Derived projections and contour materializations of saliency maps. More...

#include <HierarchySaliencyMapProjection.hpp>

Static Public Member Functions
template<class Value >
static ImagePtr< double >	edgeMapToPixelImage (const EdgeSaliencyMap< Value > &edgeMap, EdgeToPixelReducer reducer=EdgeToPixelReducer::Max)
	Rasterizes an edge-indexed map into a pixel image for display.

template<class Value , class Threshold >
static EdgeContourMap	thresholdCut (const EdgeSaliencyMap< Value > &edgeMap, Threshold threshold)
	Thresholds an edge saliency map into an edge contour set.

static NodeContourEdgeMap	nodeContourEdges (const MorphologicalTree &tree, const RegularGridAdjacency2D &adjacency)
	Projects transition adjacency edges onto their hierarchy node.

static NodeContourEdgeMap	nodeContourEdges (const MorphologicalTree &tree)
	Projects stored-adjacency transition edges onto hierarchy nodes.

template<AltitudeValue T>
static NodeContourEdgeMap	nodeContourEdges (const ValuedMorphologicalTree< T > &tree, const RegularGridAdjacency2D &adjacency)
	Projects valued-tree adjacency edges onto their hierarchy node.

template<AltitudeValue T>
static NodeContourEdgeMap	nodeContourEdges (const ValuedMorphologicalTree< T > &tree)
	Projects valued-tree stored-adjacency edges onto hierarchy nodes.

static IncrementalNodeContourMap	computeIncrementalNodeContours (const MorphologicalTree &tree, const RegularGridAdjacency2D &adjacency)
	Computes per-node incremental contour edges with explicit adjacency.

static IncrementalNodeContourMap	computeIncrementalNodeContours (const MorphologicalTree &tree)
	Computes per-node incremental contour edges using stored adjacency.

template<AltitudeValue T>
static IncrementalNodeContourMap	computeIncrementalNodeContours (const ValuedMorphologicalTree< T > &tree, const RegularGridAdjacency2D &adjacency)
	Computes valued-tree incremental contours with explicit adjacency.

template<AltitudeValue T>
static IncrementalNodeContourMap	computeIncrementalNodeContours (const ValuedMorphologicalTree< T > &tree)
	Computes valued-tree incremental contours using stored adjacency.

template<class Value >
static EdgeSaliencyMap< Value >	projectNodeValuation (const IncrementalNodeContourMap &contours, std::span< const Value > nodeValuation)
	Projects dense node valuation onto transition contour edges.

template<class Value , class Threshold >
static EdgeContourMap	thresholdByNodeValuation (const IncrementalNodeContourMap &contours, std::span< const Value > nodeValuation, Threshold threshold)
	Thresholds incremental contour edges by dense node valuation.

Detailed Description

Derived projections and contour materializations of saliency maps.

This class does not construct the formal edge-indexed saliency map. It consumes edge maps, trees, or precomputed transition-contour slices to produce display images, threshold cuts, and sparse transition-edge projections.

Definition at line 168 of file HierarchySaliencyMapProjection.hpp.

Member Function Documentation

◆ computeIncrementalNodeContours() [1/4]

static IncrementalNodeContourMap mmcfilters::HierarchySaliencyMapProjection::computeIncrementalNodeContours ( const MorphologicalTree & tree )

inlinestatic

Computes per-node incremental contour edges using stored adjacency.

Parameters

tree	Tree topology.

Returns: The computed per-node incremental contour edges using stored adjacency.

Definition at line 489 of file HierarchySaliencyMapProjection.hpp.

◆ computeIncrementalNodeContours() [2/4]

static IncrementalNodeContourMap mmcfilters::HierarchySaliencyMapProjection::computeIncrementalNodeContours	(	const MorphologicalTree &	tree,
		const RegularGridAdjacency2D &	adjacency
	)

inlinestatic

Computes per-node incremental contour edges with explicit adjacency.

The output stores one contiguous edge slice per dense internal NodeId. A transition edge belongs to node u iff its endpoint smallest nodes are distinct and u = LCA(smallestNode(source), smallestNode(target)).

Parameters

tree	Tree topology.
adjacency	Adjacency relation.

Returns: The computed per-node incremental contour edges with explicit adjacency.

Definition at line 479 of file HierarchySaliencyMapProjection.hpp.

◆ computeIncrementalNodeContours() [3/4]

template<AltitudeValue T>

static IncrementalNodeContourMap mmcfilters::HierarchySaliencyMapProjection::computeIncrementalNodeContours ( const ValuedMorphologicalTree< T > & tree )

inlinestatic

Computes valued-tree incremental contours using stored adjacency.

Parameters

tree	Tree topology.

Returns: The computed valued-tree incremental contours using stored adjacency.

Definition at line 512 of file HierarchySaliencyMapProjection.hpp.

◆ computeIncrementalNodeContours() [4/4]

template<AltitudeValue T>

static IncrementalNodeContourMap mmcfilters::HierarchySaliencyMapProjection::computeIncrementalNodeContours	(	const ValuedMorphologicalTree< T > &	tree,
		const RegularGridAdjacency2D &	adjacency
	)

inlinestatic

Computes valued-tree incremental contours with explicit adjacency.

Parameters

tree	Tree topology.
adjacency	Adjacency relation.

Returns: The computed valued-tree incremental contours with explicit adjacency.

Definition at line 501 of file HierarchySaliencyMapProjection.hpp.

◆ edgeMapToPixelImage()

template<class Value >

static ImagePtr< double > mmcfilters::HierarchySaliencyMapProjection::edgeMapToPixelImage	(	const EdgeSaliencyMap< Value > &	edgeMap,
		EdgeToPixelReducer	reducer = `EdgeToPixelReducer::Max`
	)

inlinestatic

Rasterizes an edge-indexed map into a pixel image for display.

This helper is only a visualization projection. It does not change the formal edge-indexed saliency representation. Each edge value contributes to both endpoint pixels. Max writes the maximum incident edge value per pixel, while Mean writes the arithmetic mean of incident edge values. Pixels with no incident edges keep value 0.

Parameters

edgeMap	Edge-indexed saliency or score map.
reducer	Pixel aggregation policy.

Returns: Row-major double image on the same image domain as edgeMap.

Definition at line 301 of file HierarchySaliencyMapProjection.hpp.

◆ nodeContourEdges() [1/4]

static NodeContourEdgeMap mmcfilters::HierarchySaliencyMapProjection::nodeContourEdges ( const MorphologicalTree & tree )

inlinestatic

Projects stored-adjacency transition edges onto hierarchy nodes.

Parameters

tree	Tree topology.

Returns: The projected stored-adjacency transition edges onto hierarchy nodes.

Definition at line 437 of file HierarchySaliencyMapProjection.hpp.

◆ nodeContourEdges() [2/4]

static NodeContourEdgeMap mmcfilters::HierarchySaliencyMapProjection::nodeContourEdges	(	const MorphologicalTree &	tree,
		const RegularGridAdjacency2D &	adjacency
	)

inlinestatic

Projects transition adjacency edges onto their hierarchy node.

Edges whose endpoints already have the same smallest hierarchy node are omitted. They are internal to the finest region represented by that smallest node and carry the implicit base value 0 in a full formal saliency map.

Parameters

tree	Tree topology.
adjacency	Adjacency relation.

Returns: The projected transition adjacency edges onto their hierarchy node.

Definition at line 393 of file HierarchySaliencyMapProjection.hpp.

◆ nodeContourEdges() [3/4]

template<AltitudeValue T>

static NodeContourEdgeMap mmcfilters::HierarchySaliencyMapProjection::nodeContourEdges ( const ValuedMorphologicalTree< T > & tree )

inlinestatic

Projects valued-tree stored-adjacency edges onto hierarchy nodes.

Parameters

tree	Tree topology.

Returns: The projected valued-tree stored-adjacency edges onto hierarchy nodes.

Definition at line 464 of file HierarchySaliencyMapProjection.hpp.

◆ nodeContourEdges() [4/4]

template<AltitudeValue T>

static NodeContourEdgeMap mmcfilters::HierarchySaliencyMapProjection::nodeContourEdges	(	const ValuedMorphologicalTree< T > &	tree,
		const RegularGridAdjacency2D &	adjacency
	)

inlinestatic

Projects valued-tree adjacency edges onto their hierarchy node.

Parameters

tree	Tree topology.
adjacency	Adjacency relation.

Returns: The projected valued-tree adjacency edges onto their hierarchy node.

Definition at line 454 of file HierarchySaliencyMapProjection.hpp.

◆ projectNodeValuation()

template<class Value >

static EdgeSaliencyMap< Value > mmcfilters::HierarchySaliencyMapProjection::projectNodeValuation	(	const IncrementalNodeContourMap &	contours,
		std::span< const Value >	nodeValuation
	)

inlinestatic

Projects dense node valuation onto transition contour edges.

This is a sparse projection over the transition-edge support returned by computeIncrementalNodeContours. Edges internal to the finest represented regions are omitted and have implicit value 0 in the full formal saliency map. Use HierarchySaliencyMap::computeSaliencyEdgeMap when the output must explicitly contain every graph edge.

Parameters

contours	Contour data.
nodeValuation	Dense valuation indexed by node identifier.

Returns: The projected dense node valuation onto transition contour edges.

Definition at line 530 of file HierarchySaliencyMapProjection.hpp.

◆ thresholdByNodeValuation()

template<class Value , class Threshold >

static EdgeContourMap mmcfilters::HierarchySaliencyMapProjection::thresholdByNodeValuation	(	const IncrementalNodeContourMap &	contours,
		std::span< const Value >	nodeValuation,
		Threshold	threshold
	)

inlinestatic

Thresholds incremental contour edges by dense node valuation.

This is a generic fast cut helper over precomputed transition-contour slices. It does not see the tree topology and therefore cannot validate that nodeValuation is a formal hierarchy valuation. Validate the node valuation separately when this cut is intended to represent a quasi-flat-zone saliency level.

Parameters

contours	Contour data.
nodeValuation	Dense valuation indexed by node identifier.
threshold	Threshold applied by the operation.

Returns: The thresholded incremental contour edges by dense node valuation.

Definition at line 571 of file HierarchySaliencyMapProjection.hpp.

◆ thresholdCut()

template<class Value , class Threshold >

static EdgeContourMap mmcfilters::HierarchySaliencyMapProjection::thresholdCut	(	const EdgeSaliencyMap< Value > &	edgeMap,
		Threshold	threshold
	)

inlinestatic

Thresholds an edge saliency map into an edge contour set.

A contour edge is selected when saliency(edge) >= threshold. This is the cut complement of the quasi-flat-zone convention where graph edges with saliency strictly below the level remain connected.

Parameters

edgeMap	Edge saliency map to threshold.
threshold	Threshold applied by the operation.

Returns: The thresholded edge saliency map into an edge contour set.

Definition at line 362 of file HierarchySaliencyMapProjection.hpp.

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

mmcfilters/trees/saliency/HierarchySaliencyMapProjection.hpp

Static Public Member Functions

Detailed Description

Member Function Documentation

◆ computeIncrementalNodeContours() [1/4]

◆ computeIncrementalNodeContours() [2/4]

◆ computeIncrementalNodeContours() [3/4]

◆ computeIncrementalNodeContours() [4/4]

◆ edgeMapToPixelImage()

◆ nodeContourEdges() [1/4]

◆ nodeContourEdges() [2/4]

◆ nodeContourEdges() [3/4]

◆ nodeContourEdges() [4/4]

◆ projectNodeValuation()

◆ thresholdByNodeValuation()

◆ thresholdCut()