Table 9. Runtimes for registering a point-cloud target shape. (Experiment 4).
| Exp. | Alg. | Average Runtimes (sec.) by Test Case | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | ||
| 4A | ICP | 0.009 | 0.009 | 0.010 | 0.009 | 0.010 | 0.009 | 0.009 | 0.009 | 0.009 |
| IMLP-CP | 0.015 | 0.016 | 0.019 | 0.016 | 0.020 | 0.019 | 0.015 | 0.017 | 0.015 | |
| IMLP-MD | 0.068 | 0.078 | 0.093 | 0.079 | 0.097 | 0.093 | 0.067 | 0.079 | 0.069 | |
| GICP | - | - | - | - | - | - | - | - | - | |
| CPD (2 cores) | 3.465 | 4.346 | 4.336 | 3.864 | 4.340 | 4.374 | 4.238 | 4.650 | 4.484 | |
| IMLP | 0.068 | 0.082 | 0.102 | 0.078 | 0.103 | 0.099 | 0.067 | 0.084 | 0.073 | |
| 4B | ICP | 0.013 | 0.013 | 0.013 | 0.013 | 0.013 | 0.013 | 0.012 | 0.012 | 0.013 |
| IMLP-CP | 0.023 | 0.025 | 0.028 | 0.025 | 0.028 | 0.028 | 0.024 | 0.025 | 0.024 | |
| IMLP-MD | 0.100 | 0.109 | 0.126 | 0.112 | 0.127 | 0.129 | 0.100 | 0.109 | 0.099 | |
| GICP | - | - | - | - | - | - | - | - | - | |
| CPD (2 cores) | 3.584 | 4.408 | 4.490 | 4.279 | 4.327 | 4.545 | 4.378 | 4.731 | 4.874 | |
| IMLP | 0.101 | 0.111 | 0.134 | 0.115 | 0.136 | 0.133 | 0.103 | 0.118 | 0.106 | |
Average runtimes of successful registrations from Experiment 4 are reported, where 300 randomized trials were conducted for each test case. Each test case represents a different generative noise model (Table 4) applied to points of the source shape. Results are also reported for initial shape misalignments of [15, 30] mm / degrees (Experiment 4A) and [30, 60] mm / degrees (Experiment 4B). The algorithms compared include single-threaded implementations of ICP [1], GICP [11], IMLP, and the two IMLP variants IMLP-CP and IMLP-MD. A multi-threaded implementation of CPD [20] is also reported, which made full utilization of 2 cores.