1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
|
// ----------------------------------------------------------------------------
// - Open3D: www.open3d.org -
// ----------------------------------------------------------------------------
// Copyright (c) 2018-2024 www.open3d.org
// SPDX-License-Identifier: MIT
// ----------------------------------------------------------------------------
#include "open3d/pipelines/registration/Registration.h"
#include <benchmark/benchmark.h>
#include <Eigen/Eigen>
#include "open3d/data/Dataset.h"
#include "open3d/geometry/KDTreeFlann.h"
#include "open3d/geometry/PointCloud.h"
#include "open3d/io/PointCloudIO.h"
#include "open3d/pipelines/registration/TransformationEstimation.h"
#include "open3d/utility/Logging.h"
namespace open3d {
namespace pipelines {
namespace registration {
// Testing parameters:
// ICP ConvergenceCriteria.
static const double relative_fitness = 1e-6;
static const double relative_rmse = 1e-6;
static const int max_iterations = 10;
static const double voxel_downsampling_factor = 0.02;
// NNS parameter.
static const double max_correspondence_distance = 0.05;
static std::tuple<geometry::PointCloud, geometry::PointCloud> LoadPointCloud(
const std::string& source_filename,
const std::string& target_filename,
const double voxel_downsample_factor) {
geometry::PointCloud source;
geometry::PointCloud target;
io::ReadPointCloud(source_filename, source, {"auto", false, false, true});
io::ReadPointCloud(target_filename, target, {"auto", false, false, true});
// Eliminates the case of impractical values (including negative).
if (voxel_downsample_factor > 0.001) {
source = *source.VoxelDownSample(voxel_downsample_factor);
target = *target.VoxelDownSample(voxel_downsample_factor);
} else {
utility::LogWarning(
" VoxelDownsample: Impractical voxel size [< 0.001], skipping "
"downsampling.");
}
return std::make_tuple(source, target);
}
static void BenchmarkICPLegacy(benchmark::State& state,
const TransformationEstimationType& type) {
data::DemoICPPointClouds demo_icp_pointclouds;
geometry::PointCloud source, target;
std::tie(source, target) = LoadPointCloud(demo_icp_pointclouds.GetPaths(0),
demo_icp_pointclouds.GetPaths(1),
voxel_downsampling_factor);
std::shared_ptr<TransformationEstimation> estimation;
if (type == TransformationEstimationType::PointToPlane) {
estimation = std::make_shared<TransformationEstimationPointToPlane>();
} else if (type == TransformationEstimationType::PointToPoint) {
estimation = std::make_shared<TransformationEstimationPointToPoint>();
}
Eigen::Matrix4d init_trans;
init_trans << 0.862, 0.011, -0.507, 0.5, -0.139, 0.967, -0.215, 0.7, 0.487,
0.255, 0.835, -1.4, 0.0, 0.0, 0.0, 1.0;
RegistrationResult reg_result(init_trans);
// Warm up.
reg_result = RegistrationICP(
source, target, max_correspondence_distance, init_trans,
*estimation,
ICPConvergenceCriteria(relative_fitness, relative_rmse,
max_iterations));
for (auto _ : state) {
reg_result = RegistrationICP(
source, target, max_correspondence_distance, init_trans,
*estimation,
ICPConvergenceCriteria(relative_fitness, relative_rmse,
max_iterations));
}
utility::LogDebug(" Max iterations: {}, Max_correspondence_distance : {}",
max_iterations, max_correspondence_distance);
utility::LogDebug(" Fitness: {} Inlier RMSE: {}", reg_result.fitness_,
reg_result.inlier_rmse_);
}
BENCHMARK_CAPTURE(BenchmarkICPLegacy,
PointToPlane / CPU,
TransformationEstimationType::PointToPlane)
->Unit(benchmark::kMillisecond);
BENCHMARK_CAPTURE(BenchmarkICPLegacy,
PointToPoint / CPU,
TransformationEstimationType::PointToPoint)
->Unit(benchmark::kMillisecond);
} // namespace registration
} // namespace pipelines
} // namespace open3d
|
