superdsm.atoms

class superdsm.atoms.AtomAdjacencyGraph(atoms, clusters, fg_mask, seeds, out=None)

Bases: object

Graph representation of the adjacencies of atomic image regions.

This corresponds to the adjacency graph \(\mathcal G\) as defined in Coarse-to-fine region analysis.

Parameters:

atoms – Integer-valued image representing the universe of atomic image regions. Each atomic image region has a unique label, which is the integer value.
clusters – Integer-valued image representing the regions of possibly clustered obejcts. Each region has a unique label, which is the integer value.
fg_mask – Binary image corresponding to a rough representation of the image foreground. This means that an image point \(x \in \Omega\) is True if \(Y_\omega|_{\omega=\{x\}} > 0\) and False otherwise.
seeds – The seed points which were used to determine the atomic image regions, represented by a list of tuples of coordinates. The Default pipeline only uses these for rendering the adjacency graph (see the get_edge_lines() method).
out – An instance of an Output sub-class, 'muted' if no output should be produced, or None if the default output should be used.

>>> import superdsm.atoms
>>> import numpy as np
>>> atoms = np.array([[1, 1, 2, 4],
...                   [1, 3, 2, 4],
...                   [3, 3, 3, 4]])
>>> clusters = np.array([[1, 1, 2, 2],
...                      [1, 2, 2, 2],
...                      [2, 2, 2, 2]])
>>> fg_mask = np.array([[True, False, True, False],
...                     [True, False, True,  True],
...                     [True,  True, True,  True]])
>>> seeds = [(0, 0), (0, 2), (2, 1), (1, 3)]
>>> adj = superdsm.atoms.AtomAdjacencyGraph(atoms, clusters, fg_mask, seeds, 'muted')
>>> adj[1]
set()
>>> adj[2]
{3, 4}
>>> adj[3]
{2, 4}
>>> adj[4]
{2, 3}

property atom_labels

The set of labels of all atomic image regions.

>>> import superdsm.atoms
>>> import numpy as np
>>> atoms = np.array([[1, 1, 2, 4],
...                   [1, 3, 2, 4],
...                   [3, 3, 3, 4]])
>>> clusters = np.array([[1, 1, 2, 2],
...                      [1, 2, 2, 2],
...                      [2, 2, 2, 2]])
>>> fg_mask = np.ones(atoms.shape, bool)
>>> seeds = [(0, 0), (0, 2), (2, 1), (1, 3)]
>>> adj = superdsm.atoms.AtomAdjacencyGraph(atoms, clusters, fg_mask, seeds, 'muted')
>>> adj.atom_labels
frozenset({1, 2, 3, 4})

property cluster_labels

The set of labels of all regions of possibly clustered objects.

>>> import superdsm.atoms
>>> import numpy as np
>>> atoms = np.array([[1, 1, 2, 4],
...                   [1, 3, 2, 4],
...                   [3, 3, 3, 4]])
>>> clusters = np.array([[1, 1, 2, 2],
...                      [1, 2, 2, 2],
...                      [2, 2, 2, 2]])
>>> fg_mask = np.ones(atoms.shape, bool)
>>> seeds = [(0, 0), (0, 2), (2, 1), (1, 3)]
>>> adj = superdsm.atoms.AtomAdjacencyGraph(atoms, clusters, fg_mask, seeds, 'muted')
>>> adj.cluster_labels
frozenset({1, 2})

get_atom_degree(atom_label)

Returns the number of adjacent atomic image regions.

>>> import superdsm.atoms
>>> import numpy as np
>>> atoms = np.array([[1, 1, 2, 4],
...                   [1, 3, 2, 4],
...                   [3, 3, 3, 4]])
>>> clusters = np.array([[1, 1, 2, 2],
...                      [1, 2, 2, 2],
...                      [2, 2, 2, 2]])
>>> fg_mask = np.array([[True, False, True, False],
...                     [True, False, True,  True],
...                     [True,  True, True,  True]])
>>> seeds = [(0, 0), (0, 2), (2, 1), (1, 3)]
>>> adj = superdsm.atoms.AtomAdjacencyGraph(atoms, clusters, fg_mask, seeds, 'muted')
>>> adj.get_atom_degree(1)
0
>>> adj.get_atom_degree(2)
2
>>> adj.get_atom_degree(3)
2
>>> adj.get_atom_degree(4)
2

get_atoms_in_cluster(cluster_label)

Returns the set of labels of all atomic image regions, which are part of a region of possibly clustered objects.

>>> import superdsm.atoms
>>> import numpy as np
>>> atoms = np.array([[1, 1, 2, 4],
...                   [1, 3, 2, 4],
...                   [3, 3, 3, 4]])
>>> clusters = np.array([[1, 1, 2, 2],
...                      [1, 2, 2, 2],
...                      [2, 2, 2, 2]])
>>> fg_mask = np.ones(atoms.shape, bool)
>>> seeds = [(0, 0), (0, 2), (2, 1), (1, 3)]
>>> adj = superdsm.atoms.AtomAdjacencyGraph(atoms, clusters, fg_mask, seeds, 'muted')
>>> adj.get_atoms_in_cluster(1)
{1}
>>> adj.get_atoms_in_cluster(2)
{2, 3, 4}

get_cluster_label(atom_label)

Returns the label of the region of possibly clustered objects, which an atomic image region is a part of.

>>> import superdsm.atoms
>>> import numpy as np
>>> atoms = np.array([[1, 1, 2, 4],
...                   [1, 3, 2, 4],
...                   [3, 3, 3, 4]])
>>> clusters = np.array([[1, 1, 2, 2],
...                      [1, 2, 2, 2],
...                      [2, 2, 2, 2]])
>>> fg_mask = np.ones(atoms.shape, bool)
>>> seeds = [(0, 0), (0, 2), (2, 1), (1, 3)]
>>> adj = superdsm.atoms.AtomAdjacencyGraph(atoms, clusters, fg_mask, seeds, 'muted')
>>> adj.get_cluster_label(1)
1
>>> adj.get_cluster_label(2)
2
>>> adj.get_cluster_label(3)
2
>>> adj.get_cluster_label(4)
2

get_edge_lines(accept='all', reduce=True)

Returns a list of lines corresponding to the edges of the graph.

Parameters:

accept – Must be either all or a callable. If all is used, all edges of the graph are included. Otherwise, an edge (i,j) is included only if accept(i) and accept(j) evaluate to True, where i and j are the labels of two adjacent atomic image regions.
reduce – If True, then an edge (i,j) is included only if i > j. Otherwise, both edges (i,j) and (j,i) are included.

Each line is a tuple of two seed points, and each seed point is a tuple of coordinates.

>>> import superdsm.atoms
>>> import numpy as np
>>> atoms = np.array([[1, 1, 2, 4],
...                   [1, 3, 2, 4],
...                   [3, 3, 3, 4]])
>>> clusters = np.array([[1, 1, 2, 2],
...                      [1, 2, 2, 2],
...                      [2, 2, 2, 2]])
>>> fg_mask = np.array([[True, False, True, False],
...                     [True, False, True,  True],
...                     [True,  True, True,  True]])
>>> seeds = [(0, 0), (0, 2), (2, 1), (1, 3)]
>>> adj = superdsm.atoms.AtomAdjacencyGraph(atoms, clusters, fg_mask, seeds, 'muted')
>>> adj.get_edge_lines()
[((0, 2), (2, 1)), ((0, 2), (1, 3)), ((2, 1), (1, 3))]
>>> adj.get_edge_lines(lambda i: i != 4)
[((0, 2), (2, 1))]
>>> adj.get_edge_lines(lambda i: i != 4, reduce=False)
[((0, 2), (2, 1)), ((2, 1), (0, 2))]

get_seed(atom_label)

Returns the seed point which was used to determine an atomic image region.

Returns:: Tuple of the coordinates of the seed point.

>>> import superdsm.atoms
>>> import numpy as np
>>> atoms = np.array([[1, 1, 2, 4],
...                   [1, 3, 2, 4],
...                   [3, 3, 3, 4]])
>>> clusters = np.array([[1, 1, 2, 2],
...                      [1, 2, 2, 2],
...                      [2, 2, 2, 2]])
>>> fg_mask = np.ones(atoms.shape, bool)
>>> seeds = [(0, 0), (0, 2), (2, 1), (1, 3)]
>>> adj = superdsm.atoms.AtomAdjacencyGraph(atoms, clusters, fg_mask, seeds, 'muted')
>>> adj.get_seed(1)
(0, 0)
>>> adj.get_seed(2)
(0, 2)
>>> adj.get_seed(3)
(2, 1)
>>> adj.get_seed(4)
(1, 3)

property max_degree

The maximum degree of the graph.

>>> import superdsm.atoms
>>> import numpy as np
>>> atoms = np.array([[1, 1, 2, 4],
...                   [1, 3, 2, 4],
...                   [3, 3, 3, 4]])
>>> clusters = np.array([[1, 1, 2, 2],
...                      [1, 2, 2, 2],
...                      [2, 2, 2, 2]])
>>> fg_mask = np.array([[True, False, True, False],
...                     [True, False, True,  True],
...                     [True,  True, True,  True]])
>>> seeds = [(0, 0), (0, 2), (2, 1), (1, 3)]
>>> adj = superdsm.atoms.AtomAdjacencyGraph(atoms, clusters, fg_mask, seeds, 'muted')
>>> adj.max_degree
2