31
1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif Kobbelt Jeffrey Sukharev CMPS260 Final Project

1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

Embed Size (px)

Citation preview

Page 1: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

1

efficient simplification of point-sampled geometry

From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif Kobbelt

Jeffrey Sukharev CMPS260 Final Project

Page 2: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

2

outlineintroductionsurface model & local surface analysispoint cloud simplification

hierarchical clustering iterative simplification particle simulation

measuring surface errorcomparisonconclusions

Page 3: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

3

introduction

3d content creation

acquisition renderingprocessing

many applications require coarser approximationseditingrendering

surface simplification for complexity reduction

Page 4: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

4

introduction

3d content creation

acquisition renderingprocessing

registration

raw scans

point cloud reconstruction

triangle mesh

Page 5: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

5

introduction

3d content creation

acquisition renderingprocessing

registration

raw scans

point cloud reconstruction

triangle mesh

simplification

reduced point cloud

Page 6: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

6

introduction

3d content creation

acquisition renderingprocessing

registration

raw scans

point cloud

simplification

reduced point cloud

Page 7: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

7

surface model

moving least squares (mls) approximation

Gaussian used for locality

idea: locally approximate surface with polynomial

compute reference planecompute weighted least-squares fit polynomial

implicit surface definition using a projection operator

Page 8: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

8

surface model

moving least squares (mls) approximation

idea: locally approximate surface with polynomial

compute reference planecompute weighted least-squares fit polynomial

Gaussian used for locality

implicit surface definition using a projection operator

Page 9: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

9

local surface analysis

local neighborhood (k-nearest neighbors)

Page 10: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

10

local surface analysislocal neighborhood (e.g. k-nearest)

pp

pp

pp

pp

C

n

T

n

11

covariance matrix

eigenvalue problem

lll vvC

pcentroid

Page 11: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

11

local surface analysislocal neighborhood (e.g. k-nearest)

eigenvectors span covariance ellipsoid

surface variation

smallest eigenvector is normal

in

0)(p

measures deviation from tangent plane curvature

Page 12: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

12

local surface analysis

example

original mean curvature variation n=20 variation n=50

Page 13: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

13

surface simplification

incremental clusteringhierarchical clusteringiterative simplificationparticle simulation

Page 14: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

14

incremental clustering

Clustering by growing regions start with a random seed point successively add nearest points to cluster

until cluster reaches desired maximum size

the growth of clusters can also be limited be surface variation and in that way the curvature adaptive clustering is achieved.

Page 15: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

15

incremental clustering

Incremental growth leads to some fragmentation. Therefore stray samples need to be added to closest clusters at the end of the run.

Page 16: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

16

incremental clustering

each cluster is replaced by its centroid

Origina model34,384 points

Simplified model 1,000 pts

Page 17: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

17

incremental clustering

Results from my incremental clustering implementation.

35,000 pts 1,222 pts

Page 18: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

18

surface simplification

incremental clusteringhierarchical clusteringiterative simplificationparticle simulation

Page 19: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

19

hierarchical clustering

top-down approach using binary space partition

recursively split the point cloud if: size is larger than a user-specified threshold or surface variation is above maximum threshold

split plane defined by centroid and axis of greatest variation

replace clusters by centroid

Page 20: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

20

hierarchical clustering

2d example

covariance ellipsoid split plane

centroid

root

Page 21: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

21

hierarchical clustering

2d example

Page 22: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

22

hierarchical clustering

2d example

Page 23: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

23

hierarchical clustering

2d example

Page 24: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

24

hierarchical clustering

4,280 Clusters436 Clusters43 Clusters

Page 25: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

25

surface simplification

incremental clusteringhierarchical clusteringiterative simplificationparticle simulation

Page 26: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

26

iterative simplificationiteratively contracts point pairs

each contraction reduces the number of points by one

contractions are arranged in priority queue according to quadric error metric

quadric measures cost of contraction and determines optimal position for contracted sample

equivalent to QSlim except for definition of approximating planes

Page 27: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

27

surface simplification

incremental clusteringhierarchical clusteringiterative simplificationparticle simulation

Page 28: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

28

particle simulation

Method proposed by Turk G. (for polygonal surfaces)

resample surface by distributing particles on the surface

particles move on surface according to inter-particle repelling forces

particle relaxation terminates when equilibrium is reached

can also be used for up-sampling!

Page 29: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

29

measuring error

measure distance between two point-sampled surfaces S and S’ using a sampling approach

compute set Q of points on S

maximum error:

two-sided Hausdorff distance

mean error:

area-weighted integral of point-to-surface distances

size of Q determines accuracy of error measure

),(max),(max SdSS Q qq

Q

SdQ

SSq

q ),(1

),(avg

Page 30: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

30

measuring error

d(q,S’) measures the distance of point q to surface S’ using the mls projection operator S

'S

),( Sd q

q

'q

Page 31: 1 efficient simplification of point-sampled geometry From the paper “Efficient Simplification of Point-Sampled Surfaces” by Mark Pauly, Markus Gross, Leif

31

conclusions

point cloud simplification can be useful to reduce the complexity of geometric models

early in the 3d content creation pipeline create surface hierarchies

References Mark Pauly et al “Efficient Simplification of Point Sampled

Surfaces” Mark Alexa et al “Point Set Surfaces” Levin D. “Mesh-independent surface interpolation”