def get_largest_dbscan_clusters(pointcloud,
min_return_fragment=0.7,
epsilon=0.1,
minpoints=250,
rgb_weight=0):
'''
Finds the largest clusters containing together at least min_return_fragment
of the complete point cloud. In case less points belong to clusters, all
clustered points are returned.
Parameters
----------
pointcloud : pcl.PointCloud
Input pointcloud.
min_return_fragment : float
Minimum desired fragment of pointcloud to be returned
epsilon : float
Neighborhood radius for DBSCAN.
minpoints : integer
Minimum neighborhood density for DBSCAN.
rgb_weight : float, optional
If non-zero, cluster on color information as well as location;
specifies the relative weight of the RGB components to spatial
coordinates in distance computations.
(RGB values have wildly different scales than spatial coordinates.)
Returns
-------
cluster : pcl.PointCloud
Registered pointcloud of the largest cluster found by dbscan.
'''
labels = dbscan_labels(pointcloud,
epsilon,
minpoints,
rgb_weight=rgb_weight).astype(np.int64)
selection, selected_count = _get_top_labels(labels, min_return_fragment)
# No clusters were found
if selected_count < min_return_fragment * len(labels):
return extract_mask(pointcloud, np.ones(len(pointcloud), dtype=bool))
else:
mask = [label in selection for label in labels]
return extract_mask(pointcloud, mask)
def get_largest_dbscan_clusters(pointcloud, min_return_fragment=0.7, epsilon=0.1, minpoints=250, rgb_weight=0):
"""
Finds the largest clusters containing together at least min_return_fragment
of the complete point cloud. In case less points belong to clusters, all
clustered points are returned.
Parameters
----------
pointcloud : pcl.PointCloud
Input pointcloud.
min_return_fragment : float
Minimum desired fragment of pointcloud to be returned
epsilon : float
Neighborhood radius for DBSCAN.
minpoints : integer
Minimum neighborhood density for DBSCAN.
rgb_weight : float, optional
If non-zero, cluster on color information as well as location;
specifies the relative weight of the RGB components to spatial
coordinates in distance computations.
(RGB values have wildly different scales than spatial coordinates.)
Returns
-------
cluster : pcl.PointCloud
Registered pointcloud of the largest cluster found by dbscan.
"""
labels = dbscan_labels(pointcloud, epsilon, minpoints, rgb_weight=rgb_weight).astype(np.int64)
selection, selected_count = _get_top_labels(labels, min_return_fragment)
# No clusters were found
if selected_count < min_return_fragment * len(labels):
return extract_mask(pointcloud, np.ones(len(pointcloud), dtype=bool))
else:
mask = [label in selection for label in labels]
return extract_mask(pointcloud, mask)
def get_stick_scale(pointcloud, eps=0.1, min_samples=20):
"""Takes a point cloud, as a numpy array, looks for red segments
of scale sticks and returns the scale estimation with most support.
Method:
pointcloud --dbscan--> clusters --lengthEstimation-->
lengths --ransac--> best length
Arguments:
pointcloud Point cloud containing only measuring stick segments
(only the red, or only the white parts)
eps DBSCAN parameter: Maximum distance between two samples
for them to be considered as in the same neighborhood.
min_samples DBSCAN parameter: The number of samples in a neighborhood
for a point to be considered as a core point.
Returns:
scale Estimate of the size of one actual meter in expressed
in units of the pointcloud's coordinates.
confidence A number expressing the reliability of the estimated
scale. Confidence is in [0, 1]. With a confidence greater
than .5, the estimate can be considered useable for
further calculations.
"""
# quickly return for trivial case
if pointcloud.size == 0:
return 1, 0
# find the red segments to measure
pc_reds = extract_mask(pointcloud, get_red_mask(pointcloud))
if len(pc_reds) == 0:
# unit scale, zero confidence (ie. any other estimation is better)
return 1.0, 0.0
cluster_generator = segment_dbscan(
pc_reds, eps, min_samples, algorithm='kd_tree')
sizes = [{'len': len(cluster),
'meter': measure_length(cluster) * SEGMENTS_PER_METER}
for cluster in cluster_generator]
if len(sizes) == 0:
return 1.0, 0.0
scale, votes, n_clusters = ransac(sizes)
confidence = get_confidence_level(votes, n_clusters)
return scale, confidence
def segment_dbscan(pointcloud, epsilon, minpoints, **kwargs):
"""Run the DBSCAN clustering+outlier detection algorithm on pointcloud.
Parameters
----------
pointcloud : pcl.PointCloud
Input pointcloud.
epsilon : float
Neighborhood radius for DBSCAN.
minpoints : integer
Minimum neighborhood density for DBSCAN.
**kwargs : keyword arguments, optional
arguments passed to _dbscan_labels
Returns
-------
clusters : iterable over registered PointCloud
"""
labels = dbscan_labels(pointcloud, epsilon, minpoints, **kwargs)
return (extract_mask(pointcloud, labels == label) for label in np.unique(labels[labels != -1]))
def segment_dbscan(pointcloud, epsilon, minpoints, **kwargs):
"""Run the DBSCAN clustering+outlier detection algorithm on pointcloud.
Parameters
----------
pointcloud : pcl.PointCloud
Input pointcloud.
epsilon : float
Neighborhood radius for DBSCAN.
minpoints : integer
Minimum neighborhood density for DBSCAN.
**kwargs : keyword arguments, optional
arguments passed to _dbscan_labels
Returns
-------
clusters : iterable over registered PointCloud
"""
labels = dbscan_labels(pointcloud, epsilon, minpoints, **kwargs)
return (extract_mask(pointcloud, labels == label)
for label in np.unique(labels[labels != -1]))
#!/usr/bin/env python
"""Segment points by colour from a pointcloud file and saves all reddish points
target pointcloud file. Autodectects ply, pcd and las files.
Usage: redstickdetection.py [-h] <infile> <outfile>
"""
from docopt import docopt
from patty.segmentation.segRedStick import get_red_mask
from patty.utils import extract_mask, load, save
if __name__ == '__main__':
args = docopt(__doc__)
pc = load(args['<infile>'])
red_pc = extract_mask(pc, get_red_mask(pc))
save(red_pc, args['<outfile>'])