Skip to main content

Table 2 Experimental results summary

From: Multi-view alignment with database of features for an improved usage of high-end 3D scanners

    

Pairwise alignment

Multi-view alignment

Dataset

Number of RI pairs

Avg. Number of points/RI

Param. config.

Number of RI pairs

Avg. exec. time/RI [ s]

Number of clusters

Avg. exec. time/RI [s]

Initial align. error [μm]

Final align. error [μm]

    

aligned

 

created

   

Angels

7

1M

std

7

3.6

1

4.9

62

27

Mould 1

7

410k

std

7

2.0

1

2.6

39

14

Mould 2

9

400k

std

9

1.7

1

2.2

66

11

Mould 3

9

370k

std

9

1.5

1

1.7

103

11

Mould 4

9

360k

std

9

1.6

1

2.0

165

12

Platelet

11

80k

sml

11

1.5

1

1.8

28

9

Dog

13

130k

sml

13

2.5

1

5.0

32

16

Cherub

19

650k

sml

18

1.3

1

1.6

149

20

Dolphin

19

410k

std

19

2.0

1

4.1

136

33

Capital

22

760k

std

22

3.1

1

6.6

127

23

Rose

23

170k

sml

23

2.8

1

6.1

67

13

Hurricane

31

690k

std

30

2.9

1

9.0

188

35

Decoration

47

510k

std

39

2.7

2

10.3

–

–

Venus

60

835k

std

60

3.1

1

12.1

152

17

Doga

13

130k

sml

7

2.4

1

5.1

41

16

Capitala

22

760k

std

11

2.9

1

7.5

176

23

Hurricanea

31

690k

std

17

2.9

1

9.6

255

35

Carterb

92

550k

std

64

2.5

18

16.8

–

–

  1. aAcquisition path for these datasets does not abide by the overlap constraint;
  2. bParticularly featureless dataset.