API Documentation¶

class cliquematch.Graph¶

use_heuristic¶

Search using the heuristic if true

Type: bool

use_dfs¶

Perform the depth-first search if true

Type: bool

lower_bound¶

Set the lower bound on the size of clique to find

Type: int

upper_bound¶

Set the upper bound on the size of clique to find

Type: int

time_limit¶

Set the time limit on the search

Type: float

current_vertex¶

The vertex which is about to be searched (Readonly)

Type: int

search_done¶

Whether the search has been completed (Readonly)

Type: bool

n_vertices¶

Number of vertices in the graph (Readonly)

Type: int

n_edges¶

Number of edges in the graph (Readonly)

Type: int

get_max_clique()¶

Finds a maximum clique in graph within the given bounds

Returns: the vertices in the maximum clique
Return type: list

continue_search()¶: Continue the clique search if the entire graph has not been searched

reset_search()¶: Reset the search for maximum cliques

static from_file()¶

Constructs Graph instance from reading a Matrix Market file

Parameters: str – filename
Returns: the loaded Graph

static from_edgelist()¶

Constructs Graph instance from the given edge list

Param: edgelist (numpy.ndarray of shape (n,2))
Parameters: int – num_vertices
Returns: the loaded Graph

static from_matrix()¶

Constructs Graph instance from the given boolean adjacency matrix

Param: adjmat (numpy.ndarray, bool square matrix)
Returns: the loaded Graph

static from_adjlist()¶

Constructs Graph instance from the given adjacency list

Parameters

int – num_vertices
int – num_edges
list – edges

Returns

the loaded Graph

to_file()¶

Exports Graph instance to a Matrix Market file

Parameters: str – filename

to_edgelist()¶

Exports Graph instance to an edge list

Returns: (n,2) numpy.ndarray of edges

to_matrix()¶

Exports Graph instance to a boolean matrix

Returns: square numpy.ndarray of bools

to_adjlist()¶

Exports Graph instance to an adjacency list

Returns: list of sets

class cliquematch.A2AGraph(set1, set2, d1=None, d2=None, is_d1_symmetric=True, is_d2_symmetric=True)¶

Correspondence Graph wrapper for array-to-array mappings.

S1¶

array elements are converted to numpy.float64

Type: numpy.ndarray

S2¶

array elements are converted to numpy.float64

Type: numpy.ndarray

d1¶

distance metric for elements in S1, defaults to Euclidean metric if None.

Type: callable (numpy.ndarray, int, int) -> float

d2¶

distance metric for elements in S2, defaults to Euclidean metric if None

Type: callable (numpy.ndarray, int, int) -> float

is_d1_symmetric¶

Type: bool

is_d2_symmetric¶

Type: bool

build_edges()¶

Build edges of the correspondence graph using distance metrics.

Checks d1 and d2 for defaults before passing to base class.

build_edges_with_condition(condition_func, use_cfunc_only)¶

Build edges of the correspondence graph using a given condition function.

Parameters

condition_func (callable) – must take parameters corresponding to S1, int, int, S2, int, int, and return bool
use_cfunc_only (bool) – if True, the distance metrics will not be used to filter out edges (slower)

Returns

True if construction was successful

get_correspondence(return_indices=True)¶

Get corresponding subsets between the S1 and S2.

Parameters: return_indices (bool) – if True return the indices of the corresponding elements, else return the elements

class cliquematch.L2LGraph(set1, set2, d1=None, d2=None, is_d1_symmetric=True, is_d2_symmetric=True)¶

Correspondence Graph wrapper for list-to-list mappings.

Any object can be passed for S1 and S2; the user is required to define how the elements are accessed.

S1¶

array elements are converted to numpy.float64

Type: object

S2¶

array elements are converted to numpy.float64

Type: object

d1¶

distance metric for elements in S1

Type: callable (list, int, int) -> float

d2¶

distance metric for elements in S2

Type: callable (list, int, int) -> float

is_d1_symmetric¶

Type: bool

is_d2_symmetric¶

Type: bool

build_edges()¶

Build edges of the correspondence graph using distance metrics.

Checks d1 and d2 for defaults before passing to base class.

build_edges_with_condition(condition_func, use_cfunc_only)¶

Build edges of the correspondence graph using a given condition function.

Parameters

condition_func (callable) – must take parameters corresponding to S1, int, int, S2, int, int, and return bool
use_cfunc_only (bool) – if True, the distance metrics will not be used to filter out edges (slower)

Returns

True if construction was successful

get_correspondence(return_indices=True)¶

Get corresponding subsets between the S1 and S2.

Parameters: return_indices (bool) – if True return the indices of the corresponding elements, else return the elements

class cliquematch.L2AGraph(set1, set2, d1=None, d2=None, is_d1_symmetric=True, is_d2_symmetric=True)¶

Correspondence Graph wrapper for list-to-array mappings.

Any general object can be passed for S1; the user is required to define how elements are accessed.

S1¶

Type: object

S2¶

array elements are converted to numpy.float64

Type: numpy.ndarray

d1¶

distance metric for elements in S1

Type: callable (list, int, int) -> float

d2¶

distance metric for elements in S2, defaults to Euclidean metric if None

Type: callable (numpy.ndarray, int, int) -> float

is_d1_symmetric¶

Type: bool

is_d2_symmetric¶

Type: bool

build_edges()¶

Build edges of the correspondence graph using distance metrics.

Checks d1 and d2 for defaults before passing to base class.

build_edges_with_condition(condition_func, use_cfunc_only)¶

Build edges of the correspondence graph using a given condition function.

Parameters

condition_func (callable) – must take parameters corresponding to S1, int, int, S2, int, int, and return bool
use_cfunc_only (bool) – if True, the distance metrics will not be used to filter out edges (slower)

Returns

True if construction was successful

get_correspondence(return_indices=True)¶

Get corresponding subsets between the S1 and S2.

Parameters: return_indices (bool) – if True return the indices of the corresponding elements, else return the elements

class cliquematch.A2LGraph(set1, set2, d1=None, d2=None, is_d1_symmetric=True, is_d2_symmetric=True)¶

Correspondence Graph wrapper for array-to-list mappings.

Any general object can be passed for S2; the user is required to define how elements are accessed.

S1¶

array elements are converted to numpy.float64

Type: numpy.ndarray

S2¶

Type: object

d1¶

distance metric for elements in S1, defaults to Euclidean metric if None

Type: callable (numpy.ndarray, int, int) -> float

d2¶

distance metric for elements in S2

Type: callable (list, int, int) -> float

is_d1_symmetric¶

Type: bool

is_d2_symmetric¶

Type: bool

build_edges()¶

Build edges of the correspondence graph using distance metrics.

Checks d1 and d2 for defaults before passing to base class.

build_edges_with_condition(condition_func, use_cfunc_only)¶

Build edges of the correspondence graph using a given condition function.

Parameters

condition_func (callable) – must take parameters corresponding to S1, int, int, S2, int, int, and return bool
use_cfunc_only (bool) – if True, the distance metrics will not be used to filter out edges (slower)

Returns

True if construction was successful

get_correspondence(return_indices=True)¶

Get corresponding subsets between the S1 and S2.

Parameters: return_indices (bool) – if True return the indices of the corresponding elements, else return the elements

class cliquematch.IsoGraph(set1, set2)¶

Correspondence graph for finding subgraph isomorphisms.

S1¶

Type: cliquematch.Graph

S2¶

Type: cliquematch.Graph

build_edges()¶: Build edges of the correspondence graph.

get_correspondence(return_indices=True)¶

Obtain corresponding vertices in the subgraph isomorphism.

Parameters: return_indices (bool) – if False will the return a tuple of 3 dict s: the subgraph of S1, the subgraph S2, a mapping from the vertices of S1 to S1
Returns: list or tuple (see above).

class cliquematch.AlignGraph(set1, set2)¶

Correspondence graph for aligning images using obtained interest points.

Uses a mask-based filtering method as a conditon function during construction of the graph. Default Euclidean metrics are used as distance metrics.

S1¶

array elements are converted to numpy.float64

Type: numpy.ndarray

S2¶

array elements are converted to numpy.float64

Type: numpy.ndarray

build_edges_with_filter(control_points, filter_mask, percentage)¶

Uses control points and a binary mask to filter out invalid mappings and construct a correspondence graph.

Parameters

control_points (numpy.ndarray) – control points to use in every alignment test
filter_mask (numpy.ndarray) – a boolean mask showing valid regions in the target image
percentage (float) – an alignment is valid if the number of control points that fall within the mask are greater than this value

get_correspondence(return_indices=False)¶

Find correspondence between the sets of points S1 and S2.

Parameters: return_indices (bool) – if True, returns only indices, else the below dict
Returns: dict of the two sets of corresponding points and the rotation/translation required to transform S1 to S2 (obtained via Kabsch Algorithm)