NMPTUtils Namespace Reference [Auxilliary Tools]

Auxilliary Tool: A set of functions that are commonly used by many NMPT classes and executables. More...

Functions
void	printMat (const cv::Mat &mat)
	Display a matrix to stdout in a multi line format easy to read.
void	printMat (const CvMat &mat)
	Display a matrix to stdout in a multi line format easy to read.
double	randomNormal ()
	Get a single normally distributed random number.
double	randomFloat ()
	Get a single random number distributed uniformly from 0-1.
std::string	commaSeparatedFlattenedMat (const cv::Mat &mat)
	Format a matrix in a single line, comma separated format suitable for output to a CSV file.
int	getVideoCaptureFromCommandLineArgs (cv::VideoCapture &capture, const int argc, const char **argv)
	Allows the source for an NMPT program to be specified from the command line.
double	rectAreaOverlapRatio (cv::Rect r1, cv::Rect r2)
	Computes the ratio of the area of the intersection of two rectangles to the area of the union.
double	rectAreaIntersect (cv::Rect r1, cv::Rect r2)
	Computes the area of the intersection of two rectangles.
double	notfinite (double a)
	Returns true if a is NaN or infinty.
void	distNorm (cv::Mat &normDist, const cv::Mat &unnormDist)
	Divides unnormDist by its L1-norm, effectively turning it into a probability distribution's histogram. The values of unnormDist should be all positive. If any values are NaN or inf, the probability mass is split uniformly between those values and all other values become 0.
void	sample (cv::Mat &samples, const cv::Mat &histogram, cv::Size sampleSize=cv::Size(1, 1))
	Generate a matrix of samples from a probability histogram.
int	sample (const cv::Mat &histogram)
	Generate a single sample from a probability histogram.
template<class mattype >
void	map (cv::Mat &mat, mattype(*function)(mattype))
	Apply a function (in place) to every element of a matrix. The function must take an argument of the matrix's data type, and return a value of the matrix's data type.
template<class mattype >
int	contains (const cv::Mat &mat, mattype val)
	Check to see if a matrix contains a certain value.
double	nchoosek (int n, int k)
	Quickly computer the n choose k function. The result is accurate through at least the range of 32-bit integers.
int	any (CvMat *mat)
	Checks to see if any member of mat is non-zero.
template<class mattype >
int	any (const cv::Mat &mat)
	Checks to see if any member of mat is non-zero.
void	fix (CvMat *mat)
	Round the elements of mat towards zero.
void	writeMatBinary (cv::FileStorage &storage, const std::string &name, const cv::Mat &mat)
	Write the value of a matrix to a FileStorage in a Base64 format that is much more compact than OpenCV's default format. The data is keyed by the supplied name.
void	readMatBinary (const cv::FileNode &storage, cv::Mat &mat)
	Read the value of a matrix to a FileStorage in a Base64 format that is much more compact than OpenCV's default format. The data is keyed by the supplied name. If you attempt to read a matrix that was not saved with "writeMatBinary," this will probably crash your program.
void	binaryToAscii (std::string &dest, const uchar *data, size_t bytes)
	Read the value of a matrix to a FileStorage in a Base64 format that is much more compact than OpenCV's default format. The data is keyed by the supplied name. The data is uncompressed with zlib after writing. If you attempt to read a matrix that was not saved with "writeMatBinaryCompressed," this will probably crash your program.
void	asciiToBinary (const std::string &data, uchar *dest, size_t bytes)
	Converts a Base64 string to binary data.
void	binaryToHex (std::string &dest, const uchar *data, size_t bytes)
	Converts binary data to a non-standard hex format that can be written to a text file, e.g. the YAML format written by OpenCV. This format requires 1.5 the storage space of Base64.
void	hexToBinary (const std::string &data, uchar *dest, size_t bytes)
	Converts a non-standard hex format to binary data. This format requires 1.5 the storage space of Base64.
void	splitString (std::vector< std::string > &dest, const std::string &data, size_t maxlen=512)
	Splits a string into a collection of string vectors, each with maximum length maxlen. This is useful because OpenCV imposes a maximum string length when writing to a file, somewhere between 2048 and 4096.
void	joinString (const std::vector< std::string > &data, std::string &dest)
	Rejoins a vector of strings into a single string.
void	rectangleRotated (cv::Mat &image, cv::Point center, cv::Size size, double angle, const cv::Scalar &color, int thickness=1, int lineType=8, int shift=0)
	Uses zlib to compress data. The output dest is allocated in this function and dest_size is set to the number of bytes of the compressed data. Returns a zlib error code if compression fails.
void	unIntegrate (const cv::Mat &src, cv::Mat &dest, int type=CV_8U)
	Calculate derivative image.
void	unSqIntegrate (const cv::Mat &src, cv::Mat &dest, int type=CV_8U)
	Calculate sqrt of derivative image.
cv::Mat	RBF (const cv::Mat &input, const cv::Mat &labels, const cv::Mat &weights, const cv::Mat &xqueries, double tau=.05, double eps=.001)
	Compute RBF values at supplied query points, given input data and labels.

---

Detailed Description

Auxilliary Tool: A set of functions that are commonly used by many NMPT classes and executables.

Author:

Nicholas Butko

Date:

2010

Version:

0.4

---

Function Documentation

void NMPTUtils::binaryToAscii	(	std::string &	dest,
		const uchar *	data,
		size_t	bytes
	)

Read the value of a matrix to a FileStorage in a Base64 format that is much more compact than OpenCV's default format. The data is keyed by the supplied name. The data is uncompressed with zlib after writing. If you attempt to read a matrix that was not saved with "writeMatBinaryCompressed," this will probably crash your program.

Converts binary data to a Base64 string that can be written to a text file, e.g. the YAML format written by OpenCV.

int NMPTUtils::getVideoCaptureFromCommandLineArgs	(	cv::VideoCapture &	capture,
		const int	argc,
		const char **	argv
	)

Allows the source for an NMPT program to be specified from the command line.

• If the program is invoked with no arguments, the default camera is used.
• If the program is invoked with any options (e.g. --help), a usage message is printed.
• If the program is invoked with an integer argument between 0 and 9, the camera with the corresponding number is used.
• If the program is invoked with a single string argument, it is interpererted as a path to a video file, which will be used for the source.

Returns 0 if there was an error, 1 if a video file was the source, or 2 if a camera was the source,

cv::Mat NMPTUtils::RBF	(	const cv::Mat &	input,
		const cv::Mat &	labels,
		const cv::Mat &	weights,
		const cv::Mat &	xqueries,
		double	tau = .05,
		double	eps = .001
	)

Compute RBF values at supplied query points, given input data and labels.

Parameters:

input	NxM, N data points, M Dims
labels	NxO, N data points, O outputs
weights	Nx1, 1 weight per data point
xqueries	QxM, Q query points, M Dims
tau,:	variance of gaussian weighting window
eps,:	minimal attention paid to all points

Returns:

QxO RBF values, Q query points, O output dimensions.

void NMPTUtils::rectangleRotated	(	cv::Mat &	image,
		cv::Point	center,
		cv::Size	size,
		double	angle,
		const cv::Scalar &	color,
		int	thickness = 1,
		int	lineType = 8,
		int	shift = 0
	)

Uses zlib to compress data. The output dest is allocated in this function and dest_size is set to the number of bytes of the compressed data. Returns a zlib error code if compression fails.

Uncompress data that was previously compressed. The output dest is allocated in this function, to the amount specified by dest_size. This should be the size of the original uncompressed data, which presumably you know, because you compressed it. Returns a zlib error code if decompression fails. Draw a rotated rectangle on an image. The point is to be like opencv's cv::rectangle function, but to allow rotation.