distance from points to skeleton surface or centerline

[huangziwei]

Distance from multiple points to one skeleton

To measure the distance between some points to a skeleton, you can use dx.distance(skel, points), or directly from a Skeleton object:

import skeliner as sk

skel = sk.io.load_npz("../temp_io/720575940550605504.npz")
skel.distance(skel.nodes, mode="centerline")

will returns 0 for all nodes, as the nodes are the centerline.

if we change it a bit:

skel.distance(skel.nodes[100]+ 163, mode="surface"), skel.distance(skel.nodes[100] + 141, mode="centerline")

will return

(7.226243952925699, 148.54367737047386)

Batch distance

a new batch.py module can also find the top k nearest skeletons for multiple points

import skeliner as sk

# load some cells
dsgc = sk.io.load_npz("../../../data/DSGC/720575940567182923/skeleton.npz")
sac_offs = [sk.io.load_npz(f"../../../data/SAC-OFF/{cid}/skeleton.npz") for cid in [720575940547803011, 720575940554656742, 720575940572154471]]

# picks some points for testing
points = []
for i in range(3):
    contact_seeds = sk.pair.find_contact_seeds(dsgc, sac_offs[i])
    points.append(contact_seeds.pos[2])

sk.batch.nearest_skeletons(skeletons=[dsgc] + sac_offs, points=points, k=2)

this will return

[{'point_index': 0,
  'point': [499.73781427336235, 471.3932988369569, 43.056525600559425],
  'matches': [{'skeleton_index': 0,
    'skeleton_id': 720575940567182923,
    'distance': 0.0},
   {'skeleton_index': 1, 'skeleton_id': 720575940547803011, 'distance': 0.0}]},
 {'point_index': 1,
  'point': [420.4119215774796, 399.5634949644302, 55.745175985484856],
  'matches': [{'skeleton_index': 0,
    'skeleton_id': 720575940567182923,
    'distance': 0.0},
   {'skeleton_index': 2, 'skeleton_id': 720575940554656742, 'distance': 0.0}]},
 {'point_index': 2,
  'point': [403.9949115638043, 385.27809952845183, 57.7712639337827],
  'matches': [{'skeleton_index': 0,
    'skeleton_id': 720575940567182923,
    'distance': 0.0},
   {'skeleton_index': 3, 'skeleton_id': 720575940572154471, 'distance': 0.0}]}]

as expected, the test points are the contact seeds between the DSGC and three different OFF SAC, so the DSGC will always be the closest cell to all three points, and each SAC will be closest to one of the point.