# -*- python -*-
#
# Copyright INRIA - CIRAD - INRA
#
# Distributed under the Cecill-C License.
# See accompanying file LICENSE.txt or copy at
# http://www.cecill.info/licences/Licence_CeCILL-C_V1-en.html
#
# ==============================================================================
from __future__ import division, print_function, absolute_import
import numpy
from ..object import Image3D
# ==============================================================================
[docs]
def remove_internal(image_3d):
len_x, len_y, len_z = image_3d.shape
im = Image3D.zeros(
(len_x + 2, len_y + 2, len_z + 2), voxels_size=image_3d.voxels_size
)
im[1:-1, 1:-1, 1:-1] = image_3d
xx, yy, zz = numpy.where(image_3d == 1)
ijk = [
(-1, -1, -1),
(-1, -1, 0),
(-1, -1, 1),
(-1, 0, -1),
(-1, 0, 0),
(-1, 0, 1),
(-1, 1, -1),
(-1, 1, 0),
(-1, 1, 1),
(0, -1, -1),
(0, -1, 0),
(0, -1, 1),
(0, 0, -1),
(0, 0, 0),
(0, 0, 1),
(0, 1, -1),
(0, 1, 0),
(0, 1, 1),
(1, -1, -1),
(1, -1, 0),
(1, -1, 1),
(1, 0, -1),
(1, 0, 0),
(1, 0, 1),
(1, 1, -1),
(1, 1, 0),
(1, 1, 1),
]
def removable(x, y, z):
for i, j, k in ijk:
if im[x + i, y + j, z + k] == 0:
return False
return True
im2 = Image3D.zeros_like(im)
for i in range(len(xx)):
x, y, z = xx[i], yy[i], zz[i]
if removable(x, y, z):
im2[(x, y, z)] = 1
result = im - im2
return result[1:-1, 1:-1, 1:-1]
[docs]
def labeling_connected_component(image_3d):
len_x, len_y, len_z = image_3d.shape
im = Image3D.zeros((len_x + 2, len_y + 2, len_z + 2))
im[1:-1, 1:-1, 1:-1] = image_3d
xx, yy, zz = numpy.where(im == 1)
mat = Image3D.zeros_like(im)
def get_neighbors(x, y, z):
neighbors = []
for i in [-1, 0, 1]:
for j in [-1, 0, 1]:
for k in [-1, 0, 1]:
ind = x + i, y + j, z + k
if im[ind] == 1:
neighbors.append(ind)
return neighbors
num_label = 1
for i, xx_val in enumerate(xx):
x, y, z = xx_val, yy[i], zz[i]
if mat[x, y, z] == 0:
mat[x, y, z] = num_label
neighbors = get_neighbors(x, y, z)
while neighbors:
xxx, yyy, zzz = neighbors.pop()
if mat[xxx, yyy, zzz] == 0:
mat[xxx, yyy, zzz] = num_label
neighbors += get_neighbors(xxx, yyy, zzz)
num_label += 1
return mat[1:-1, 1:-1, 1:-1]
[docs]
def kept_biggest_connected_component(image_3d):
mat = labeling_connected_component(image_3d)
max_value = 0
save = None
for i in range(1, numpy.max(mat) + 1):
nb = len(numpy.where(mat == i)[0])
if nb > max_value:
max_value = nb
save = i
mat[mat != save] = 0
mat[mat == save] = 1
return mat
