lti::colorModelEstimator Class Reference

This functor is used to create three-dimensional histograms for the colors of images. More...

#include <ltiColorModelEstimator.h>

Inheritance diagram for lti::colorModelEstimator:

[legend]

Collaboration diagram for lti::colorModelEstimator:

[legend]

List of all members.

Classes
class	parameters
	The parameters for the class colorModelEstimator. More...
Public Member Functions
	colorModelEstimator ()
	colorModelEstimator (const colorModelEstimator &other)
virtual	~colorModelEstimator ()
virtual const char *	getTypeName () const
bool	reset ()
bool	consider (const rgbPixel &pix, const int times=1)
bool	smoothConsider (const rgbPixel &pix, const int times=1)
bool	consider (const float &c1, const float &c2, const float &c3, const int times=1)
bool	consider (const palette &pal)
bool	consider (const image &img)
bool	consider (const channel &c1, const channel &c2, const channel &c3)
bool	consider (const image &img, const channel8 &mask)
bool	consider (const image &img, const imatrix &mask)
bool	consider (const channel &c1, const channel &c2, const channel &c3, const channel8 &mask)
bool	consider (const channel &c1, const channel &c2, const channel &c3, const imatrix &mask)
bool	consider (const image &img, const channel8 &mask, const ubyte label)
bool	consider (const image &img, const imatrix &mask, const int label)
bool	consider (const std::list< image > &imgs)
bool	consider (const std::list< image > &imgs, const std::list< channel8 > &masks)
const thistogram< double > &	getColorModel ()
bool	apply (thistogram< double > &model) const
bool	apply (trgbPixel< double > &mean, dmatrix &covar) const
colorModelEstimator &	copy (const colorModelEstimator &other)
colorModelEstimator &	operator= (const colorModelEstimator &other)
virtual functor *	clone () const
virtual bool	updateParameters ()
const parameters &	getParameters () const
Protected Member Functions
void	outer (const trgbPixel< double > &a, const trgbPixel< double > &b, dmatrix &mat) const
Protected Attributes
thistogram< double >	theModel
trgbPixel< double >	sum
dmatrix	sumOuter
int	n

---

Detailed Description

This functor is used to create three-dimensional histograms for the colors of images.

The histograms are used as color models in functors like lti::colorModelSelector and lti::probabilityMap.

It is possible to provide whole images (usefull to create histograms of backgrounds, if they can be captured as a whole image), or images with corresponding masks, that allow to consider only those pixels with a mask value different than zero.

It is also possible to provide directly a list of images.

This functor generates also simple statistics for a multivariate gaussian model for the color (mean color and covariance matrix)

---

Constructor & Destructor Documentation

lti::colorModelEstimator::colorModelEstimator

(

)

Default constructor.

lti::colorModelEstimator::colorModelEstimator

(

const colorModelEstimator &

other

)

Copy constructor.

Parameters:

other

the object to be copied

virtual lti::colorModelEstimator::~colorModelEstimator

(

)

[virtual]

Destructor.

---

Member Function Documentation

bool lti::colorModelEstimator::apply	(	trgbPixel< double > &	mean,
		dmatrix &	covar
	)			const

Get the mean color and covariance matrix of the internal model.

Parameters:

	mean	the mean color
	covar	covariance matrix of the color (3x3 matrix)

Returns:

true if apply successful, false otherwise.

bool lti::colorModelEstimator::apply

(

thistogram< double > &

model

)

const

Get a copy of the internal model in the given parameter.

Parameters:

model

the color model (as 3D histogram) computed until now

Returns:

true if apply successful or false otherwise.

virtual functor* lti::colorModelEstimator::clone

(

)

const [virtual]

Returns a pointer to a clone of this functor.

Reimplemented from lti::statisticsFunctor.

bool lti::colorModelEstimator::consider	(	const std::list< image > &	imgs,
		const std::list< channel8 > &	masks
	)

Consider all the images in the given list.

The mask list must have the same size, and the corresponding images must also have the same size.

Parameters:

	imgs	list of images to be considered.
	masks	list of all masks.

Returns:

true if successful or false otherwise.

bool lti::colorModelEstimator::consider

(

const std::list< image > &

imgs

)

Consider all the images in the given list.

Parameters:

imgs

list of images to be considered

Returns:

true if successful or false otherwise.

bool lti::colorModelEstimator::consider	(	const image &	img,
		const imatrix &	mask,
		const int	label
	)

Consider the pixels in the image, for which the corresponding pixel in the mask contains the value given by label.

The mask and the image must have the same size.

This is seldom used. See lti::computePalette, maybe that is what you are looking for!

Parameters:

	img	image with colors to be considered
	mask	imatrix with the mask.
	label	only those pixel whose value in the mask are equal to this parameters will be considered.

Returns:

true if successful or false otherwise.

bool lti::colorModelEstimator::consider	(	const image &	img,
		const channel8 &	mask,
		const ubyte	label
	)

Consider the pixels in the image, for which the corresponding pixel in the mask contains the value given by label.

The mask and the image must have the same size.

This is seldom used. See lti::computePalette, maybe that is what you are looking for!

Parameters:

	img	image with colors to be considered
	mask	channel8 with the mask.
	label	only those pixel whose value in the mask are equal to this parameters will be considered.

Returns:

true if successful or false otherwise.

bool lti::colorModelEstimator::consider	(	const channel &	c1,
		const channel &	c2,
		const channel &	c3,
		const imatrix &	mask
	)

Consider the pixels in the image, for which the corresponding pixel in the mask is not zero.

The mask and the image must have the same size.

Parameters:

	c1	first channel.
	c2	second channel.
	c3	third channel.
	mask	imatrix with the mask. A value of zero means, that the corresponding pixel in the image must be ignored.

Returns:

true if successful or false otherwise.

bool lti::colorModelEstimator::consider	(	const channel &	c1,
		const channel &	c2,
		const channel &	c3,
		const channel8 &	mask
	)

Consider the pixels in the image, for which the corresponding pixel in the mask is not zero.

The mask and the image must have the same size.

Parameters:

	c1	first channel.
	c2	second channel.
	c3	third channel.
	mask	channel8 with the mask. A value of zero means, that the corresponding pixel in the image must be ignored.

Returns:

true if successful or false otherwise.

bool lti::colorModelEstimator::consider	(	const image &	img,
		const imatrix &	mask
	)

Consider the pixels in the image, for which the corresponding pixel in the mask is not zero.

The mask and the image must have the same size.

Parameters:

	img	image with colors to be considered
	mask	imatrix with the mask. A value of zero means, that the corresponding pixel in the image must be ignored.

Returns:

true if successful or false otherwise.

bool lti::colorModelEstimator::consider	(	const image &	img,
		const channel8 &	mask
	)

Consider the pixels in the image, for which the corresponding pixel in the mask is not zero.

The mask and the image must have the same size.

Parameters:

	img	image with colors to be considered
	mask	channel8 with the mask. A value of zero means, that the corresponding pixel in the image must be ignored.

Returns:

true if successful or false otherwise.

bool lti::colorModelEstimator::consider	(	const channel &	c1,
		const channel &	c2,
		const channel &	c3
	)

Consider all pixels defined by the correspondent elements in the three given channels.

Parameters:

	c1	first channel.
	c2	second channel.
	c3	third channel.

Returns:

true if successful or false otherwise.

bool lti::colorModelEstimator::consider

(

const image &

img

)

Consider all pixels in the image.

Parameters:

img

image with colors to be considered

Returns:

true if successful or false otherwise.

bool lti::colorModelEstimator::consider

(

const palette &

pal

)

Consider a pixel for the model.

Parameters:

pal

the color palette to be considered

Returns:

true if successful or false otherwise.

bool lti::colorModelEstimator::consider	(	const float &	c1,
		const float &	c2,
		const float &	c3,
		const int	times = 1
	)

Consider a three dimensional pixel value given its three components.

It is assumed that the value range of all channels lies between 0.0 and 1.0.

Parameters:

	c1	first color component.
	c2	second color component.
	c3	third color component.
	times	number of times the given color point must be consider.

bool lti::colorModelEstimator::consider	(	const rgbPixel &	pix,
		const int	times = 1
	)

Consider a pixel for the model.

Parameters:

	pix	pixel to be considered
	times	how many times should the pixel be considered

Returns:

true if successful or false otherwise.

colorModelEstimator& lti::colorModelEstimator::copy

(

const colorModelEstimator &

other

)

Copy data of "other" functor.

Parameters:

other

the functor to be copied

Returns:

a reference to this functor object

Reimplemented from lti::statisticsFunctor.

const thistogram<double>& lti::colorModelEstimator::getColorModel

(

)

Get a constant reference to the internal model.

const parameters& lti::colorModelEstimator::getParameters

(

)

const

Returns used parameters.

Reimplemented from lti::statisticsFunctor.

virtual const char* lti::colorModelEstimator::getTypeName

(

)

const [virtual]

Returns the name of this type ("colorModelEstimator").

Reimplemented from lti::statisticsFunctor.

colorModelEstimator& lti::colorModelEstimator::operator=

(

const colorModelEstimator &

other

)

Alias for copy member.

Parameters:

other

the functor to be copied

Returns:

a reference to this functor object

Reimplemented from lti::functor.

void lti::colorModelEstimator::outer	(	const trgbPixel< double > &	a,
		const trgbPixel< double > &	b,
		dmatrix &	mat
	)			const [protected]

Outer product of two colors.

The matrix MUST be 3x3 before calling this method.

bool lti::colorModelEstimator::reset

(

)

Reset the internal histogram to consider new elements.

The internal model will be empty, but with the correct size.

bool lti::colorModelEstimator::smoothConsider	(	const rgbPixel &	pix,
		const int	times = 1
	)

Consider a pixel for the model.

This more time consuming consider method will not only add 1 to the "winner" bin, but will also distribute the "1" vote using tri-linear interpolation among the seven neighbors.

Parameters:

	pix	pixel to be considered
	times	how many times should the pixel be considered

Returns:

true if successful or false otherwise.

virtual bool lti::colorModelEstimator::updateParameters

(

)

[virtual]

Set functor's parameters.

This member makes a copy of theParam: the functor will keep its own copy of the parameters!

This overloading will initialize the size of the internal color model

Returns:

true if successful, false otherwise

Reimplemented from lti::functor.

---

Member Data Documentation

int lti::colorModelEstimator::n [protected]

Number of pixels considered until now.

trgbPixel<double> lti::colorModelEstimator::sum [protected]

Sum of colors until now.

dmatrix lti::colorModelEstimator::sumOuter [protected]

Sum of outer-product of the colors until now (required for the computation of the covariance matrix.

thistogram<double> lti::colorModelEstimator::theModel [protected]

The color model used.

If empty, it is assumed to be uninitialized.

---

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

• ltiColorModelEstimator.h