• No results found

The ICP algorithm and its extensions Local, Rigid, Pairwise

N/A
N/A
Protected

Academic year: 2022

Share "The ICP algorithm and its extensions Local, Rigid, Pairwise"

Copied!
47
0
0

Laster.... (Se fulltekst nå)

Fulltekst

(1)

Local, Rigid, Pairwise

The ICP algorithm and its extensions

(2)
(3)

(4)
(5)
(6)

(7)

k

2

Rp

i

t q

i

E

(8)

1. Selecting source points (from one or both meshes) 2. Matching to points in the other mesh

3. Weighting the correspondences

4. Rejecting certain (outlier) point pairs

5. Assigning an error metric to the current transform 6. Minimizing the error metric w.r.t. transformation

(9)

(10)

1. Selecting source points (from one or both meshes) 2. Matching to points in the other mesh

3. Weighting the correspondences

4. Rejecting certain (outlier) point pairs

5. Assigning an error metric to the current transform 6. Minimizing the error metric w.r.t. transformation

(11)
(12)

R

    

( )

2

Rp

i

t q

i

n

i

E

 



 



 

 

z y x i

i i

i i

i

r r r r

n t n

p r

n q

p

E ( ) ( )

2

, where

(13)

 

 

 

 

 

 

 

 

 

 

2 2

1 1 1 2

2 2

1 1

2) (

) 1 (

, ,

,

n q p

n q p

t t t r r r

n n

p

n n

p

b x

b x

z y x z y x

A

A

  b

x

b x

1 T T

T T

A A

A A A

A

(14)
(15)

1. Selecting source points (from one or both meshes) 2. Matching to points in the other mesh

3. Weighting the correspondences

4. Rejecting certain (outlier) point pairs

5. Assigning an error metric to the current transform 6. Minimizing the error metric w.r.t. transformation

(16)

(17)

1. Selecting source points (from one or both meshes) 2. Matching to points in the other mesh

3. Weighting the correspondences

4. Rejecting certain (outlier) point pairs

5. Assigning an error metric to the current transform 6. Minimizing the error metric w.r.t. transformation

(18)

(19)

Uniform Sampling Stable Sampling

(20)

A

T

Ax = A

T

b

C = A

T

A

 

 

 

 

 

 

 

 

 

 

2 2

1 1 1 2

2 2

1 1

2) (

) 1 (

,

,

p q n

n q p

t t t r r r

n n

p

n n

p

b x

z y x z y x

A

(21)

C

3 small eigenvalues 3 rotation

3 small eigenvalues 2 translation

1 rotation

2 small eigenvalues 1 translation

1 rotation

1 small eigenvalue 1 rotation

1 small eigenvalue 1 translation

[Gelfand et al. ‘04]

(22)

Key: 3 DOFs stable 4 DOFs stable

5 DOFs stable 6 DOFs stable

(23)

• C

(24)

Random sampling Normal-space sampling

(25)
(26)

1. Selecting source points (from one or both meshes) 2. Matching to points in the other mesh

3. Weighting the correspondences

4. Rejecting certain (outlier) point pairs

5. Assigning an error metric to the current transform 6. Minimizing the error metric w.r.t. transformation

(27)

(28)

(29)
(30)

(31)
(32)
(33)
(34)
(35)

f(x)

(36)

f’(x)

(37)

(38)
(39)
(40)
(41)
(42)
(43)

(44)

2D 3D [Mitra et al. 2004]

(45)

(46)

Translation in x-z plane.

Rotation about y-axis.

Converges

Does not converge

(47)

distance-field

formulation

Plane-to-plane ICP

Referanser

RELATERTE DOKUMENTER

– extensions to the decimation method which support global error control have been proposed (Global error is defined here in opposition to local error, i.e. whether the

Any implementation re- quires methods for selecting points from a given finite data set, choosing an interpolant for each Voronoi cell, and choosing an error metric to determine

Adjusting an error metric derived from the hierarchy (e.g., in terms of the RMS deviation of scalar values contained in a node from their average value) permits one to select from

The error metric (see Sec. 4.1.1 ) we adopt assures an overestimation of the error introduced by the approximation but requires to be able to move eas- ily from sons to

From Section 3 to Sec- tion 6, we describe our parallel multiresolution volume ren- dering algorithm, including the construction of the wavelet tree with hierarchical error

Although both using the same greedy framework, our PSP algorithm is slower than IPS which has adopted efficient quadric error metric (QEM) [GH97], as it has to compose and

We evaluate several shadow map algorithms based on warping and partitioning using the maximum perspective aliasing error over the entire view frustum.. With respect to our error

Even though the actual illumination condi- tions are disregarded in such an interpolation error metric, it generates a record density that gracefully follows the indi- rect