Modified Moravec Interest Operator based image area matching. More...

#include <TePDIMMIOMatching.hpp>

Inheritance diagram for TePDIMMIOMatching:

Classes
class	CalcCCorrelationMtxParams
class	ExtractLocalMaximasParams
struct	GenerateCorrelationFeaturesParams
class	LoadImageParams
struct	MaximasListNodeT
Public Types
enum	FeatMatchingMethod { NormCrossCorrMethod }
Public Member Functions
	TePDIMMIOMatching ()
	~TePDIMMIOMatching ()
bool	CheckParameters (const TePDIParameters &parameters) const
	Checks if the supplied parameters fits the requirements of each PDI algorithm implementation.
Protected Types
typedef TePDIMatrix< unsigned char >	MaskMatrixT
typedef std::list < MaximasListNodeT >	MaximasListT
typedef std::pair< unsigned int, unsigned int >	MaximasMapNodeT
typedef std::multimap< double, MaximasMapNodeT >	MaximasMapT
Protected Member Functions
bool	RunImplementation ()
	Runs the current algorithm implementation.
void	ResetState (const TePDIParameters &params)
	Reset the internal state to the initial state.
void	moravecRepeatability (const MaximasListT &inputList, const MaximasListT &constraintList, unsigned int constraintImageWidth, unsigned int constraintImageHeight, double moravecRepeatabilityMinTolerance, double moravecReSampleFactor, MaximasListT &outputList)
	Applies a Moravec repeatability constraint over an image maximas list.
bool	matchFeatures (FeatMatchingMethod matching_method, const TePDIMatrix< double > &img1featMtx, const MaximasListT &img1_maximas_list, const TePDIMatrix< double > &img2featMtx, const MaximasListT &img2_maximas_list, TeCoordPairVect &tiePointsVec, std::vector< double > &tiePointsWeights)
	Features matching and tie-points generation.
Static Protected Member Functions
static void	loadImage (void *paramsPtr)
	A job function load raster data into a simple matrix.
static void	createImgMatrix (TeDataType dataType, TeSharedPtr< TePDIMtxDoubleAdptInt > &matrixPtr)
static void	extractLocalMaximas (void *paramsPtr)
	Extract local interest maxima points using a Modified Moravec Interest Operator aproach.
static bool	extractLocalMaximas (TePDIMtxDoubleAdptInt const imgMatrixPtr, MaskMatrixT const maskMatrixPtr, const unsigned int &xStart, const unsigned int &yStart, const unsigned int &width, const unsigned int &height, TePDIMatrix< double > &bufferMatrix, const unsigned int &moravecWindowSize, const unsigned int &maximasNumber, MaximasListT &outMaximasList, TePDIPIManager &progress)
	Extract local interest maxima points from an sub-image sub-region using a Modified Moravec Interest Operator aproach.
static void	generateCorrelationFeatures (void *paramsPtr)
	Generate correlation features.
static void	calcCCorrelationMtx (void *paramsPtr)
	cross-correlation matrix calcule.
static void	matrix2Tiff (const TePDIMtxDoubleAdptInt &input_matrix, const std::string &out_file_name, const MaximasListT &maxima_points=MaximasListT())
	Save the output matrix to a geotiff file.
static void	matrix2Tiff (const TePDIMtxDoubleAdptInt &input_matrix, const std::string &out_file_name, const TeCoordPairVect &tiepoints, bool usePt1)
	Save the output matrix to a geotiff file.
static void	doublesMatrix2Tiff (const TePDIMatrix< double > &input_matrix, const std::string &out_file_name, const MaximasListT &maxima_points=MaximasListT())
	Save the output matrix to a geotiff file.
static void	features2Tiff (unsigned int corr_window_width, const MaximasListT &img_maxima_points, TePDIMatrix< double > &img_features_matrix, const std::string &filenameaddon)
	Save the generated features to tif files.
static bool	bicubicResampleMatrix (const TePDIMtxDoubleAdptInt &inputMatrix, unsigned int outLines, unsigned int outCols, bool progressEnabled, TeMutex &globalMutex, double outMinValue, double outMaxValue, TePDIMtxDoubleAdptInt &outputMatrix)
	Input matrix bicubic resample.
static bool	generateMaximasRaster (const TePDIMtxDoubleAdptInt &inputMatrix, const MaximasListT &maximaPoints, TeRaster &outRaster)
	Generate the output maximas image ( 3 RGB bands where the maximas points are marked with value R255G0B0).
Protected Attributes
const double	moravecDownSampleFactor_
const double	moravecUpSampleFactor_
const double	moravecRepeatabilityMinTolerance_
TeThreadJobsManager	jobsMan_
TeMutex	globalMutex_

Detailed Description

Modified Moravec Interest Operator based image area matching.

Author:: Emiliano F. Castejon <castejon@dpi.inpe.br>.

Note:: The required parameters are:

Parameters:

input_image1_ptr	(TePDITypes::TePDIRasterPtrType) - The input image 1.
input_channel1	( unsigned int ) - Band to process from input_image1.
input_image2_ptr	(TePDITypes::TePDIRasterPtrType) - The input image 2.
input_channel2	( unsigned int ) - Band to process from input_image2.
out_tie_points_ptr	( TeSharedPtr< TeCoordPairVect > ) - The output tie-points where TeCoordPair.pt1 are input_image1 matricial indexes and TeCoordPair.pt2 are input_image2 matricial indexes.

Note:: The Optional parameters are:

Parameters:

matching_method	(TePDIMMIOMatching::FeatMatchingMethod) Features matching method (default: TePDIMMIOMatching::NormCrossCorrMethod).
pixel_x_relation	(double) - The pixel resolution relation pixel_x_relation = img1_pixel_res_x / img2_pixel_res_x (default=1.0).
pixel_y_relation	(double) - The pixel resolution relation pixel_y_relation = img1_pixel_res_y / img2_pixel_res_y (default=1.0).
gt_params	( TeGTParams ) - The geometric transformation parameters to be used (if not supplied, the dafult TeGTParams will be used - affine transformation, max direct mapping error = 2 pixels, RANSACRemotion outliers remotion ).
out_gt_params_ptr	( TeGTParams::pointer ) - The output parameters of internally generated geometric transformation (only valid when geometrical filtering is enabled).
input_box1	( TeBox ) - Box (image matrix coords) to process from input_image1 ( the entire image will be used if no box was supplied ).
input_box2	( TeBox ) - Box (image matrix coords) to process from input_image2 ( the entire image will be used if no box was supplied ).
enable_multi_thread	(int) - If present (any value) a multi-threaded processing will be used. (default: multi-thread disabled).
enable_threaded_raster_read	(int) - If present (any value) a multi-threaded input rasters data reading will be performed if enable_multi_thread is also enabled; Some TeDecoders do not support multi-thread so use it with caution (default: multi-threaded raster read disabled).
max_tie_points	(unsigned int) - The maximum number of generated tie-points (default=1000).
skip_geom_filter	(int) - If present (any value) will disable the geometric filtering ( default: geometric filtering enabled ).
enable_upsampled_filter	(int) - If present (any value) will enable the upsampled image Moravec repeatability filter ( default: upsampled filtering disabled - only the downsampled image filter will be applied ).
disable_reap_filter	(int) - If present (any value) will disable the image Moravec repeatability filters (upsample and downsample) ( default: filtering enabled - filter will be applied to avoid creation of tie-points over noisy areas).
corr_window_width	(unsigned int) - The correlation window width used to correlate points between images (Must be an odd number, minimum 13, default: 21).
moravec_window_width	(unsigned int) - The Moravec window width used to locate canditate tie-points (Must be an odd number, minimum 11, default: 11 ).
out_tie_points_weights_ptr	( TeSharedPtr< std::vector< double > > ) - The output tie- points weights vector (normalized between 0 and 1).
mask_image1_ptr	(TePDITypes::TePDIRasterPtrType) - Mask image 1 pointer (tie-points falling inside mask image areas marked with zeroes will not be generated; this image must have the same number of lines/columns as input_image1).
mask_image2_ptr	(TePDITypes::TePDIRasterPtrType) - Mask image 2 pointer (tie-points falling inside mask image areas marked with zeroes will not be generated; this image must have the same number of lines/columns as input_image1).
disable_subimage_search	(int) - If present (any value) the sub-image search will not enabled (the sub-image search improves the tie-point distribution homogeneity - default:sub-image search enabled).
out_maximas_image1_ptr	(TePDITypes::TePDIRasterPtrType) - This parameters is for visualization and test purposes and if this parameter is present and the pointer is valid (pointing to a valid raster instance) the output maximas image (frim input image 1) will be generated an copied to the pointed raster (this raster will be initiated to 3 RGB bands where the maximas points are marked with value R255G0B0).
out_maximas_image2_ptr	- The same as the out_maximas_image1_ptr parameter but for the second input image.

Examples:: TePDIMMIOMatching_test.cpp.

Member Typedef Documentation

typedef TePDIMatrix< unsigned char > TePDIMMIOMatching::MaskMatrixT [protected]

Mask matrix type.

typedef std::list< MaximasListNodeT > TePDIMMIOMatching::MaximasListT [protected]

Maximas list type

typedef std::pair< unsigned int, unsigned int > TePDIMMIOMatching::MaximasMapNodeT [protected]

Ordered maximas map node type

Parameters:

unsigned	int (first) Column - X.
unsigned	int (second) Line - Y.

typedef std::multimap< double, MaximasMapNodeT > TePDIMMIOMatching::MaximasMapT [protected]

Ordered maximas map type.

Member Enumeration Documentation

enum TePDIMMIOMatching::FeatMatchingMethod

Features matching methods.

Enumerator:

NormCrossCorrMethod

Features normalized cross-correlation method

Constructor & Destructor Documentation

TePDIMMIOMatching::TePDIMMIOMatching ( )

TePDIMMIOMatching::~TePDIMMIOMatching ( )

Member Function Documentation

bool TePDIMMIOMatching::bicubicResampleMatrix	(	const TePDIMtxDoubleAdptInt &	inputMatrix,
		unsigned int	outLines,
		unsigned int	outCols,
		bool	progressEnabled,
		TeMutex &	globalMutex,
		double	outMinValue,
		double	outMaxValue,
		TePDIMtxDoubleAdptInt &	outputMatrix
	)		`[static, protected]`

Input matrix bicubic resample.

Parameters:

inputMatrix	Input matrix.
outLines	Output number of lines.
outCols	Output number of columns.
progressEnabled	Progress enabled flag.
globalMutex	Global mutex.
outMinValue	The allowed output min value.
outMaxValue	The allowed output max value.
outputMatrix	Output matrix.

Returns:: true if OK, false on errors.

void TePDIMMIOMatching::calcCCorrelationMtx ( void * paramsPtr ) [static, protected]

cross-correlation matrix calcule.

Parameters:

paramsPtr The parameters needed for the thread execution (a pointer to a CalcCCorrelationMtxParams struct).

bool TePDIMMIOMatching::CheckParameters ( const TePDIParameters & parameters ) const [virtual]

Checks if the supplied parameters fits the requirements of each PDI algorithm implementation.

Note:: Error log messages must be generated. No exceptions generated.

Parameters:

parameters The parameters to be checked.

Returns:: true if the parameters are OK. false if not.

Implements TePDIAlgorithm.

void TePDIMMIOMatching::createImgMatrix	(	TeDataType	dataType,
		TeSharedPtr< TePDIMtxDoubleAdptInt > &	matrixPtr
	)		`[static, protected]`

Instantiate an image matrix capable of store pixel data of the supplied type.

Parameters:

dataType	Pixel type.
matrixPtr	A pointer for the new created matrix.

void TePDIMMIOMatching::doublesMatrix2Tiff	(	const TePDIMatrix< double > &	input_matrix,
		const std::string &	out_file_name,
		const MaximasListT &	maxima_points = `MaximasListT()`
	)		`[static, protected]`

Save the output matrix to a geotiff file.

Parameters:

input_matrix	Input matrix.
out_file_name	Output file name.
maxima_points	The maxima points to be filled with 255

bool TePDIMMIOMatching::extractLocalMaximas	(	TePDIMtxDoubleAdptInt const *	imgMatrixPtr,
		MaskMatrixT const *	maskMatrixPtr,
		const unsigned int &	xStart,
		const unsigned int &	yStart,
		const unsigned int &	width,
		const unsigned int &	height,
		TePDIMatrix< double > &	bufferMatrix,
		const unsigned int &	moravecWindowSize,
		const unsigned int &	maximasNumber,
		MaximasListT &	outMaximasList,
		TePDIPIManager &	progress
	)		`[static, protected]`

Extract local interest maxima points from an sub-image sub-region using a Modified Moravec Interest Operator aproach.

Parameters:

imgMatrixPtr	Input image matrix pointer.
maskMatrixPtr	Input mask matrix pointer (or null if mask is not used).
xStart	Image sub-region initial X coordinate.
yStart	Image sub-region initial y coordinate.
width	Image sub-region width.
height	Image sub-region height.
bufferMatrix	Pre-initiated buffer with at least width x height.
moravecWindowSize	Moravec window size
maximasNumber	The number of maximas to append to outMaximasMap.
outMaximasList	The found maximas will be appended to this container.
progres	Progress interface manager reference.
return	true if OK, false on errors.

void TePDIMMIOMatching::extractLocalMaximas ( void * paramsPtr ) [static, protected]

Extract local interest maxima points using a Modified Moravec Interest Operator aproach.

Parameters:

paramsPtr The parameters needed for the thread execution (a pointer to a ExtractLocalMaximasParams struct).

void TePDIMMIOMatching::features2Tiff	(	unsigned int	corr_window_width,
		const MaximasListT &	img_maxima_points,
		TePDIMatrix< double > &	img_features_matrix,
		const std::string &	filenameaddon
	)		`[static, protected]`

Save the generated features to tif files.

Parameters:

corr_window_width	The correlation windows width.
img_maxima_points	The image maxima points (these are the correlation matrix centers).
img_features_matrix	The output image features matrix ( each line is a stacked version of one rotated correlation window ). A string to be appended to the file name of each feature file.

Returns:: true if OK, false on errors.

void TePDIMMIOMatching::generateCorrelationFeatures ( void * paramsPtr ) [static, protected]

Generate correlation features.

Parameters:

paramsPtr The parameters needed for the thread execution (a pointer to a ExtractLocalMaximasParams struct).

bool TePDIMMIOMatching::generateMaximasRaster	(	const TePDIMtxDoubleAdptInt &	inputMatrix,
		const MaximasListT &	maximaPoints,
		TeRaster &	outRaster
	)		`[static, protected]`

Generate the output maximas image ( 3 RGB bands where the maximas points are marked with value R255G0B0).

Parameters:

inputMatrix	Input image data matrix.
maximaPoints	Input maximas point list.
outRaster	The ouptput raster.

Returns:: true if ok, false on errors.

void TePDIMMIOMatching::loadImage ( void * paramsPtr ) [static, protected]

A job function load raster data into a simple matrix.

Parameters:

paramsPtr The parameters needed for the thread execution (a pointer to a LoadImageParams struct).

bool TePDIMMIOMatching::matchFeatures	(	FeatMatchingMethod	matching_method,
		const TePDIMatrix< double > &	img1featMtx,
		const MaximasListT &	img1_maximas_list,
		const TePDIMatrix< double > &	img2featMtx,
		const MaximasListT &	img2_maximas_list,
		TeCoordPairVect &	tiePointsVec,
		std::vector< double > &	tiePointsWeights
	)		`[protected]`

Features matching and tie-points generation.

Parameters:

matching_method	Features matching method.
img1featMtx	Image 1 features matrix.
img1_maximas_list	Image 1 maximas list.
img2featMtx	Image 2 features matrix.
img2_maximas_list	Image 2 maximas list.
tiePointsVec	Output tie-points vector.
tiePointsWeights	Output tie-points weights vector ( normalized between DBL_MIN and 1 ).

Returns:: true if OK, false on errors.

void TePDIMMIOMatching::matrix2Tiff	(	const TePDIMtxDoubleAdptInt &	input_matrix,
		const std::string &	out_file_name,
		const TeCoordPairVect &	tiepoints,
		bool	usePt1
	)		`[static, protected]`

Save the output matrix to a geotiff file.

Parameters:

input_matrix	Input matrix.
out_file_name	Output file name.
tiepoints	The tie oints to be filled with 255
usePt1	Use pt1 from tiepoints, otherwise use pt1

void TePDIMMIOMatching::matrix2Tiff	(	const TePDIMtxDoubleAdptInt &	input_matrix,
		const std::string &	out_file_name,
		const MaximasListT &	maxima_points = `MaximasListT()`
	)		`[static, protected]`

Save the output matrix to a geotiff file.

Parameters:

input_matrix	Input matrix.
out_file_name	Output file name.
maxima_points	The maxima points to be filled with 255

void TePDIMMIOMatching::moravecRepeatability	(	const MaximasListT &	inputList,
		const MaximasListT &	constraintList,
		unsigned int	constraintImageWidth,
		unsigned int	constraintImageHeight,
		double	moravecRepeatabilityMinTolerance,
		double	moravecReSampleFactor,
		MaximasListT &	outputList
	)		`[protected]`

Applies a Moravec repeatability constraint over an image maximas list.

Parameters:

inputList	Original image maximas list.
constraintList	Constraint list.
constraintImageWidth	Constraint image width.
constraintImageHeight	Constraint image height.
moravecRepeatabilityMinTolerance	Error tolerance (pixels) constraint image reference.
moravecReSampleFactor	Moravec resample factor (factor = inputList_scal / constraintList_scale).
outputList	Output maximas list from input list.

void TePDIMMIOMatching::ResetState ( const TePDIParameters & params ) [protected, virtual]

Reset the internal state to the initial state.

Parameters:

params The new parameters referente at initial state.

Implements TePDIAlgorithm.

bool TePDIMMIOMatching::RunImplementation ( ) [protected, virtual]

Runs the current algorithm implementation.

Returns:: true if OK. false on error.

Implements TePDIAlgorithm.

Member Data Documentation

TeMutex TePDIMMIOMatching::globalMutex_ [protected]

A global syncronization mutex.

TeThreadJobsManager TePDIMMIOMatching::jobsMan_ [protected]

Internal threaded jobs manager.

const double TePDIMMIOMatching::moravecDownSampleFactor_ [protected]

Moravec downsample factor = original_image / downsampled_image

const double TePDIMMIOMatching::moravecRepeatabilityMinTolerance_ [protected]

Moravec repeatability tolerance

const double TePDIMMIOMatching::moravecUpSampleFactor_ [protected]

Moravec upsample factor = original_image / upsampled_image

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

E:/Projetos_Primeiro_Semestre_2012/TerraView/terralib/src/terralib/image_processing/TePDIMMIOMatching.hpp
E:/Projetos_Primeiro_Semestre_2012/TerraView/terralib/src/terralib/image_processing/TePDIMMIOMatching.cpp

Classes

Public Types

Public Member Functions

Protected Types

Protected Member Functions

Static Protected Member Functions

Protected Attributes

Detailed Description

Member Typedef Documentation

Member Enumeration Documentation

Constructor & Destructor Documentation

Member Function Documentation

Member Data Documentation