Make fragments
The first step of the scene reconstruction system is to create fragments from short RGBD sequences.
Input arguments
The script runs with python run_system.py [config] --make
. In [config]
,
["path_dataset"]
should have subfolders image
and depth
to store the
color images and depth images respectively. We assume the color images and the
depth images are synchronized and registered. In [config]
, the optional
argument ["path_intrinsic"]
specifies the path to a json file that stores
the camera intrinsic matrix (See
/tutorial/pipelines/rgbd_odometry.ipynb#read-camera-intrinsic for
details). If it is not given, the PrimeSense factory setting is used instead.
Register RGBD image pairs
46
47 o3d.pipelines.odometry.RGBDOdometryJacobianFromHybridTerm(),
48 option)
49 return [success, trans, info]
50 return [False, np.identity(4), np.identity(6)]
51 else:
52 odo_init = np.identity(4)
53 [success, trans, info] = o3d.pipelines.odometry.compute_rgbd_odometry(
54 source_rgbd_image, target_rgbd_image, intrinsic, odo_init,
55 o3d.pipelines.odometry.RGBDOdometryJacobianFromHybridTerm(), option)
56 return [success, trans, info]
57
58
59def make_posegraph_for_fragment(path_dataset, sid, eid, color_files,
60 depth_files, fragment_id, n_fragments,
61 intrinsic, with_opencv, config):
62 o3d.utility.set_verbosity_level(o3d.utility.VerbosityLevel.Error)
63 pose_graph = o3d.pipelines.registration.PoseGraph()
64 trans_odometry = np.identity(4)
65 pose_graph.nodes.append(
66 o3d.pipelines.registration.PoseGraphNode(trans_odometry))
67 for s in range(sid, eid):
68 for t in range(s + 1, eid):
69 # odometry
70 if t == s + 1:
71 print(
72 "Fragment %03d / %03d :: RGBD matching between frame : %d and %d"
73 % (fragment_id, n_fragments - 1, s, t))
74 [success, trans,
75 info] = register_one_rgbd_pair(s, t, color_files, depth_files,
76 intrinsic, with_opencv, config)
The function reads a pair of RGBD images and registers the source_rgbd_image
to the target_rgbd_image
. The Open3D function compute_rgbd_odometry
is
called to align the RGBD images. For adjacent RGBD images, an identity matrix is
used as the initialization. For non-adjacent RGBD images, wide baseline matching
is used as the initialization. In particular, the function pose_estimation
computes OpenCV ORB feature to match sparse features over wide baseline images,
then performs 5-point RANSAC to estimate a rough alignment, which is used as
the initialization of compute_rgbd_odometry
.
Multiway registration
76
77 trans_odometry = np.dot(trans, trans_odometry)
78 trans_odometry_inv = np.linalg.inv(trans_odometry)
79 pose_graph.nodes.append(
80 o3d.pipelines.registration.PoseGraphNode(
81 trans_odometry_inv))
82 pose_graph.edges.append(
83 o3d.pipelines.registration.PoseGraphEdge(s - sid,
84 t - sid,
85 trans,
86 info,
87 uncertain=False))
88
89 # keyframe loop closure
90 if s % config['n_keyframes_per_n_frame'] == 0 \
91 and t % config['n_keyframes_per_n_frame'] == 0:
92 print(
93 "Fragment %03d / %03d :: RGBD matching between frame : %d and %d"
94 % (fragment_id, n_fragments - 1, s, t))
95 [success, trans,
96 info] = register_one_rgbd_pair(s, t, color_files, depth_files,
97 intrinsic, with_opencv, config)
98 if success:
99 pose_graph.edges.append(
100 o3d.pipelines.registration.PoseGraphEdge(
101 s - sid, t - sid, trans, info, uncertain=True))
102 o3d.io.write_pose_graph(
103 join(path_dataset, config["template_fragment_posegraph"] % fragment_id),
104 pose_graph)
105
106
107def integrate_rgb_frames_for_fragment(color_files, depth_files, fragment_id,
108 n_fragments, pose_graph_name, intrinsic,
109 config):
110 pose_graph = o3d.io.read_pose_graph(pose_graph_name)
111 volume = o3d.pipelines.integration.ScalableTSDFVolume(
112 voxel_length=config["tsdf_cubic_size"] / 512.0,
113 sdf_trunc=0.04,
114 color_type=o3d.pipelines.integration.TSDFVolumeColorType.RGB8)
115 for i in range(len(pose_graph.nodes)):
116 i_abs = fragment_id * config['n_frames_per_fragment'] + i
117 print(
118 "Fragment %03d / %03d :: integrate rgbd frame %d (%d of %d)." %
119 (fragment_id, n_fragments - 1, i_abs, i + 1, len(pose_graph.nodes)))
120 rgbd = read_rgbd_image(color_files[i_abs], depth_files[i_abs], False,
121 config)
122 pose = pose_graph.nodes[i].pose
123 volume.integrate(rgbd, intrinsic, np.linalg.inv(pose))
This script uses the technique demonstrated in
/tutorial/pipelines/multiway_registration.ipynb. The function
make_posegraph_for_fragment
builds a pose graph for multiway registration of
all RGBD images in this sequence. Each graph node represents an RGBD image and
its pose which transforms the geometry to the global fragment space.
For efficiency, only key frames are used.
Once a pose graph is created, multiway registration is performed by calling the
function optimize_posegraph_for_fragment
.
51 pose_graph = o3d.io.read_pose_graph(pose_graph_name)
52 preference_loop_closure = \
53 config["preference_loop_closure_registration"])
54
55
56def optimize_posegraph_for_refined_scene(path_dataset, config):
57 pose_graph_name = join(path_dataset, config["template_refined_posegraph"])
58 pose_graph_optimized_name = join(
59 path_dataset, config["template_refined_posegraph_optimized"])
60 run_posegraph_optimization(pose_graph_name, pose_graph_optimized_name,
61 max_correspondence_distance = config["voxel_size"] * 1.4,
62 preference_loop_closure = \
63 config["preference_loop_closure_registration"])
This function calls global_optimization
to estimate poses of the RGBD images.
Make a fragment
124
125 mesh.compute_vertex_normals()
126 return mesh
127
128
129def make_pointcloud_for_fragment(path_dataset, color_files, depth_files,
130 fragment_id, n_fragments, intrinsic, config):
131 mesh = integrate_rgb_frames_for_fragment(
132 color_files, depth_files, fragment_id, n_fragments,
133 join(path_dataset,
134 config["template_fragment_posegraph_optimized"] % fragment_id),
135 intrinsic, config)
136 pcd = o3d.geometry.PointCloud()
137 pcd.points = mesh.vertices
138 pcd.colors = mesh.vertex_colors
139 pcd_name = join(path_dataset,
140 config["template_fragment_pointcloud"] % fragment_id)
141 o3d.io.write_point_cloud(pcd_name, pcd, False, True)
142
143
144def process_single_fragment(fragment_id, color_files, depth_files, n_files,
145 n_fragments, config):
146 if config["path_intrinsic"]:
Once the poses are estimates, /tutorial/pipelines/rgbd_integration.ipynb is used to reconstruct a colored fragment from each RGBD sequence.
Batch processing
181
182 else:
183 for fragment_id in range(n_fragments):
184 process_single_fragment(fragment_id, color_files, depth_files,
185 n_files, n_fragments, config)
The main function calls each individual function explained above.
Results
Fragment 000 / 013 :: RGBD matching between frame : 0 and 1
Fragment 000 / 013 :: RGBD matching between frame : 0 and 5
Fragment 000 / 013 :: RGBD matching between frame : 0 and 10
Fragment 000 / 013 :: RGBD matching between frame : 0 and 15
Fragment 000 / 013 :: RGBD matching between frame : 0 and 20
:
Fragment 000 / 013 :: RGBD matching between frame : 95 and 96
Fragment 000 / 013 :: RGBD matching between frame : 96 and 97
Fragment 000 / 013 :: RGBD matching between frame : 97 and 98
Fragment 000 / 013 :: RGBD matching between frame : 98 and 99
The following is a log from optimize_posegraph_for_fragment
.
[GlobalOptimizationLM] Optimizing PoseGraph having 100 nodes and 195 edges.
Line process weight : 389.309502
[Initial ] residual : 3.223357e+05, lambda : 1.771814e+02
[Iteration 00] residual : 1.721845e+04, valid edges : 157, time : 0.022 sec.
[Iteration 01] residual : 1.350251e+04, valid edges : 168, time : 0.017 sec.
:
[Iteration 32] residual : 9.779118e+03, valid edges : 179, time : 0.013 sec.
Current_residual - new_residual < 1.000000e-06 * current_residual
[GlobalOptimizationLM] total time : 0.519 sec.
[GlobalOptimizationLM] Optimizing PoseGraph having 100 nodes and 179 edges.
Line process weight : 398.292104
[Initial ] residual : 5.120047e+03, lambda : 2.565362e+02
[Iteration 00] residual : 5.064539e+03, valid edges : 179, time : 0.014 sec.
[Iteration 01] residual : 5.037665e+03, valid edges : 178, time : 0.015 sec.
:
[Iteration 11] residual : 5.017307e+03, valid edges : 177, time : 0.013 sec.
Current_residual - new_residual < 1.000000e-06 * current_residual
[GlobalOptimizationLM] total time : 0.197 sec.
CompensateReferencePoseGraphNode : reference : 0
The following is a log from integrate_rgb_frames_for_fragment
.
Fragment 000 / 013 :: integrate rgbd frame 0 (1 of 100).
Fragment 000 / 013 :: integrate rgbd frame 1 (2 of 100).
Fragment 000 / 013 :: integrate rgbd frame 2 (3 of 100).
:
Fragment 000 / 013 :: integrate rgbd frame 97 (98 of 100).
Fragment 000 / 013 :: integrate rgbd frame 98 (99 of 100).
Fragment 000 / 013 :: integrate rgbd frame 99 (100 of 100).
The following images show some of the fragments made by this script.



