DistanceTransform Class Reference

an abstract distance transform class More...

#include <DistanceTransform.h>

Inheritance diagram for DistanceTransform:

List of all members.

Public Member Functions
	DistanceTransform (const int xSize, const int ySize, const bool unload=true)
	construct a distance transform object for an image of size (xSize,ySize).
virtual void	doTransform (const unsigned char *const I)=0
	calculate the distance transform
double	getD (const int x, const int y) const
	return the distance value assigned to a particular position
virtual bool	getP (const int x, const int y, int &px, int &py) const
	return the "parent" (border point) associated with this point
Static Public Attributes
static const int	IntInfinity = INT_MAX / 2
static const double	FloatInfinity = FLT_MAX / 2.0
Protected Member Functions
int	sub (const int x, const int y) const
template<class T>
void	initImmediate (const unsigned char *const I, T *d, const T halfDx=0, const T halfDy=0)
	init elements of the immediate interior (i.e., border points).
template<class T>
void	check (T *d, const int center, const int X, const int Y, const T Delta)
virtual void	borderCheck (const unsigned char *const I)
void	cleanUp ()
void	finish (const unsigned char *const I, double *d)
void	finish (const unsigned char *const I, int *d, const int dx, const int dy)
Protected Attributes
int	xSize
	integer size (width) in pixels
int	ySize
	integer size (height) in pixels
double *	dD
	result of distance transform (as doubles)
int *	iD
	the calculated distance transform (as ints)
Private Attributes
int *	rowOffsets
	used to speed up 2d array indexing
bool	unloadFlag
	convert ints to doubles (if necessary)
Classes
class	P
	DistanceTransform::P class. Simply a 2D point. More...

---

Detailed Description

an abstract distance transform class

---

Constructor & Destructor Documentation

DistanceTransform::DistanceTransform (  const int  xSize, const int  ySize, const bool  unload = true )

construct a distance transform object for an image of size (xSize,ySize). if 'unload' is true, transforms that use ratios of integers should convert their integer results to the actual double values (of the ratios).

---

Member Function Documentation

void DistanceTransform::finish (  const unsigned char *const   I, int *  d, const int  dx, const int  dy )  [protected]

called to finish up after distance transform employing ints

void DistanceTransform::finish (  const unsigned char *const   I, double *  d )  [protected]

called to finish up after distance transform employing doubles

virtual bool DistanceTransform::getP (  const int  x, const int  y, int &  px, int &  py )  const [virtual]

return the "parent" (border point) associated with this point Parameters: x  is the x location y  is the y location px  is the parent's x location to be returned py  is the parent's y location to be returned Returns: true if the parent is know; otherwise false. Reimplemented in CSED, DeadReckoning, DijkstraVectors, EightSED, errorfreeCSED, ModifiedDijkstra, Simple, and SimpleList.

int DistanceTransform::sub (  const int  x, const int  y )  const [protected]

given a point (x,y) in an image, this method returns the corresponding 1d subscript into the 1d image array

---

Member Data Documentation

const double DistanceTransform::FloatInfinity = FLT_MAX / 2.0 [static]

a double representing infinity (but one such that smaller values may still be added to it)

const int DistanceTransform::IntInfinity = INT_MAX / 2 [static]

an integer representing infinity (but one such that smaller values may still be added to it)

---

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

• DistanceTransform.h
• DistanceTransform.cpp

---