CINNICstatsRun Class Reference

Collaboration diagram for CINNICstatsRun:

Public Member Functions
	CINNICstatsRun ()
	default constructor
	~CINNICstatsRun ()
	default destructor
void	setConfig (readConfig &config, readConfig &config2)
	sets the config and loads files
void	setStuff (readConfig &fileList)
	sets a bunch of parameters and junk
void	setStuff ()
	single file option
void	runStandardStats (readConfig &fileList)
	run standard stats battery on images with contour traces
void	runPointAndFlood (readConfig &fileList, const char *param)
	run salience point analysis on image data
void	randomMatch (float likelyhood, long posRegionCount, long *totalCount)
	Run through the input image. Find likelyhood of a random match.
void	preProcess ()
	method for comp image pre-processing
void	checkSize ()
	checks what to do about images that are of different size
PixRGB< float >	colorTable (int i)
	this is a table of colors to be used for circles
void	pointAndFloodImage (Image< float > test_image, Image< float > sal_map, int points, char *filename, float floodv, float floodt)
	pointAndFlood on one image, no stats
void	pointAndFloodImage (Image< float > test_image, Image< float > sal_map, char *filename)
	pointAndFlood on one image, no stats
void	pointAndFlood (const char *filename, int i, bool standalone)
	this is the core to point and flood
float	polatSagi2AFC (Image< float > targetImage, Image< float > notargetImage)
	this will take in an image and make a 2AFC, see Itti Thesis 145-149
float	getMu1 ()
	Return mu1.
float	getMu2 ()
	Return mu2.
Protected Attributes
std::string	statsFile
	the output file
int	ii
	used for something
int	jj
int	decSize
	variables that used to be defined, but are now in stats.conf
int	unEvenDec
int	lowThresh
int	highThresh
int	chopVal
int	imageCount
	This is the number of images to be compaired.
int	zeroCount
int	qualityImage
	these are used for high quality comparison images
int	qChop
int	doResize
	this is true if the image is to be resized
int	pointsNum
	the number of salience points to find
int	circRad
	width of circles to draw out in sal map
int	rankThresh
	threshold at which to consider a point
int	preProcessPNF
	this tells if point and flood should pre-process image #
int	useDrawDisk
	this is set if you want to use draw disk rather then flood. Use w caution
int	diskSize
	the size of the disk to be drawn (r)
int	monoColor
	use this to produce thick mono color circle plots in figure
int	centerOffset
	offset for center of image element in output image
int	edgeAtten
	how much to attenuate borders for edge effects
float	totalErrorRatio
	the total error ratio, mean Chamfer34 value
float	meanChamfer
float	totalMeanError
	the mean error ratio, other stats
float	totalStdError
float	maxMean
float	minMean
float	totEucDist
	The total euclidian distance.
float	meanEucDist
	the mean euclidian distance, other stats
float	stdEucDist
float	maxEuc
float	minEuc
float	totReg
	total regression
float	meanReg
	mean regression, other stats
float	stdReg
float	maxReg
float	minReg
float	floodThresh
	flood variabels for image flooding
float	floodVal
float	foundTotal
	stats values for point and flood
float	foundMean
float	foundSTD
float	rankTotal
float	rankMean
float	rankSTD
float	PSError
	values for Polat and Sagi
float	mu1
	mu1 and mu2 as the population mean
float	mu2
Image< float >	testImage
	Image to test.
Image< float >	compImage
	Image to compare against.
Image< float >	salMap
	sal map for PointAndFlood
Image< byte >	tempImage1
	Temp image 1.
Image< byte >	tempImage2
	Temp image 2.
Image< PixRGB< byte > >	tempImage1c
	Temp image 1c.
Image< PixRGB< float > >	tempImage1cf
	temp image 1cf
Image< PixRGB< float > >	outImageTemplate
	output image of points
Image< PixRGB< float > >	outImageSource
	output image of points
Image< float >	maskImage
	track the mask image for stats
std::vector< int >	errorCount
	the count of error pixels in each image
std::vector< int >	totCount
	the count of total positive pixels in each image
std::vector< float >	totalVal
	The total pixel value (/n = average pixel value).
std::vector< float >	errorRatio
	The ratio of error pixels to non-error pixels.
std::vector< float >	testImageVector
	place holder for vectored images
std::vector< float >	compImageVector
	place holder for vectored images
std::vector< float >	testMean
	More stats.
std::vector< float >	testStd
	More stats.
std::vector< float >	compMean
	More stats.
std::vector< float >	compStd
	More stats.
std::vector< float >	regression
	Regression.
std::vector< double >	eucDist
	Eucliduan Distance between images.
stats< float >	Stats
	General stats object.
stats< double >	dStats
	Another stats object.
std::vector< Point2D< int > >	pointVec
	holds max n points from image
std::vector< std::vector < Point2D< int > > >	maxPointList
	holds max points for all images
std::vector< float >	postChamferVal
	holds max val for all chamfer34 processed images
std::vector< float >	pointsFound
	holds the count of correct salient points
std::vector< int >	pointsFoundNum
	how many images found this number of salient points
std::vector< float >	pointRank
	holds the rank of the most salient point on the object
std::vector< int >	pointRankNum
	how many images are at this rank
Point2D< int > *	point2D
PixRGB< float >	pixRGB
	a lovelly pixRGB instance
std::vector< long >	candPixels
	std::vector for number of candidate pixels for point and flood
std::vector< long >	realPixels
	std::vector for the number of real pixels on the contour
std::vector< double >	strikeP
	Vector to hold p for striking a pixel.
std::vector< double >	bernoulliP
	Bernoulli distribution P for the nth strike.

Detailed Description

Definition at line 51 of file CINNICstatsRun.H.

Constructor & Destructor Documentation

CINNICstatsRun::CINNICstatsRun ( )

default constructor

Definition at line 54 of file CINNICstatsRun.C.

CINNICstatsRun::~CINNICstatsRun ( )

default destructor

Definition at line 58 of file CINNICstatsRun.C.

Member Function Documentation

void CINNICstatsRun::checkSize ( )

checks what to do about images that are of different size

Definition at line 452 of file CINNICstatsRun.C.

References ASSERT, doResize, Image< T >::getHeight(), Image< T >::getWidth(), rescale(), tempImage1c, and tempImage2.

Referenced by runPointAndFlood(), and runStandardStats().

PixRGB< float > CINNICstatsRun::colorTable ( int i )

this is a table of colors to be used for circles

Definition at line 476 of file CINNICstatsRun.C.

References PixRGB< T >::setBlue(), PixRGB< T >::setGreen(), and PixRGB< T >::setRed().

Referenced by pointAndFlood().

float CINNICstatsRun::getMu1 ( )

Return mu1.

Definition at line 645 of file CINNICstatsRun.C.

References mu1.

Referenced by main().

float CINNICstatsRun::getMu2 ( )

Return mu2.

Definition at line 650 of file CINNICstatsRun.C.

Referenced by main().

void CINNICstatsRun::pointAndFlood	(	const char *	filename,
		int	i,
		bool	standalone
	)

this is the core to point and flood

Definition at line 541 of file CINNICstatsRun.C.

References candPixels, circRad, colorTable(), compImage, diskSize, drawArrow(), drawCircle(), drawDisk(), findMax(), flood(), floodThresh, foundTotal, Image< T >::getVal(), Point2D< T >::i, ii, maskImage, monoColor, outImageSource, outImageTemplate, pixRGB, pointRank, pointRankNum, pointsFound, pointsFoundNum, pointsNum, randomMatch(), rankThresh, realPixels, salMap, Image< T >::setVal(), sformat(), testImage, useDrawDisk, and Raster::WriteRGB().

Referenced by pointAndFloodImage(), and runPointAndFlood().

void CINNICstatsRun::pointAndFloodImage	(	Image< float >	test_image,
		Image< float >	sal_map,
		char *	filename
	)

pointAndFlood on one image, no stats

Definition at line 527 of file CINNICstatsRun.C.

References Image< T >::getHeight(), Image< T >::getWidth(), maskImage, outImageSource, outImageTemplate, pointAndFlood(), Image< T >::resize(), salMap, and testImage.

void CINNICstatsRun::pointAndFloodImage	(	Image< float >	test_image,
		Image< float >	sal_map,
		int	points,
		char *	filename,
		float	floodv,
		float	floodt
	)

pointAndFlood on one image, no stats

Definition at line 513 of file CINNICstatsRun.C.

References circRad, floodThresh, monoColor, pointsNum, and useDrawDisk.

Referenced by main().

float CINNICstatsRun::polatSagi2AFC	(	Image< float >	targetImage,
		Image< float >	notargetImage
	)

this will take in an image and make a 2AFC, see Itti Thesis 145-149

This takes in an image and performs a simulation of Polat and Sagi 1994 two alternative forced choice. It takes in an image and defines an error in judgement based upon the 2AFC paradigm. It returns the error. This should be compaired to the staircase method described by Polat and Sagi in obtaining threshold.

Definition at line 631 of file CINNICstatsRun.C.

References ASSERT, centerOffset, stats< T >::getErrorGGC_2AFC(), Image< T >::getHeight(), Image< T >::getVal(), Image< T >::getWidth(), mu1, PSError, sqrt(), and Stats.

Referenced by main().

void CINNICstatsRun::preProcess ( )

method for comp image pre-processing

Definition at line 407 of file CINNICstatsRun.C.

References compImage, Image< T >::getHeight(), Image< T >::getVal(), Image< T >::getWidth(), lowPass5(), qualityImage, rescale(), Image< T >::setVal(), and testImage.

Referenced by runPointAndFlood(), and runStandardStats().

void CINNICstatsRun::randomMatch	(	float *	likelyhood,
		long *	posRegionCount,
		long *	totalCount
	)

Run through the input image. Find likelyhood of a random match.

This method will take the compare image and find what is the p that a random pixel will land on a pixel that is a contour candidate pixel. Takes in pointers that return p, the number of candidate pixels and the total number of pixels

Parameters:

	likelyhood	This is the p that a random pixel will fall on a candidate contour
	posRegionCount	This is the number of candidate pixels in the image
	totalCount	This is the total number of pixels in the image

Definition at line 381 of file CINNICstatsRun.C.

References compImage, Image< T >::getHeight(), Image< T >::getVal(), Image< T >::getWidth(), maskImage, and rankThresh.

Referenced by pointAndFlood().

void CINNICstatsRun::runPointAndFlood	(	readConfig &	fileList,
		const char *	param
	)

run salience point analysis on image data

This method will attempt to find the most salient points in an input image, then it will compair thos points to points that are expected. It does this by first finding a point, then finding its distance to a contour using champher, then the point are is flooded to prevent it from being found again. This is repeated n times

Parameters:

fileList

This is the list of input images

Definition at line 292 of file CINNICstatsRun.C.

References bernoulliP, candPixels, checkSize(), compImage, edgeAtten, stats< T >::findS(), foundTotal, Image< T >::getHeight(), Image< T >::getWidth(), inplaceAttenuateBorders(), maskImage, outImageSource, outImageTemplate, pointAndFlood(), pointRank, pointRankNum, pointsFound, pointsFoundNum, pointsNum, preProcess(), preProcessPNF, RASFMT_PNM, readConfig::readFileTrue(), readConfig::readFileValueC(), readConfig::readFileValueNameC(), Raster::ReadGray(), Raster::ReadRGB(), realPixels, Image< T >::resize(), salMap, Stats, statsFile, strikeP, tempImage1c, tempImage1cf, tempImage2, and testImage.

Referenced by main().

void CINNICstatsRun::runStandardStats ( readConfig & fileList )

run standard stats battery on images with contour traces

this method finds how similar an image is to another image based upon its corralation, euclidian distance and mean error. It takes in two image files and does the compairison. It can run a whole batch of images (what its made to do) by setting up the file names in a config file

Parameters:

fileList

This is the parsed list of files

point2D object

Definition at line 146 of file CINNICstatsRun.C.

References Image< T >::begin(), checkSize(), compImage, compImageVector, compMean, compStd, dStats, edgeAtten, Image< T >::end(), errorCount, errorRatio, eucDist, stats< T >::findS(), Image< T >::getHeight(), Image< T >::getVal(), Image< T >::getWidth(), imageCount, inplaceAttenuateBorders(), readConfig::itemCount(), stats< T >::mean(), meanEucDist, meanReg, preProcess(), RASFMT_PNM, readConfig::readFileTrue(), readConfig::readFileValueC(), readConfig::readFileValueNameC(), Raster::ReadGray(), Raster::ReadRGB(), regression, stats< T >::rRegression(), Image< T >::setVal(), sqrt(), Stats, tempImage1c, tempImage1cf, tempImage2, testImage, testImageVector, testMean, testStd, totalErrorRatio, totalMeanError, totCount, totEucDist, and totReg.

Referenced by main().

void CINNICstatsRun::setConfig	(	readConfig &	config,
		readConfig &	config2
	)

sets the config and loads files

Read in the configuration file values

Parameters:

config

the parsed config file

Definition at line 64 of file CINNICstatsRun.C.

References centerOffset, circRad, decSize, diskSize, doResize, edgeAtten, floodThresh, readConfig::getItemValueF(), readConfig::getItemValueS(), monoColor, pointsNum, preProcessPNF, PSError, qualityImage, rankThresh, statsFile, and useDrawDisk.

Referenced by main().

void CINNICstatsRun::setStuff ( )

single file option

Definition at line 129 of file CINNICstatsRun.C.

References foundTotal, pointRankNum, pointsFoundNum, pointsNum, pointVec, Image< T >::resize(), salMap, tempImage1c, tempImage2, totalErrorRatio, totEucDist, and totReg.

void CINNICstatsRun::setStuff ( readConfig & fileList )

sets a bunch of parameters and junk

called automatically by constructor, but needs to be called if one of the methods such as runStandardStats or runPointAndFlood is called more then once.

Definition at line 94 of file CINNICstatsRun.C.

References bernoulliP, candPixels, compMean, compStd, errorCount, errorRatio, eucDist, foundTotal, readConfig::itemCount(), maxPointList, pointRank, pointRankNum, pointsFound, pointsFoundNum, pointsNum, pointVec, postChamferVal, realPixels, regression, Image< T >::resize(), salMap, strikeP, tempImage1c, tempImage2, testMean, testStd, totalErrorRatio, totalVal, totCount, totEucDist, and totReg.