Table 1 Performance on 3DMatch and 3DLoMatch datasets

From: Enhancing point cloud registration with transformer: cultural heritage protection of the Terracotta Warriors

Method

3DMatch

3DLoMatch

Param.(M)

Time(s)

RR(%)

RRE(\( ^{\circ }\))

RTE(m)

RR(%)

RRE(\( ^{\circ }\))

RTE(m)

FCGF[36]

85.1

1.949

0.066

40.1

3.147

0.100

8.76

0.16

D3Feat[28]

81.6

2.161

0.067

37.2

3.361

0.103

24.3

0.40

OMNet[37]

90.5

4.166

0.105

8.40

7.299

0.151

–

–

Predator[29]

89.0

2.029

0.064

59.8

3.048

0.093

7.43

0.54

REGTR[35]

92.0

1.567

0.049

64.8

2.827

0.077

–

–

GeoTrans[38]

92.0

1.808

0.063

74.0

2.934

0.089

9.83

0.23

UDPReg[39]

91.4

1.642

0.064

64.3

2.951

0.086

–

–

RoReg[40]

93.2

1.840

0.063

71.2

3.090

0.093

–

–

MAC[41]

93.7

1.890

0.062

59.8

3.500

0.098

–

–

RIGA[42]

89.3

1.798

0.056

65.1

3.016

0.089

–

–

Our

94.3

1.497

0.041

76.1

2.765

0.068

7.5

0.16

  1. The best results are in bold