SoSupervisedTextureClassificationProcessing2d Class Reference
[Classification]

SoSupervisedTextureClassificationProcessing2d engine More...

#include <ImageViz/Engines/ImageSegmentation/Classification/SoSupervisedTextureClassificationProcessing2d.h>

Inheritance diagram for SoSupervisedTextureClassificationProcessing2d:

Public Types
enum	FeatureGroup { DIRECTIONAL_COOCCURRENCE = 1, ROTATION_INVARIANT_COOCCURRENCE = 2, FIRST_ORDER_STATISTICS = 4, HISTOGRAM_STATISTICS = 8, INTENSITY = 16 }
enum	CoocTextonShape { CUBE = 0, SPHERE = 1, BALL = 2 }
enum	OutMapType { CLOSEST_DISTANCE = 0, RELATIVE_DISTANCE = 1, CLASS_DISTANCE = 2, NONE = 3 }
Public Member Functions
	SoSupervisedTextureClassificationProcessing2d ()
Public Attributes
SoSFImageDataAdapter	inImage
SoSFImageDataAdapter	inTrainingImage
SoSFBitMask	featureGroup
SoSFVec2i32	radiusRange
SoSFUInt32	radiusStep
SoSFUInt32	coocRadius
SoSFEnum	coocTextonShape
SoSFUInt32	coocTextonSize
SoSFDouble	minSeparationPercentage
SoImageVizEngineOutput < SoSFImageDataAdapter, SoImageDataAdapter * >	outLabelImage
SoSFEnum	outMapType
SoImageVizEngineOutput < SoSFImageDataAdapter, SoImageDataAdapter * >	outMapImage

Detailed Description

SoSupervisedTextureClassificationProcessing2d engine

The SoSupervisedTextureClassificationProcessing3d engine realizes a segmentation based on a texture model automatically built from a training input image.

It chains automatically the 3 steps of the texture classification workflow: model creation, texture model learning and model application.

For an introduction see the Texture classification section

FILE FORMAT/DEFAULT

inImage	NULL
inTrainingImage	NULL
featureGroup	DIRECTIONAL_COOCCURRENCE \| ROTATION_INVARIANT_COOCCURRENCE \| FIRST_ORDER_STATISTICS \| HISTOGRAM_STATISTICS \| INTENSITY
radiusRange	2 14
radiusStep	4
coocRadius	10
coocTextonShape	SPHERE
coocTextonSize	4
minSeparationPercentage	3.0
outMapType	CLOSEST_DISTANCE

Library references: SupervisedTextureClassification2d

Member Enumeration Documentation

enum SoSupervisedTextureClassificationProcessing2d::CoocTextonShape

This enum defines all type of measures used for texture classification.

Enumerator:

CUBE

The set of all points associated to corners, middles of edges and middles of faces of a cube of size textonSize.

SPHERE

The set of all points situated at the same euclidean distance textonSize from the center.

This mode must be used when a repetitive texture is mono-scale.

BALL

The set of all points situated at a distance less or equal to textonSize from the center.

This mode can be useful to classify a multi-scale repetitive texture but may be very time consuming.

enum SoSupervisedTextureClassificationProcessing2d::FeatureGroup

This enum defines all type of measures used for texture classification.

Enumerator:

DIRECTIONAL_COOCCURRENCE	Features based on co-occurrence's matrix. One feature is extracted from each co-occurrence vector.
ROTATION_INVARIANT_COOCCURRENCE	Features based on co-occurrence's matrix. 3 statistics features are extracted from all vectors.
FIRST_ORDER_STATISTICS	Features based on first order statistics which are not computed using an histogram.
HISTOGRAM_STATISTICS	Features based on histogram statistics.
INTENSITY	Feature based on the intensity value of the input image.

enum SoSupervisedTextureClassificationProcessing2d::OutMapType

Enumerator:

CLOSEST_DISTANCE	The outMapImage represents the mahalanobis distance to the class selected by the classification. The closer to 0 this metric is, the more confident the classification is.
RELATIVE_DISTANCE	The outMapImage represents the mahalanobis distance to the class selected by the classification( d1 ) weighted by the gap with the second closest distance( d2 ). The smaller this metric is, the more confident and less ambiguous the classification is. Map Value = log( d1 / ( d2 - d1 ) ).
CLASS_DISTANCE	The outMapImage is a multichannel image where each channel represents the distance to the corresponding class.
NONE	No output map.

Constructor & Destructor Documentation

SoSupervisedTextureClassificationProcessing2d::SoSupervisedTextureClassificationProcessing2d ( )

Constructor.

Member Data Documentation

SoSFUInt32 SoSupervisedTextureClassificationProcessing2d::coocRadius

Radius of the circular neighborhood used by the cooccurrence features.

Default value is 10.

SoSFEnum SoSupervisedTextureClassificationProcessing2d::coocTextonShape

The shape of the co-occurrence texton, i.e., the pattern defined by the set of co-occurrence vectors.

This shape represents the distribution of points around the target point for computing the co-occurrence matrix. The shape, associated to the texton size, defines the set of vectors that are used for computing co-occurrence features. For instance, in 2D, a Cube shape of size 3 defines the co-occurrence vectors (-3, -3) , (0, -3), (3, -3), (-3, 0) , (3, 0), (-3, 3) , (0, 3) and (3, 3). This parameter is ignored if none of the co-occurrence measure types is selected. Use enum CoocTextonShape. Default is SPHERE

SoSFUInt32 SoSupervisedTextureClassificationProcessing2d::coocTextonSize

The size of the texton shape for co-occurrence features.

This size is constrained by the radius parameter. The constraint depends on the texton shape. For instance, with a square texton, the texton size cannot exceed the rounded value of

$radius \times \sqrt{2}$

. This parameter is ignored if none of the co-occurrence measure types is selected. Default value is 4.

SoSFBitMask SoSupervisedTextureClassificationProcessing2d::featureGroup

The groups of textural features to compute.

Use enum FeatureGroup. Default is DIRECTIONAL_COOCCURRENCE | ROTATION_INVARIANT_COOCCURRENCE | FIRST_ORDER_STATISTICS | HISTOGRAM_STATISTICS | INTENSITY

SoSFImageDataAdapter SoSupervisedTextureClassificationProcessing2d::inImage

The input grayscale image to segment.

Default value is NULL. Supported types include: grayscale image.

SoSFImageDataAdapter SoSupervisedTextureClassificationProcessing2d::inTrainingImage

The input label training image (16 or 32 bits).

Each label represents a class sample for the learning step. Default value is NULL. Supported types include: label image.

SoSFDouble SoSupervisedTextureClassificationProcessing2d::minSeparationPercentage

This parameter controls the rejection criteria of the feature selection algorithm (FS).

A measure is rejected if its contribution does not increase enough the separation power of the classification model. This ratio indicates the minimal relative growth required to keep a measure. Please refer to Feature Selection section for more information about this parameter. The value must be greater than or equal to 0.0. Default value is 3.0.

SoImageVizEngineOutput<SoSFImageDataAdapter,SoImageDataAdapter*> SoSupervisedTextureClassificationProcessing2d::outLabelImage

The output label image representing the texture classification result.

Default value is NULL. Supported types include: label image.

SoImageVizEngineOutput<SoSFImageDataAdapter,SoImageDataAdapter*> SoSupervisedTextureClassificationProcessing2d::outMapImage

Output map image.

The type of the image will be float. Default value is NULL. Supported types include: grayscale binary label color image.

SoSFEnum SoSupervisedTextureClassificationProcessing2d::outMapType

The Output map image type.

Default is CLOSEST_DISTANCE. Use enum OutMapType. Default is CLOSEST_DISTANCE

SoSFVec2i32 SoSupervisedTextureClassificationProcessing2d::radiusRange

The minimum and maximum radius of the circular neighborhoods used for computing textural features.

Default value is SbVec2i32(2,14).