Correspondences Class Reference

Class that compares two boards and generates correspondences across them. More...

Public Member Functions
def	__init__ (self, theParams=CfgCorrespondences(), initBoard=None)
	Constructor for the board matcher class. More...
def	adapt (self)
	Perform adaptation based on correspondence as applicable. More...
def	correct (self)
	Perform correspondence correction if sensible. More...
def	correspond (self)
	Get correspondences based on the most recent measurement (contained). More...
def	greedyAssignment (self, scoreTable_shape, scoreTable_color, scoreTable_edge_color)
	Run the greedyAssignment for the score table. More...
def	HungarianAssignment (self, scoreTable, getInverse=False)
	Run Hungarian Assignment for the score table. More...
def	matchPieces (self)
	Match all the measured puzzle pieces with board prior in a pairwise manner to get meas to prior. More...
def	process (self, bMeas)
	Run correspondence pipeline to preserve identity of puzzle pieces. More...

Static Public Member Functions
def	buildMatcher (typeStr, matchConfig=None)
	Build out a Matcher instance for data association / puzzle correspondences. More...

Public Attributes
	boardEstimate
	boardMeasurement
	matcher
	params
	pAssignments
	pAssignments_rotation
	scoreTable
	scoreTable_shape
	scoreType
	skipList

Detailed Description

Class that compares two boards and generates correspondences across them.

The comparison technique and properties are completely up to the programmer/engineer to establish. This class simply structures and faclitates the implementation. There is an implicit static assumption underlying the elements temporal evolution.

Constructor & Destructor Documentation

__init__()

def __init__	(		self,
			theParams = CfgCorrespondences(),
			initBoard = None
	)

Constructor for the board matcher class.

Parameters

[in]

solution

A solution/calibrated board instance. theParams: Any additional parameters in a structure.

Member Function Documentation

adapt()

def adapt

(

self

)

Perform adaptation based on correspondence as applicable.

If something special in the matching model depends on an underlying representation that is evolving over time and needs adaptation, then this is the place to implement. Adaptation differs from correction in that it may represent meta-parameters that support correspondence. These meta-parameters may need adjustment also.

buildMatcher()

def buildMatcher (   typeStr,   matchConfig = None  )

static

Build out a Matcher instance for data association / puzzle correspondences.

Parameters

[in]	typeStr	Matching approach as string specification.
[in]	matchConfig	Matching configuration (optional: uses default)

correct()

def correct

(

self

)

Perform correspondence correction if sensible.

If the correspondences rely on information that evolves in time and should be filtered or updated, this is where that gets implemented. Actions to occur here:

1. Update puzzle piece information (raw and feature).
2. Relabel the measured board to reflect association with known pieces.
3. Relabel status of estimated board to reflect status of (not) measured pieces.

correspond()

def correspond

(

self

)

Get correspondences based on the most recent measurement (contained).

greedyAssignment()

def greedyAssignment	(		self,
			scoreTable_shape,
			scoreTable_color,
			scoreTable_edge_color
	)

Run the greedyAssignment for the score table.

Args: scoreTable_shape: The score table for the pairwise comparison (shape). scoreTable_color: The score table for the pairwise comparison (color). scoreTable_edge_color: The score table for the pairwise comparison (edge_color).

Returns: matched_id: The matched pair dict.

HungarianAssignment()

def HungarianAssignment	(		self,
			scoreTable,
			getInverse = False
	)

Run Hungarian Assignment for the score table.

Atempts to associate row elements with column elements to define an assignment mapping that gives column elements (ideally) the same ID as the row elements. Per scipy documentation for linear_sum_assignment:

Can also solve a generalization of the classic assignment problem where the cost matrix is rectangular. If it has more rows than columns, then not every row needs to be assigned to a column, and vice versa.

The score table shape will influence the operation of the Hungarian Assignment implementation. If the matrix is:

• square, then association will be a 1-1 assignment;
• wide, then the associated will be a 1-1 assignment with unassigned columns;
• tall, then the association is permissive of unassigned row elements, not 1-1.

It is up to the outer scope to establish what is the best way to pass in the scoreTable and then to interpret the results. If the 1-1 association should operate in the opposite direction, then set getInverse to True.

Returning assignments as a dictionary to permit non-assignment.

Parameters

[in]	scoreTAble	Score table for the pairwise comparison.
[out]	matched_id	Matched pair dict.

matchPieces()

def matchPieces

(

self

)

Match all the measured puzzle pieces with board prior in a pairwise manner to get meas to prior.

Only gets matches, does not act on them.

The matching outcome is internally stored in member variable (pAssignments).

process()

def process	(		self,
			bMeas
	)

Run correspondence pipeline to preserve identity of puzzle pieces.

Parameters

[in]

bMeas

Measured board.

Member Data Documentation

boardEstimate

boardEstimate

boardMeasurement

boardMeasurement

matcher

matcher

params

params

pAssignments

pAssignments

pAssignments_rotation

pAssignments_rotation

scoreTable

scoreTable

scoreTable_shape

scoreTable_shape

scoreType

scoreType

skipList

skipList

---

The documentation for this class was generated from the following file:

• /home/pvela/python/puzzle_solver/puzzle/board.py