featureClusterFilters.H File Reference

#include "Util/Assert.H"
#include "Image/Image.H"
#include <vector>

Include dependency graph for featureClusterFilters.H:

This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Functions
template<class T_or_RGB >
T_or_RGB	filterAtLocation (Image< T_or_RGB > *image, Image< float > *kernel, Point2D< int > *location)
	this method will apply a filter at a location and return a value
template<class T_or_RGB >
void	filterAtLocationBatch (const Image< T_or_RGB > *image, const Image< float > *kernel, const std::vector< Point2D< int > * > *location, std::vector< T_or_RGB > *results, const unsigned int size)
	this method will apply a filter at a location and return a value
template<class T_or_RGB >
void	multiScaleBatchFilter (const Image< T_or_RGB > *image, const std::vector< Point2D< int > * > *location, const unsigned int *locations, const std::vector< Image< float > > *filters, const unsigned int scales, const bool quarterSize, const bool normalize, std::vector< std::vector< T_or_RGB > > *results)
	This method applies filter on an image at multiple scales at select locals.
template<class T >
void	multiScaleJunctionFilter (const unsigned int size, const unsigned int scales, const std::vector< std::vector< PixH2SV2< T > > > *input, std::vector< std::vector< PixH2SV2< T > > > *results)
	this will find junctions after calling multiScaleBatchFilter
template<class T >
void	multiScaleJunctionFilter (const unsigned int size, const unsigned int scales, const std::vector< std::vector< T > > *input, std::vector< std::vector< T > > *results)
	call this optionally after multiScaleBatchFilter

---

Detailed Description

Filters for featureClusterVision

Definition in file featureClusterFilters.H.

---

Function Documentation

template<class T_or_RGB >

T_or_RGB filterAtLocation	(	Image< T_or_RGB > *	image,
		Image< float > *	kernel,
		Point2D< int > *	location
	)			[inline]

this method will apply a filter at a location and return a value

This method is useful if your need to filter an image in a non-linear method or only need to extract filtered patches. Other than that, you should use the other methods above.

Definition at line 137 of file featureClusterFilters.H.

References abs(), ASSERT, Image< T >::beginw(), Image< T >::getHeight(), Image< T >::getWidth(), Point2D< T >::i, and sum().

template<class T_or_RGB >

void filterAtLocationBatch	(	const Image< T_or_RGB > *	image,
		const Image< float > *	kernel,
		const std::vector< Point2D< int > * > *	location,
		std::vector< T_or_RGB > *	results,
		const unsigned int	size
	)			[inline]

this method will apply a filter at a location and return a value

This method is useful if your need to filter an image in a non-linear method or only need to extract filtered patches. Other than that, you should use the other methods above. This method differs from filterAtLocation in that it accepts a batch of locations.

Definition at line 203 of file featureClusterFilters.H.

References abs(), ASSERT, Image< T >::begin(), Image< T >::getHeight(), Image< T >::getWidth(), and sum().

template<class T_or_RGB >

void multiScaleBatchFilter	(	const Image< T_or_RGB > *	image,
		const std::vector< Point2D< int > * > *	location,
		const unsigned int *	locations,
		const std::vector< Image< float > > *	filters,
		const unsigned int	scales,
		const bool	quarterSize,
		const bool	normalize,
		std::vector< std::vector< T_or_RGB > > *	results
	)			[inline]

This method applies filter on an image at multiple scales at select locals.

When called this method will run each of the supplied filters on the image at the scales ranging from full scale to size/2^n. To keep the workings straight, supply a name for each filter and then you can be sure how the results allign in the output vector. The method for each filter is basic convolution, but it limited to each location rather than the whole image.

Parameters:

	image	this is the input raw image
	location	This is a list of the locations at which to apply filters
	locations	This is how many locations are in the list
	filterNames	a list of what each filter is called for labels
	filters	these are several kernels for filter use
	scales	how man scales down do you want to filter each image
	quarterSize	scale image by sqrt(.5)(quarter-ish) or 1/2
	normalize	normalize over the filter outputs
	results	what you get back from all this
	resultsLables	User supplied descriptors for each filter

Definition at line 305 of file featureClusterFilters.H.

References abs(), Image< T >::begin(), Image< T >::beginw(), decXY(), Image< T >::getHeight(), Image< T >::getWidth(), sqrt(), and sum().

Referenced by featureClusterVision< FLOAT >::fCVrunStandAloneMSBatchFilter(), and featureClusterVision< FLOAT >::fCVrunStandAloneMSBatchTest().

template<class T >

void multiScaleJunctionFilter	(	const unsigned int	size,
		const unsigned int	scales,
		const std::vector< std::vector< T > > *	input,
		std::vector< std::vector< T > > *	results
	)			[inline]

call this optionally after multiScaleBatchFilter

Definition at line 568 of file featureClusterFilters.H.

template<class T >

void multiScaleJunctionFilter	(	const unsigned int	size,
		const unsigned int	scales,
		const std::vector< std::vector< PixH2SV2< T > > > *	input,
		std::vector< std::vector< PixH2SV2< T > > > *	results
	)			[inline]

this will find junctions after calling multiScaleBatchFilter

This is optional and may be called to find junctions if multiScaleBatchFilter was called using gabor filters at 0,45,90 and 135 degrees. Call this after calling multiScaleBatchFilter if wanted.

Definition at line 468 of file featureClusterFilters.H.

References abs().

Referenced by featureClusterVision< FLOAT >::fCVrunStandAloneMSBatchFilter(), and featureClusterVision< FLOAT >::fCVrunStandAloneMSBatchTest().