IEEE Computational Intelligence Magazine - November 2023 - 73
TABLE I Results obtained for the testing point clouds from the full models on ModelNet40 in the full-category setting [44]. Bold
indicates the best result and underline indicates the second-best result.
METHOD
ICP [20]
FGR [40]
FPFH [41]
PointNetLK [11]
FMR [39]
PCRNet [30]
PRNet [18]
DCP-v2 [15]
IDAM-GNN [29]
RPMNet [28]
OMNet [42]
CFNet(ours)
(a) UNSEEN MODELS
(b) UNSEEN CATEGORIES
(c) GAUSSIAN NOISE
RMSE(R) MAE(R) RMSE(t) MAE(t) RMSE(R) MAE(R) RMSE(t) MAE(t) RMSE(R) MAE(R) RMSE(t) MAE(t)
12.353 5.617 0.1668 0.0695 13.298 6.241 0.1834 0.0789 13.828 7.117 0.2902 0.0878
7.381
3.381
9.259
6.852
4.220
3.205
2.926
2.838
1.634
1.133
1.106
5.381 0.0117 0.0062 7.344
2.961 0.0152 0.0132 3.213
5.363 0.0131 0.0068 7.387
2.821 0.0158 0.0134 5.067
0.0105
0.911 0.0097 0.0056
1.006
0.0025 2.484
2.094
1.803 0.5667 0.475 7.633
2.885 0.0412 0.0469 4.524
2.315 0.0218 0.0171 4.328
3.207 0.0236 0.0127 3.083
0.715 0.0251 0.0047 3.671
0.539
0.0169
1.193 0.0206 0.0146
0.0036 0.0024 1.272
by Gaussian noise, which will be discussed
in detail in Section IV-E.
B. Full Categories for Training &
Testing
In the first experiment, all categories of
point clouds in ModelNet40 are used.
1.024
0.0045 2.217
1.583
5.356 0.0105 0.0058
4.196 0.0215 0.0188
1.253 0.5764 0.4763 11.045 2.628 0.5761 0.4764 12.177 2.865 0.6673 0.5374
1.596 0.5851 0.4886 7.579
3.015 0.0458 0.0484 4.895
1.415 0.0168 0.0124 4.995
1.273 0.0214 0.0123 4.258
0.742 0.0128 0.0034 2.462
0.287
2.295 0.5797 0.4844
3.257 0.0468 0.0505
2.056 0.0170 0.0123
1.277 0.0105 0.0091
1.026 0.0236 0.0061
0.570
1.162
0.0056
0.0032 0.0027 1.348
9843/2468 samples are used for training
and testing, respectively. Table I(a) evaluates
the performance of our method
and its counterparts in this experiment
(ICP and PointNetLK almost fail).
Note that the unsupervised-based registration
methods [36], [37] outperform
1.234
0.0033
0.0028
most supervised-based registration
methods in terms of most metrics, and
our method can theoretically further
stimulate their potential. From the
experimental results, CFNet is superior
to the other methods under most performance
indicators, showing strong
0.0161 0.0048
0.0024
FIGURE 6. Registration results (cyan: source point cloud, magenta: target point cloud, and gray: transformed point cloud). The first two are
correspondence- or feature descriptor-based methods, and the last three are deep learning-based methods where the target point cloud is the same
but the initial poses of the source point clouds are different to distinguish the registration effects of the different methods.
NOVEMBER 2023 | IEEE COMPUTATIONAL INTELLIGENCE MAGAZINE 73
IEEE Computational Intelligence Magazine - November 2023
Table of Contents for the Digital Edition of IEEE Computational Intelligence Magazine - November 2023
Contents
IEEE Computational Intelligence Magazine - November 2023 - Cover1
IEEE Computational Intelligence Magazine - November 2023 - Cover2
IEEE Computational Intelligence Magazine - November 2023 - Contents
IEEE Computational Intelligence Magazine - November 2023 - 2
IEEE Computational Intelligence Magazine - November 2023 - 3
IEEE Computational Intelligence Magazine - November 2023 - 4
IEEE Computational Intelligence Magazine - November 2023 - 5
IEEE Computational Intelligence Magazine - November 2023 - 6
IEEE Computational Intelligence Magazine - November 2023 - 7
IEEE Computational Intelligence Magazine - November 2023 - 8
IEEE Computational Intelligence Magazine - November 2023 - 9
IEEE Computational Intelligence Magazine - November 2023 - 10
IEEE Computational Intelligence Magazine - November 2023 - 11
IEEE Computational Intelligence Magazine - November 2023 - 12
IEEE Computational Intelligence Magazine - November 2023 - 13
IEEE Computational Intelligence Magazine - November 2023 - 14
IEEE Computational Intelligence Magazine - November 2023 - 15
IEEE Computational Intelligence Magazine - November 2023 - 16
IEEE Computational Intelligence Magazine - November 2023 - 17
IEEE Computational Intelligence Magazine - November 2023 - 18
IEEE Computational Intelligence Magazine - November 2023 - 19
IEEE Computational Intelligence Magazine - November 2023 - 20
IEEE Computational Intelligence Magazine - November 2023 - 21
IEEE Computational Intelligence Magazine - November 2023 - 22
IEEE Computational Intelligence Magazine - November 2023 - 23
IEEE Computational Intelligence Magazine - November 2023 - 24
IEEE Computational Intelligence Magazine - November 2023 - 25
IEEE Computational Intelligence Magazine - November 2023 - 26
IEEE Computational Intelligence Magazine - November 2023 - 27
IEEE Computational Intelligence Magazine - November 2023 - 28
IEEE Computational Intelligence Magazine - November 2023 - 29
IEEE Computational Intelligence Magazine - November 2023 - 30
IEEE Computational Intelligence Magazine - November 2023 - 31
IEEE Computational Intelligence Magazine - November 2023 - 32
IEEE Computational Intelligence Magazine - November 2023 - 33
IEEE Computational Intelligence Magazine - November 2023 - 34
IEEE Computational Intelligence Magazine - November 2023 - 35
IEEE Computational Intelligence Magazine - November 2023 - 36
IEEE Computational Intelligence Magazine - November 2023 - 37
IEEE Computational Intelligence Magazine - November 2023 - 38
IEEE Computational Intelligence Magazine - November 2023 - 39
IEEE Computational Intelligence Magazine - November 2023 - 40
IEEE Computational Intelligence Magazine - November 2023 - 41
IEEE Computational Intelligence Magazine - November 2023 - 42
IEEE Computational Intelligence Magazine - November 2023 - 43
IEEE Computational Intelligence Magazine - November 2023 - 44
IEEE Computational Intelligence Magazine - November 2023 - 45
IEEE Computational Intelligence Magazine - November 2023 - 46
IEEE Computational Intelligence Magazine - November 2023 - 47
IEEE Computational Intelligence Magazine - November 2023 - 48
IEEE Computational Intelligence Magazine - November 2023 - 49
IEEE Computational Intelligence Magazine - November 2023 - 50
IEEE Computational Intelligence Magazine - November 2023 - 51
IEEE Computational Intelligence Magazine - November 2023 - 52
IEEE Computational Intelligence Magazine - November 2023 - 53
IEEE Computational Intelligence Magazine - November 2023 - 54
IEEE Computational Intelligence Magazine - November 2023 - 55
IEEE Computational Intelligence Magazine - November 2023 - 56
IEEE Computational Intelligence Magazine - November 2023 - 57
IEEE Computational Intelligence Magazine - November 2023 - 58
IEEE Computational Intelligence Magazine - November 2023 - 59
IEEE Computational Intelligence Magazine - November 2023 - 60
IEEE Computational Intelligence Magazine - November 2023 - 61
IEEE Computational Intelligence Magazine - November 2023 - 62
IEEE Computational Intelligence Magazine - November 2023 - 63
IEEE Computational Intelligence Magazine - November 2023 - 64
IEEE Computational Intelligence Magazine - November 2023 - 65
IEEE Computational Intelligence Magazine - November 2023 - 66
IEEE Computational Intelligence Magazine - November 2023 - 67
IEEE Computational Intelligence Magazine - November 2023 - 68
IEEE Computational Intelligence Magazine - November 2023 - 69
IEEE Computational Intelligence Magazine - November 2023 - 70
IEEE Computational Intelligence Magazine - November 2023 - 71
IEEE Computational Intelligence Magazine - November 2023 - 72
IEEE Computational Intelligence Magazine - November 2023 - 73
IEEE Computational Intelligence Magazine - November 2023 - 74
IEEE Computational Intelligence Magazine - November 2023 - 75
IEEE Computational Intelligence Magazine - November 2023 - 76
IEEE Computational Intelligence Magazine - November 2023 - 77
IEEE Computational Intelligence Magazine - November 2023 - 78
IEEE Computational Intelligence Magazine - November 2023 - 79
IEEE Computational Intelligence Magazine - November 2023 - 80
IEEE Computational Intelligence Magazine - November 2023 - Cover3
IEEE Computational Intelligence Magazine - November 2023 - Cover4
https://www.nxtbook.com/nxtbooks/ieee/computationalintelligence_202311
https://www.nxtbook.com/nxtbooks/ieee/computationalintelligence_202308
https://www.nxtbook.com/nxtbooks/ieee/computationalintelligence_202305
https://www.nxtbook.com/nxtbooks/ieee/computationalintelligence_202302
https://www.nxtbook.com/nxtbooks/ieee/computationalintelligence_202211
https://www.nxtbook.com/nxtbooks/ieee/computationalintelligence_202208
https://www.nxtbook.com/nxtbooks/ieee/computationalintelligence_202205
https://www.nxtbook.com/nxtbooks/ieee/computationalintelligence_202202
https://www.nxtbook.com/nxtbooks/ieee/computationalintelligence_202111
https://www.nxtbook.com/nxtbooks/ieee/computationalintelligence_202108
https://www.nxtbook.com/nxtbooks/ieee/computationalintelligence_202105
https://www.nxtbook.com/nxtbooks/ieee/computationalintelligence_202102
https://www.nxtbook.com/nxtbooks/ieee/computationalintelligence_202011
https://www.nxtbook.com/nxtbooks/ieee/computationalintelligence_202008
https://www.nxtbook.com/nxtbooks/ieee/computationalintelligence_202005
https://www.nxtbook.com/nxtbooks/ieee/computationalintelligence_202002
https://www.nxtbook.com/nxtbooks/ieee/computationalintelligence_201911
https://www.nxtbook.com/nxtbooks/ieee/computationalintelligence_201908
https://www.nxtbook.com/nxtbooks/ieee/computationalintelligence_201905
https://www.nxtbook.com/nxtbooks/ieee/computationalintelligence_201902
https://www.nxtbook.com/nxtbooks/ieee/computationalintelligence_201811
https://www.nxtbook.com/nxtbooks/ieee/computationalintelligence_201808
https://www.nxtbook.com/nxtbooks/ieee/computationalintelligence_201805
https://www.nxtbook.com/nxtbooks/ieee/computationalintelligence_201802
https://www.nxtbook.com/nxtbooks/ieee/computational_intelligence_winter17
https://www.nxtbook.com/nxtbooks/ieee/computational_intelligence_fall17
https://www.nxtbook.com/nxtbooks/ieee/computational_intelligence_summer17
https://www.nxtbook.com/nxtbooks/ieee/computational_intelligence_spring17
https://www.nxtbook.com/nxtbooks/ieee/computational_intelligence_winter16
https://www.nxtbook.com/nxtbooks/ieee/computational_intelligence_fall16
https://www.nxtbook.com/nxtbooks/ieee/computational_intelligence_summer16
https://www.nxtbook.com/nxtbooks/ieee/computational_intelligence_spring16
https://www.nxtbook.com/nxtbooks/ieee/computational_intelligence_winter15
https://www.nxtbook.com/nxtbooks/ieee/computational_intelligence_fall15
https://www.nxtbook.com/nxtbooks/ieee/computational_intelligence_summer15
https://www.nxtbook.com/nxtbooks/ieee/computational_intelligence_spring15
https://www.nxtbook.com/nxtbooks/ieee/computational_intelligence_winter14
https://www.nxtbook.com/nxtbooks/ieee/computational_intelligence_fall14
https://www.nxtbook.com/nxtbooks/ieee/computational_intelligence_summer14
https://www.nxtbook.com/nxtbooks/ieee/computational_intelligence_spring14
https://www.nxtbook.com/nxtbooks/ieee/computational_intelligence_winter13
https://www.nxtbook.com/nxtbooks/ieee/computational_intelligence_fall13
https://www.nxtbook.com/nxtbooks/ieee/computational_intelligence_summer13
https://www.nxtbook.com/nxtbooks/ieee/computational_intelligence_spring13
https://www.nxtbook.com/nxtbooks/ieee/computational_intelligence_winter12
https://www.nxtbook.com/nxtbooks/ieee/computational_intelligence_fall12
https://www.nxtbookmedia.com