With other feature preserving filter Like anisotropic MCF bilateral filter median filter
zl54 Comparison With other feature preserving filter Fandisk model A MCFBilateralMedian Sharpness Gaussian, sh=23.4, 16 steps
zl55 Comparison With other feature preserving filter Two hole structure Laplacian, sh=32, 97 steps
zl56 Comparison With other feature preserving filter Golf driver head Bilateral
zl57 Comparison With other feature preserving filter Golf driver head A MCF
zl58 Comparison With other feature preserving filter Golf driver head Sharpness
zl59 Comparison With other feature preserving filter Guardian lion Bilateral
zl60 Comparison With other feature preserving filter Guardian lion A MFC
zl61 Comparison With other feature preserving filter Guardian lion Sharpness
zl62 Comparison How about shrinkage? Little volume shrinkage, nearly intact
zl63 Comparison How about execution time? 2.8 GHz Pentium 4 processor with 1 GB RAM
zl64 Outline Introduction about authors Introduction about works Sharpness dependent filter Sharpness dependent weighting function Comparison Conclusion
zl65 Conclusion Highlights Define sharpness to measure feature areas of models Use sharpness dependent weighting function to automatically select filter to smooth for different feature Experiments to evaluate weighting function
zl66 ThankU