Computing Shapes and Their Features from Point Samples

120
Tamal K. Dey The Ohio State University Computing Shapes and Their Features from Point Samples

description

Computing Shapes and Their Features from Point Samples. Tamal K. Dey The Ohio State University. `. Surface Reconstruction. Point Cloud. Surface Reconstruction. Algorithms. Alpha-shapes (Edelsbrunner, M ü cke 94) Crust (Amenta, Bern 98) Natural Neighbors (Boissonnat, Cazals 00) - PowerPoint PPT Presentation

Transcript of Computing Shapes and Their Features from Point Samples

Page 1: Computing Shapes and Their Features from Point Samples

Tamal K. Dey The Ohio State University

Computing Shapes and Their Features from Point Samples

Page 2: Computing Shapes and Their Features from Point Samples

2/52Department of Computer and Information Science

Surface Reconstruction

`

Point Cloud

Surface Reconstruction

Page 3: Computing Shapes and Their Features from Point Samples

3/52Department of Computer and Information Science

Algorithms

1. Alpha-shapes (Edelsbrunner, Mücke 94)

2. Crust (Amenta, Bern 98)

3. Natural Neighbors (Boissonnat, Cazals 00)

4. Cocone (Amenta, Choi, Dey, Leekha, 00)

5. Tight Cocone (Dey, Goswami, 02)

6. Power Crust (Amenta, Choi, Kolluri 01)

Page 4: Computing Shapes and Their Features from Point Samples

4/52Department of Computer and Information Science

Basic Topology

d-ball Bd {x in Rd | ||x|| ≤ 1} d-sphere Sd {x in Rd+1 | ||x||=1}

Homeomorphism h: T1 → T2 where

h is continuous, bijective and has continuous inverse

K-Manifold : neighborhoods homeomorphic to open k-ball2-sphere, torus, double torus are 2-manifolds

K-manifold with boundary: interior points, boundary pointsBd is a d-manifold with boundary where bd(Bd)=S(d-1)

Page 5: Computing Shapes and Their Features from Point Samples

5/52Department of Computer and Information Science

Basic Topology Smooth Manifolds Triangulation

k-simplex Simplicial complex K: (i) t in K if t is a face of t' in

K

(ii) t1, t2 in K => t1∩ t2 face of both

K is a triangulation of a topological space T if

T ≈ |K|

Page 6: Computing Shapes and Their Features from Point Samples

6/52Department of Computer and Information Science

Sampling

Page 7: Computing Shapes and Their Features from Point Samples

7/52Department of Computer and Information Science

Medial Axis

Page 8: Computing Shapes and Their Features from Point Samples

8/52Department of Computer and Information Science

f(x) is the distance to medial axis

Local Feature SizeAmenta-Bern-Eppstein 98

f(x)

Page 9: Computing Shapes and Their Features from Point Samples

9/52Department of Computer and Information Science

Each x has a sample within f(x) distance

-samplingAmenta-Bern-Eppstein 98

x

Page 10: Computing Shapes and Their Features from Point Samples

10/52Department of Computer and Information Science

-sampleε-sample is also ε'-sample for ε' > ε

Page 11: Computing Shapes and Their Features from Point Samples

11/52Department of Computer and Information Science

Lipschitz Property of f()Lemma (Lipschitz Continuity):

f(x) ≤ f(y) + ||x-y||

Proof: Let m be a point on M with f(y)=||y-m||

By triangular inequality

||x-m|| ≤ ||y-m|| + ||x-y||

f(x) ≤ ||x-m|| ≤ f(y)+||x-y||

Page 12: Computing Shapes and Their Features from Point Samples

12/52Department of Computer and Information Science

FTL: Feature translation lemma

Lemma (Feature Translation): If ||x-y|| ≤ ε f(x) then

1/(1+ ε)f(y) ≤ f(x) ≤ 1/(1- ε)f(y)

Exercise 1: Prove it. Also prove ||x-y|| ≤ ε /(1-

ε)f(y)

Page 13: Computing Shapes and Their Features from Point Samples

13/52Department of Computer and Information Science

FBL: Feature Ball LemmaLemma (Feature Ball): If a d-ball B intersects a k-

manifold Σ at more than two points with either (i) B∩Σ is not a k-ball, or (ii) bd(B)∩Σ is not a (k-1)-sphere, then B contains a medial axis point. Exercise 2: Prove

Page 14: Computing Shapes and Their Features from Point Samples

14/52Department of Computer and Information Science

Voronoi/Delaunay Diagrams

Voronoi diagram VP: collection of Voronoi cells {Vp}

Vp={x in R3 | ||x-p|| ≤ ||x-q|| for all q in P}

Voronoi facet, Voronoi edge, Voronoi vertex

Delaunay triangulation DP: Dual of VP, a simplicial complex

Delaunay edge, triangle, tetrahedra

Page 15: Computing Shapes and Their Features from Point Samples

15/52Department of Computer and Information Science

Delaunay propertiesEmptiness : A simplex t is in DP if and only if there is

a circumscribing ball of t that does not contain any point of P inside.

Proof: If a k-simplex t, 0 ≤ k ≤ 3, is in a DP, its dual (3-k)-dimensional Voronoi element has a point x that is equidistant from the (k+1) vertices of t. Also these vertices are closest to x among all points of P. This only means the ball centered at x with the vertices of t on the bounding sphere is empty.

Exercise 3: Show that if t has an empty circumball, t is in DP

Page 16: Computing Shapes and Their Features from Point Samples

16/52Department of Computer and Information Science

Restricted Voronoi/Delaunay

Restricted Voronoi: VP,Σ = {Vp Σ =Vp∩ Σ | p in P}

Restricted Delaunay: Dp, Σ ={A k-simplex is Conv R where

∩Vp, Σ ≠ Ø for p in R}

Page 17: Computing Shapes and Their Features from Point Samples

17/52Department of Computer and Information Science

Curve samples and Voronoi

Page 18: Computing Shapes and Their Features from Point Samples

18/52Department of Computer and Information Science

Crust Algorithm (2D)Amenta-Bern-Eppstein 98

Compute VP

Add Voronoi vertices

Compute Delaunay

Retain edges between

samples only

Page 19: Computing Shapes and Their Features from Point Samples

19/52Department of Computer and Information Science

Nearest Neighbor AlgorithmDey-Kumar 99

Compute DP

For each p, compute nearest neighbor

For each p, compute its half-neighbor.

Page 20: Computing Shapes and Their Features from Point Samples

20/52Department of Computer and Information Science

Difficulties in 3DVoronoi vertices can come

close to the surface… slivers are nasty.

There is no unique `correct’ surface for reference

or

or……

Voronoi vertex

Page 21: Computing Shapes and Their Features from Point Samples

21/52Department of Computer and Information Science

Normals and Voronoi Cells(3D) Amenta-Bern 98

Page 22: Computing Shapes and Their Features from Point Samples

22/52Department of Computer and Information Science

Long Voronoi cellsLemma (Medial): Let m1 and m2 be the centers of two medial balls at p. Vp contains m1, m2.Exercise 4: Prove it

Page 23: Computing Shapes and Their Features from Point Samples

23/52Department of Computer and Information Science

NL : Normal LemmaLemma (Normal) : Let v be a point in Vp with ||v-p||>μf(p). Then,

angle((v-p),np)≤ arcsin ε/μ(1- ε) + arcsin ε /(1- ε). Exercise 5: Prove NL

Page 24: Computing Shapes and Their Features from Point Samples

24/52Department of Computer and Information Science

NVL: Normal Variation lemma

Lemma (Normal Variation) : Let x and y be two points with ||x-y||≤r f(x) for r< 1/3. Then, angle(nx,ny) ≤ r/(1-3r).

Page 25: Computing Shapes and Their Features from Point Samples

25/52Department of Computer and Information Science

ENL: Edge Normal LemmaLemma (Edge Normal): angle((p-q),np) > /2 – arcsin ||p-q||/2f(p).

Proof: sin θ = ||p-q||/2||m-p|| ≤ ||p-q||/2f(p)

Page 26: Computing Shapes and Their Features from Point Samples

26/52Department of Computer and Information Science

TNL: Triangle Normal Lemma

Lemma (Triangle Normal) : angle(npqr,np) ≤ α + arcsin((2/√3) sin 2α)where α ≤ arcsin d/f(p) and d, the circumradius, is sufficiently small.

Page 27: Computing Shapes and Their Features from Point Samples

27/52Department of Computer and Information Science

TopologyClosed Ball property (Edelsbrunner, Shah 94): If restricted Voronoi cellis a closed ball in each dimension, then DP, Σ is homeomorphic to Σ.

Assume P is an e-sample of Σ where e is sufficiently small. It can be shown that (P, Σ) satisfies the closed ball property.(proof from Cheng-Dey-Edelsbrunner-Sullivan 02)

Page 28: Computing Shapes and Their Features from Point Samples

28/52Department of Computer and Information Science

SDL: Short Distance Lemma

Lemma Short Distance : x, y two points in Vp,Σ. (i) ||x-p||< ε/(1- ε)f(p)(ii) ||x-y|| < 2ε/(1- ε)f(x).

Exercise 6: Prove it.

Page 29: Computing Shapes and Their Features from Point Samples

29/52Department of Computer and Information Science

LDL: Long Distance Lemma

Lemma (Long Distance) : Suppose L intersects S in two points x, y andmakes angle less than ξ with nx. Then ||x-y||>2f(x)cos ξ.

Page 30: Computing Shapes and Their Features from Point Samples

30/52Department of Computer and Information Science

VEL: Voronoi Edge LemmaLemma (Voronoi Edge) : A Voronoi edge intersects Σ in a single point.

Proof: x ≤ angle(npqr,np) + angle(np,nx) ≤ O(ε) + O(ε) by TNL and NVL.

2f(x)cos O(ε) ≤ ||x-y|| ≤ O(ε) f(x)by SDL and LDL

Contradiction when ε is sufficiently smallExercise 7: Prove it

Page 31: Computing Shapes and Their Features from Point Samples

31/52Department of Computer and Information Science

VFL: Voronoi Facet LemmaLemma (Voronoi Facet): A Voronoi facet intersects Σ in a 1-ball.

Proof: angle(Lnx)≤ angle(Lnp) + angle(np,nx)

≤ O(ε) + O(ε) by ENL and NVL. 2f(x)cos O(ε) ≤ ||x-y|| ≤ O(ε) f(x)

by SDL and LDL Contradiction when ε is sufficiently small

Page 32: Computing Shapes and Their Features from Point Samples

32/52Department of Computer and Information Science

VCL: Voronoi Cell LemmaLemma (Voronoi Cell) : A Voronoi cell intersects Σ in a 2-ball.

Proof: show that handles and connected components of Σ cannot be in the cell. Then show that if the cell intersects Σ in multiple disks, we reach a contradiction with SDL and LDL.

Page 33: Computing Shapes and Their Features from Point Samples

33/52Department of Computer and Information Science

PolesP+

P-

Page 34: Computing Shapes and Their Features from Point Samples

34/52Department of Computer and Information Science

PVL: Pole Vector LemmaP+

P-

npvp

Lemma (Pole Vector) : angle((p+-p),np)=2arcsin ε /(1- ε).Proof: ||p+-p||> f(p) since Vp contains a medial axis point (Medial Lemma). Plug this in Normal Lemma.

Page 35: Computing Shapes and Their Features from Point Samples

35/52Department of Computer and Information Science

Crust in 3DAmenta-Bern 98

Introduce poles

Filter crust triangles from Delaunay

Filter by normals

Extract manifold

Page 36: Computing Shapes and Their Features from Point Samples

36/52Department of Computer and Information Science

Manifold Extraction: Prunning

Remove Sharp edges with their triangles

Page 37: Computing Shapes and Their Features from Point Samples

37/52Department of Computer and Information Science

Why Prunning Works?Crust triangles include restricted Delaunay triangles

The underlying space of the restricted Delaunay

triangles is homeomorphic to the sampled surface

No edge of the restricted triangles is sharp

After prunning, at least the surface made by the

restricted Delaunay triangles remains

Page 38: Computing Shapes and Their Features from Point Samples

38/52Department of Computer and Information Science

Manifold Extraction: Walk

Walk inside or outside the possibly thickened surface

Page 39: Computing Shapes and Their Features from Point Samples

39/52Department of Computer and Information Science

Cocone AlgorithmAmenta-Choi-Dey-Leekha 00

Simplified/improved the Crust

Only single Voronoi computation

Analysis is simpler

No normal filtering step

Proof of homeomorphism

Page 40: Computing Shapes and Their Features from Point Samples

40/52Department of Computer and Information Science

Cocone

vp= p+ - p is the pole vector

Space spanned by vectors

within the Voronoi cell making

angle > 3/8 with vp or -vp

Page 41: Computing Shapes and Their Features from Point Samples

41/52Department of Computer and Information Science

Cocone Algorithm

Page 42: Computing Shapes and Their Features from Point Samples

42/52Department of Computer and Information Science

Candidate triangles computatione=(a,b); a=a-p; b= b-p

Page 43: Computing Shapes and Their Features from Point Samples

43/52Department of Computer and Information Science

Candidate Triangle Properties

Candidate triangles include the restricted

Delaunay triangles

Their circumradii are small O()f(p)

Their normals make only O() angle with the

surface normals at the vertices

Page 44: Computing Shapes and Their Features from Point Samples

44/52Department of Computer and Information Science

Restricted Delaunay property

Claim: Let y in Vp∩ Σ . Then, angle(np,(y-p)) > /2- εExercise 8: Prove it.

Lemma (Restricted Delaunay): All restricted triangles are in T for ε <0.1.Proof: Let y in e∩Σ where e is the dual edge for a triangle.

angle((y-p),vp)>angle((y-p),np)-angle(np,vp)

> /2- ε -angle(np,vp)

> 3/8 by PVL for ε < 0.1.

Page 45: Computing Shapes and Their Features from Point Samples

45/52Department of Computer and Information Science

No sharp edge

Lemma Sharp: No restricted Delaunay triangle has a sharp edge for ε < 0.06

Page 46: Computing Shapes and Their Features from Point Samples

46/52Department of Computer and Information Science

Small radius and flatnessLemma (Small Triangle): The circumradius r of any candidate triangle is O(ε)f(p) where p is any of its vertex and ε < 0.06.

Proof: There is y in dual edge so that angle(vp,(y-p))>3/8.By PVL angle(np,(y-p)) > 3/8-2arcsin ε /(1- ε).Use contrapositive of NL to conclude ||y-p||=O(ε)f(p) for ε <0.06.

Lemma (Flat Triangle): For each candidate triangle pqrangle(npqr,np)=O(ε)

Proof: Follows from STL and TNL.

Page 47: Computing Shapes and Their Features from Point Samples

47/52Department of Computer and Information Science

Homeomorphism

Let M be the triangulated surface obtained after the manifold extraction.

Define h: R3 -> Σ where h(q) is the closest point on Σ. h is welldefined except at the medial axis points.

Lemma Homeomorphism: The restriction of h to M, h: M -> Σ,is a homeomorphism.

Proof: Use STL, FTL for the proof, see ACDL00.

Page 48: Computing Shapes and Their Features from Point Samples

48/52Department of Computer and Information Science

Cocone GuaranteesTheorem:

Any point x is within O(f(x) distance from a point in the output. Conversely, any point of the output surface has a point x within O()f(x) distance for ε <0.06.

Theorem:

The output surface computed by Cocone from an -sample is homeomorphic to the sampled surface for ε < 0.06.

Page 49: Computing Shapes and Their Features from Point Samples

49/52Department of Computer and Information Science

Undersampling Dey-Giesen 01

Boundaries

Small features

Non-smoothness

Page 50: Computing Shapes and Their Features from Point Samples

50/52Department of Computer and Information Science

Boundaries

Page 51: Computing Shapes and Their Features from Point Samples

51/52Department of Computer and Information Science

Small Features High curvature regions are often undersampled

Page 52: Computing Shapes and Their Features from Point Samples

52/52Department of Computer and Information Science

Well Sampled Patch and Boundary Vertices

is well sampled if ε-sampling

holds for

Restricted Voronoi on defines boundary vertices

p is interior if restricted cell has no boundary point otherwise p is boundary vertex

Page 53: Computing Shapes and Their Features from Point Samples

53/52Department of Computer and Information Science

Radius and HeightRadius r(p): radius of cocone

Height h(p): distance to the negative pole p-

cocone neighbors Np

Page 54: Computing Shapes and Their Features from Point Samples

54/52Department of Computer and Information Science

Flatness Condition

Vertex p is flat if

1. Ratio condition: r(p) h(p)

2. Normal condition: vp,vq q with pNq

Page 55: Computing Shapes and Their Features from Point Samples

55/52Department of Computer and Information Science

Boundary Detection(1st phase)

IsFlat(p,,) check ratio and normal condition for Vp;

if both are satisfied

return true

else

return false

end

Page 56: Computing Shapes and Their Features from Point Samples

56/52Department of Computer and Information Science

Boundary Detection(2nd phase)

Boundary(P,,) Compute the set R of flat vertices; while pR and pNq with qR and r(p)h(p) and vp,vq R:=Rp; endwhile return P\Rend

Page 57: Computing Shapes and Their Features from Point Samples

57/52Department of Computer and Information Science

ReconstructionCocone(P, ,)

Compute VP;

for each pP if pB compute T of triangles with duals intersecting Cp; endifenfor;

Extract manifold;end

B:= Boundary(P,,)

Page 58: Computing Shapes and Their Features from Point Samples

58/52Department of Computer and Information Science

Data Set Sat

Page 59: Computing Shapes and Their Features from Point Samples

59/52Department of Computer and Information Science

Data Set Engine

Page 60: Computing Shapes and Their Features from Point Samples

60/52Department of Computer and Information Science

Nonsmoothness

Page 61: Computing Shapes and Their Features from Point Samples

61/52Department of Computer and Information Science

Watertight Surfaces

Page 62: Computing Shapes and Their Features from Point Samples

62/52Department of Computer and Information Science

Tight Cocone Dey-Goswami 02

Page 63: Computing Shapes and Their Features from Point Samples

63/52Department of Computer and Information Science

Tight COCONE Principle Compute the Delaunay triangulation of the input point set.

Use COCONE along with detection of undersampling to get an initial

surface with undersampled regions identified.

Stitch the holes from the existing Delaunay triangles without inserting any

new point.

Effectively, the output surface bounds one or more solids.

Page 64: Computing Shapes and Their Features from Point Samples

64/52Department of Computer and Information Science

Result Sharp corners and edges of AutoPart can be reconstructed.

Page 65: Computing Shapes and Their Features from Point Samples

65/52Department of Computer and Information Science

Dinosaur

Page 66: Computing Shapes and Their Features from Point Samples

66/52Department of Computer and Information Science

Large Data

Octree subdivision

Page 67: Computing Shapes and Their Features from Point Samples

67/52Department of Computer and Information Science

Cracks Cracks appear in surface computed from octree boxes

Page 68: Computing Shapes and Their Features from Point Samples

68/52Department of Computer and Information Science

Padding Include a fraction from the neighbors to form the extended box

Page 69: Computing Shapes and Their Features from Point Samples

69/52Department of Computer and Information Science

Surface Matching

Page 70: Computing Shapes and Their Features from Point Samples

70/52Department of Computer and Information Science

Experimental DataPentium III,733Mhz,512Mb

Time

Memory

Page 71: Computing Shapes and Their Features from Point Samples

71/52Department of Computer and Information Science

Lucy25

3.5 million points, 198 mints

Page 72: Computing Shapes and Their Features from Point Samples

72/52Department of Computer and Information Science

David’s Head

2 mil points, 93 minutes

Page 73: Computing Shapes and Their Features from Point Samples

73/52Department of Computer and Information Science

Noisy Data - Bunny

Front view Rear view

Page 74: Computing Shapes and Their Features from Point Samples

74/52Department of Computer and Information Science

Noisy Data – Ram Head

Front view Rear view

Page 75: Computing Shapes and Their Features from Point Samples

75/52Department of Computer and Information Science

Example movie file

Mannequin

Page 76: Computing Shapes and Their Features from Point Samples

76/52Department of Computer and Information Science

• Female

Point data Tight Cocone Robust Cocone

Examples

Page 77: Computing Shapes and Their Features from Point Samples

77/52Department of Computer and Information Science

• Mannequin

Point data Tight Cocone Robust Cocone

Examples

Page 78: Computing Shapes and Their Features from Point Samples

78/52Department of Computer and Information Science

Cocone Software

Cocone: Reconstructs surfaces with boundaries.

Tight Cocone: Reconstructs watertight surfaces.

Available fromhttp://www.cis.ohio-state.edu/~tamaldey/cocone.html

Acknowledgement: CGAL

Page 79: Computing Shapes and Their Features from Point Samples

79/52Department of Computer and Information Science

Medial axis from point sample

• Earlier work did not have guarantees[Attali-Montanvert-Lachaud 01]

• Power shape : guarantees topology, uses power diagram[Amenta-Choi-Kolluri 01]

• Medial : Approximates the medial axis as a Voronoi subcomplex and has converegence guarantee.[Dey-Zhao 02]

Page 80: Computing Shapes and Their Features from Point Samples

80/52Department of Computer and Information Science

Medial Axis• Medial Ball• Medial Axis -Sampling

Page 81: Computing Shapes and Their Features from Point Samples

81/52Department of Computer and Information Science

Geometric Definitions

• Pole and Pole Vector

• Tangent Polygon • Umbrella Up

Page 82: Computing Shapes and Their Features from Point Samples

82/52Department of Computer and Information Science

Filtering conditions

• Medial axis point m• Medial angle θ• Angle and Ratio

Conditions

Our goal: : approximate the medial axis as a approximate the medial axis as a subset of Voronoi facetssubset of Voronoi facets..

Page 83: Computing Shapes and Their Features from Point Samples

83/52Department of Computer and Information Science

Angle Condition• Angle Condition

[θ ]:

Max{σ in Up angle(nσ,(q-p)) }< /2- θ

Page 84: Computing Shapes and Their Features from Point Samples

84/52Department of Computer and Information Science

‘Only Angle Condition’ Results

= 18 degrees

= 3 degrees = 32 degrees

Page 85: Computing Shapes and Their Features from Point Samples

85/52Department of Computer and Information Science

‘Only Angle Condition’ Results

= 15 degrees

= 20 degrees

= 30 degrees

Page 86: Computing Shapes and Their Features from Point Samples

86/52Department of Computer and Information Science

Ratio Condition

• Ratio Condition []:

Min{σ in Up ||p-q||/Rσ >

Page 87: Computing Shapes and Their Features from Point Samples

87/52Department of Computer and Information Science

‘Only Ratio Condition’ Results

= 2

= 4

= 8

Page 88: Computing Shapes and Their Features from Point Samples

88/52Department of Computer and Information Science

‘Only Ratio Condition’ Results

= 2

= 4

= 6

Page 89: Computing Shapes and Their Features from Point Samples

89/52Department of Computer and Information Science

Algorithm• Each of Angle and Ratio conditions

individually is not sufficient. • Combination of both conditions

• First Angle, then Ratio• The Angle condition captures the Delaunay edges

which lie away from the surface.• The Ratio condition captures the the Delaunay edges

which make small angles with the umbrella triangles but are comparatively larger than their circumradii.

• Allows fixed values of θ and

Page 90: Computing Shapes and Their Features from Point Samples

90/52Department of Computer and Information Science

Algorithm

)cloure(Output 11endfor10

endfor9endif8

Dual78 Condtion or Ratio Condition Angle satisfies if6

edgeDelaunayeachfor5; Compute4

eachfor3;2

;and Compute1

8

F

pqF:Fpq

UpqUPp

FDV

P

p

p

PP

)(M EDIAL

Page 91: Computing Shapes and Their Features from Point Samples

91/52Department of Computer and Information Science

AnalysisLemma 1: If w lies in the segment mm’, 2 tan(arcsin2 ) .p pw w

p = O ()

S

m1

m’

Page 92: Computing Shapes and Their Features from Point Samples

92/52Department of Computer and Information Science

Analysis (continued)

Lemma 2: Let F = Dual pq be a Voronoi facet

where pq satisfies the angle condition [ ] with

2p + p. Any point w in F with

is at a distance

from m where m and are the center and radius

of the medial ball at p with .

w p 1 (2 tan(arcsin 2 ))

sin( ) p pp

0pm pwt t

p, p = O ()

>( - p )

Page 93: Computing Shapes and Their Features from Point Samples

93/52Department of Computer and Information Science

Analysis (continued)Lemma 3: Let w be any point in the Voronoi facet F=Dual pq with

where is the radius of the medial ball at p with center m so that .

If pq satisfies the Ratio condition then either

0pm pwt t

0[ 8]

1/8( )p pw m O or

1/8( ) .pw p

w p

Proof

If , Lemma 2 gives 7 /8p 1/8( )p pO

If , ……7 /8p 1/8( )p

Page 94: Computing Shapes and Their Features from Point Samples

94/52Department of Computer and Information Science

Analysis (continued)

Lemma 4: Let w Vp be a point such that , where is the

radius of a medial ball at p with the center m and , and .

Then, for sufficiently small >0, if the medial angle of p

is larger than 1/3.

w p

0pm pwt t 1/ 4

3/ 4( )w m O

Page 95: Computing Shapes and Their Features from Point Samples

95/52Department of Computer and Information Science

TheoremsTheorem:

if by Angle: (i) , apply Lemma 4

Otherwise Lemma 2

if by Ratio: (i) , apply Lemma 4

Otherwise Lemma 3

Theorem:

0lim .L M

0lim .M L

w p

w p

Page 96: Computing Shapes and Their Features from Point Samples

96/52Department of Computer and Information Science

Experimental Results

Page 97: Computing Shapes and Their Features from Point Samples

97/52Department of Computer and Information Science

Experimental Results

Page 98: Computing Shapes and Their Features from Point Samples

98/52Department of Computer and Information Science

Experimental Results

Page 99: Computing Shapes and Their Features from Point Samples

99/52Department of Computer and Information Science

Experimental Results

Page 100: Computing Shapes and Their Features from Point Samples

100/52Department of Computer and Information Science

Medial Axis from a CAD model

CAD model

Point Sampling Medial Axis

Page 101: Computing Shapes and Their Features from Point Samples

101/52Department of Computer and Information Science

Medial Axis

Medial Axis from a CAD model

CAD model

Point Sampling

Page 102: Computing Shapes and Their Features from Point Samples

102/52Department of Computer and Information Science

Medial Axis

Medial Axis from a CAD model

CAD model

Point Sampling

Page 103: Computing Shapes and Their Features from Point Samples

103/52Department of Computer and Information Science

Example movie file

Anchor Medial

Page 104: Computing Shapes and Their Features from Point Samples

104/52Department of Computer and Information Science

Segmentation and matchingDey-Giesen-Goswami 03

• Segment a shape into `features’• Match two shapes based on the

segmentation

Page 105: Computing Shapes and Their Features from Point Samples

105/52Department of Computer and Information Science

Feature definitionFlow

Continuous

Discrete flow

Discretization

Page 106: Computing Shapes and Their Features from Point Samples

106/52Department of Computer and Information Science

FlowShape :

dp xxpxh R allfor inf)( 2

xpxA p minarg)(• Anchor set:

• Height fuinction:

Page 107: Computing Shapes and Their Features from Point Samples

107/52Department of Computer and Information Science

Flow

• Vector field v :

)()()(

xdxxdxxv

if x is regular and 0 otherwise

• Flow induced by v Fix points of are the critical points of h

Page 108: Computing Shapes and Their Features from Point Samples

108/52Department of Computer and Information Science

Features• F(x) = closure(S(x)) for a maximum x

Page 109: Computing Shapes and Their Features from Point Samples

109/52Department of Computer and Information Science

Flow by discrete set• Driver d(x): closest point on dual to the Voronoi

object containing x•

Vector field:

• This also induces a flow )()()(

xdxxdxxv

Page 110: Computing Shapes and Their Features from Point Samples

110/52Department of Computer and Information Science

Stable manifolds

• Gabriel edges are stable manifolds of saddles• Stable manifolds of maxima are shaded

Page 111: Computing Shapes and Their Features from Point Samples

111/52Department of Computer and Information Science

Stable manifolds• Feature F(x) = closure(S(x)) for a maximum x

Page 112: Computing Shapes and Their Features from Point Samples

112/52Department of Computer and Information Science

Flow relation• t < t’ if the circumcenters of t and t’ lie on the same side of

the edge shared by them.• Collect all triangles related by the transitive closure of <

Page 113: Computing Shapes and Their Features from Point Samples

113/52Department of Computer and Information Science

Flow relation in 3D• In 2D there is at most one t’ so that t< t’• Exercise 9: Show an example in 3D where a tetrahedron t <

t1 and t < t2. Show that there cannot be any third tetrahedron t3 so that t< t3.

Page 114: Computing Shapes and Their Features from Point Samples

114/52Department of Computer and Information Science

Algorithm for )(~ xF

Page 115: Computing Shapes and Their Features from Point Samples

115/52Department of Computer and Information Science

Merging• Small perturbations create insignificant features• Sampling artifacts introduce more segmentations

• Merge stable manifolds

Page 116: Computing Shapes and Their Features from Point Samples

116/52Department of Computer and Information Science

Results (2D)

Page 117: Computing Shapes and Their Features from Point Samples

117/52Department of Computer and Information Science

Results (3D)

Page 118: Computing Shapes and Their Features from Point Samples

118/52Department of Computer and Information Science

Results (2D)

Page 119: Computing Shapes and Their Features from Point Samples

119/52Department of Computer and Information Science

Results (3D)

Page 120: Computing Shapes and Their Features from Point Samples

120/52Department of Computer and Information Science

Open problems• Design an algorithm to reconstruct non-smooth surface

• Design an algorithm for medial axis approximaionwith topological guarantee

• Prove an approximation result for feature segmentation

Software: Cocone, Medial : http://www.cis.ohio-state.edu/~tamaldey/cocone.htmlSegmatch: http://www.cis.ohio-state.edu/~tamaldey/segmatch.html