Transforms.H File Reference

---

Detailed Description

Transformations on Image

Definition in file Transforms.H.

#include "Util/Promotions.H"

Include dependency graph for Transforms.H:

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

Go to the source code of this file.

Functions
template<class T>
Image< byte >	segmentObject (const Image< T > &src, const Point2D< int > &center)
	Segment object using an adaptive thresholding & region growing method.
template<class T>
int	segmentLandmark (const Image< T > &src, const Point2D< int > &seed, Image< byte > &target, double &activity, double &standout, float &area_percentage)
	Segment landmark using an adaptive thresholding & region growing method.
template<class T>
Image< byte >	contour2D (const Image< T > &src, const byte onval=255, const byte offval=0)
	Compute 2D 4-neighbor contours of binary objects in image.
template<class T>
Image< T >	chamfer34 (const Image< T > &src, const T dmax=255)
	3/4-chamfer distance transform; object = non 0
template<class T>
Image< T >	saliencyChamfer34 (const Image< T > &src, const T dmax=255)
	3/4-chamfer saliency distance transform; object = non 0
template<class T>
Image< typename promote_trait< T, float >::TP >	dct (const Image< T > &src, const int offx, const int offy, const int size)
	Computes 2D discrete cosine transform.
template<class T>
float	infoFFT (const Image< T > &src, const float eps, const Rectangle &rect)
	FFT-based information measure.
template<class T>
Image< float >	infoFFT (const Image< T > &src, const float eps)
	FFT-based information measure.
double	learningCoeff (const Image< float > &featureMap, const Image< byte > &targetDMap, const bool softmask=false, const int in_thresh=20, const int out_thresh=255)
	Examine local maxes in/out of targets and compute a learning weight.
template<class T_or_RGB>
Image< T_or_RGB >	scaleBlock (const Image< T_or_RGB > &src, const Dims newDims)
	scale up an image using blocks
template<class T>
int	floodClean (const Image< T > &src, Image< T > &dest, const Point2D< int > &seed, const T thresh, const T val, int numdirs=8)
	Floods object >= thresh with val from seed and returns area.
template<class T>
int	floodCleanBB (const Image< T > &src, Image< T > &dest, const Point2D< int > &seed, const T thresh, const T val, Rectangle &bbox)
	same as floodClean, only finds the bounding box of the region in addition
template<class T>
Image< byte >	segmentObjectClean (const Image< T > &src, const Point2D< int > &seed, int numdirs=8)
	same as segmentObject, only with floodClean and different parameters
template<class T>
Image< float >	distDegrade (const Image< T > &src, const Point2D< int > &foa, const float area)
	degrades a binary image mask s.t. it decays away from the FOA
template<class T>
Image< T >	highThresh (const Image< T > &src, const T thresh, const T val)
	set all values >= thresh to val in src and return the result
template<class T_or_RGB>
Image< T_or_RGB >	replaceVals (const Image< T_or_RGB > &src, const T_or_RGB from, const T_or_RGB to, const T_or_RGB other)
	replace all values "from" with "to" and all other values with "other"
template<class T_or_RGB>
Image< T_or_RGB >	replaceVals (const Image< T_or_RGB > &src, const T_or_RGB from, const T_or_RGB to)
	replace all values "from" with "to" and leave all other values in place
template<class T_or_RGB>
Image< T_or_RGB >	composite (const Image< T_or_RGB > &fg, const Image< T_or_RGB > &bg, const T_or_RGB transparent=T_or_RGB())
	Composite of the non-transparent pixels of fg on top of bg.
template<class T>
Image< byte >	makeBinary (const Image< T > &src, const T &threshold, const byte lowVal=byte(0), const byte highVal=byte(255))
	make a binary version of the source image
template<class T>
Image< byte >	makeBinary2 (const Image< T > &src, const T &lowThresh, const T &highThresh, const byte lowVal=byte(0), const byte highVal=byte(255))
	make a binary version of the source image
template<class T_or_RGB>
void	pasteImage (Image< T_or_RGB > &background, const Image< T_or_RGB > &foreground, const T_or_RGB &transparent, const Point2D< int > location, float opacity=1.0F)
	paste foreground into background image at (location)
template<class T>
void	inplacePasteGabor (Image< T > &dst, const Image< T > &gabor, const Point2D< int > &pos, const T background)
	Paste a gabor into an image area. Account for overlap.
template<class T>
int	flood (const Image< T > &src, Image< T > &dest, const Point2D< int > &seed, const T thresh, const T val)
	Floods object >= thresh with val from seed and returns area.
template<class T>
int	countParticles (const Image< T > &src, const T thresh)
	How many particles >= thresh ?
template<class T>
void	inplaceAddBGnoise (Image< T > &dst, const float range)
	Add background noise to image.
template<class T>
void	inplaceAddBGnoise2 (Image< T > &dst, const float range)
	Add background noise to image.

---

Function Documentation

template<class T> Image<T> chamfer34 (  const Image< T > &  src, const T  dmax = 255 )

3/4-chamfer distance transform; object = non 0 Definition at line 257 of file Transforms.C. References ASSERT, Image< T >::begin(), Image< T >::beginw(), Image< T >::getDims(), Image< T >::getHeight(), Image< T >::getSize(), Image< T >::getWidth(), Dims::h(), i, Image< T >::initialized(), rutz::min(), Dims::w(), and w. Referenced by VisualBufferStd::findMostInterestingTarget(), foveate(), highlightRegions(), VisualBufferStd::input(), TaskRelevanceMapKillN::integrate(), WinnerTakeAllGreedy::integrate(), main(), BeoSubCanny::runCanny(), and runCanny().

template<class T_or_RGB> Image<T_or_RGB> composite (  const Image< T_or_RGB > &  fg, const Image< T_or_RGB > &  bg, const T_or_RGB  transparent = T_or_RGB() )

Composite of the non-transparent pixels of fg on top of bg. Definition at line 1360 of file Transforms.C. References ASSERT, Image< T >::begin(), Image< T >::beginw(), Image< T >::end(), Image< T >::getDims(), Image< T >::isSameSize(), NO_INIT, and lobot::stop(). Referenced by SimulationViewerStd::drawMegaCombo(), SimulationViewerEyeMvt::getTraj(), and SimulationViewerStd::getTraj().

template<class T> Image<byte> contour2D (  const Image< T > &  src, const byte  onval = 255, const byte  offval = 0 )

Compute 2D 4-neighbor contours of binary objects in image. Points on the contour will be given value onval, and points not on the contour will have offval. Definition at line 210 of file Transforms.C. References ASSERT, Image< T >::begin(), Image< T >::beginw(), Image< T >::getDims(), Image< T >::getHeight(), Image< T >::getWidth(), h, i, Image< T >::initialized(), NO_INIT, w, and z. Referenced by SimulationViewerNerdCam::drawMaskOutline(), SimulationViewerStd::drawMaskOutline(), SimulationViewerCompress::getTraj(), and Image_xx_contour2D_xx_1().

template<class T> int countParticles (  const Image< T > &  src, const T  thresh )

How many particles >= thresh ? Definition at line 1562 of file Transforms.C. References fill(), flood(), Image< T >::getHeight(), Image< T >::getVal(), Image< T >::getWidth(), Point2D< T >::i, Point2D< T >::j, LFATAL, and p.

template<class T> Image<typename promote_trait<T, float>::TP> dct (  const Image< T > &  src, const int  offx, const int  offy, const int  size )

Computes 2D discrete cosine transform. of square image patch size*size at offset (offx, offy) Definition at line 527 of file Transforms.C. References ASSERT, cols, lobot::cos(), e, Image< T >::getHeight(), Image< T >::getVal(), Image< T >::getWidth(), i, Image< T >::initialized(), l, log(), n, N, NO_INIT, p, rows, s, Image< T >::setVal(), lobot::sin(), and x. Referenced by infoMeasure().

template<class T> Image<float> distDegrade (  const Image< T > &  src, const Point2D< int > &  foa, const float  area )

degrades a binary image mask s.t. it decays away from the FOA

template<class T> int flood (  const Image< T > &  src, Image< T > &  dest, const Point2D< int > &  seed, const T  thresh, const T  val )

Floods object >= thresh with val from seed and returns area.

template<class T> int floodClean (  const Image< T > &  src, Image< T > &  dest, const Point2D< int > &  seed, const T  thresh, const T  val, int  numdirs = 8 )

Floods object >= thresh with val from seed and returns area. a clean version of Image::flood - problems solved: correct boundary checking, allow for thresh = 0. Parameters: dest  if dest is not initialized, it will be initialized to ZEROS. ATTENTION: If you want to have the old behavior of dest being initialized to src, you have to this explicitely before calling floodClean

template<class T> int floodCleanBB (  const Image< T > &  src, Image< T > &  dest, const Point2D< int > &  seed, const T  thresh, const T  val, Rectangle &  bbox )

same as floodClean, only finds the bounding box of the region in addition

template<class T> Image<T> highThresh (  const Image< T > &  src, const T  thresh, const T  val )

set all values >= thresh to val in src and return the result Definition at line 1314 of file Transforms.C. References ASSERT, Image< T >::beginw(), Image< T >::endw(), Image< T >::initialized(), and lobot::stop(). Referenced by EntropyChannel::doInput(), and main().

template<class T> Image<float> infoFFT (  const Image< T > &  src, const float  eps )

FFT-based information measure. Definition at line 869 of file Transforms.C. References data, Image< T >::getDims(), Image< T >::getHeight(), Image< T >::getVal(), Image< T >::getWidth(), Dims::h(), i, NO_INIT, Image< T >::setVal(), lobot::sin(), and Dims::w().

template<class T> float infoFFT (  const Image< T > &  src, const float  eps, const Rectangle &  rect )

FFT-based information measure. Definition at line 858 of file Transforms.C. References crop(), getMinMax(), and infoFFT().

template<class T> void inplaceAddBGnoise (  Image< T > &  dst, const float  range )

Add background noise to image. Range should be max value for any image of the type to which we add noise Definition at line 1585 of file Transforms.C. References inplaceAddBGnoise2(). Referenced by RawVisualCortex::combineOutputs(), BeoSubSaliency::computeCMAP(), computeCMAP(), main(), RegSaliency::postChannel(), SingleChannel::postProcessMap(), and BeobotBrainMT::threadCompute().

template<class T> void inplaceAddBGnoise2 (  Image< T > &  dst, const float  range )

Add background noise to image. Range should be max value for any image of the type to which we add noise similar to addBGnoise, but eliminates BNG.. constant Definition at line 1595 of file Transforms.C. References ASSERT, Image< T >::beginw(), Image< T >::getHeight(), Image< T >::getWidth(), h, i, Image< T >::initialized(), rutz::min(), randomDouble(), and w. Referenced by inplaceAddBGnoise().

template<class T> void inplacePasteGabor (  Image< T > &  dst, const Image< T > &  gabor, const Point2D< int > &  pos, const T  background )

Paste a gabor into an image area. Account for overlap. this is a simple extention to paste, but it tests for gabor overlap and attempts rectify similar to solution for MakeSnake by taking the pixel value furthest from the background value Parameters: background  This is the color of the image background

double learningCoeff (  const Image< float > &  featureMap, const Image< byte > &  targetDMap, const bool  softmask = false, const int  in_thresh = 20, const int  out_thresh = 255 )

Examine local maxes in/out of targets and compute a learning weight. Valid ranges for in_thresh and out_thresh are 0..255, plus: use -1 for in_thresh to learn from the background only, or use -1 for out_thresh to learn from the target only.

template<class T> Image<byte> makeBinary (  const Image< T > &  src, const T &  threshold, const byte  lowVal = byte(0), const byte  highVal = byte(255) )

make a binary version of the source image Parameters: src  source image threshold  the threshold used for binarization lowVal  value for all points <= threshold highVal  value for all points > threshold Definition at line 1384 of file Transforms.C. References Image< T >::begin(), Image< T >::beginw(), Image< T >::endw(), Image< T >::getDims(), NO_INIT, and lobot::stop(). Referenced by VisualBufferStd::findMostInterestingTarget(), Image_xx_makeBinary_xx_1(), main(), and Raster_xx_img_gray_write_bw_read_gray_xx_1().

template<class T> Image<byte> makeBinary2 (  const Image< T > &  src, const T &  lowThresh, const T &  highThresh, const byte  lowVal = byte(0), const byte  highVal = byte(255) )

make a binary version of the source image Parameters: src  source image highThresh  the threshold used for binarization lowThresh  the threshold used for binarization lowVal  value for all points <= threshold highVal  value for all points > threshold Definition at line 1397 of file Transforms.C. References Image< T >::begin(), Image< T >::beginw(), Image< T >::endw(), Image< T >::getDims(), NO_INIT, and lobot::stop().

template<class T_or_RGB> void pasteImage (  Image< T_or_RGB > &  background, const Image< T_or_RGB > &  foreground, const T_or_RGB &  transparent, const Point2D< int >  location, float  opacity = 1.0F )

paste foreground into background image at (location) Parameters: transparent  pixels with this value are not replaced in background

template<class T_or_RGB> Image<T_or_RGB> replaceVals (  const Image< T_or_RGB > &  src, const T_or_RGB  from, const T_or_RGB  to )

replace all values "from" with "to" and leave all other values in place Definition at line 1345 of file Transforms.C. References Image< T >::begin(), Image< T >::beginw(), Image< T >::endw(), Image< T >::getDims(), and NO_INIT.

template<class T_or_RGB> Image<T_or_RGB> replaceVals (  const Image< T_or_RGB > &  src, const T_or_RGB  from, const T_or_RGB  to, const T_or_RGB  other )

replace all values "from" with "to" and all other values with "other" Definition at line 1330 of file Transforms.C. References Image< T >::begin(), Image< T >::beginw(), Image< T >::endw(), Image< T >::getDims(), and NO_INIT.

template<class T> Image<T> saliencyChamfer34 (  const Image< T > &  src, const T  dmax )

3/4-chamfer saliency distance transform; object = non 0 * reset borders */ Definition at line 375 of file Transforms.C. References ASSERT, Image< T >::beginw(), Image< T >::getDims(), Image< T >::getHeight(), Image< T >::getSize(), Image< T >::getWidth(), Dims::h(), h, i, Image< T >::initialized(), Dims::w(), w, x, and y. Referenced by V1::evolveCanny(), and V1::evolveSobel().

template<class T_or_RGB> Image<T_or_RGB> scaleBlock (  const Image< T_or_RGB > &  src, const Dims  newDims )

scale up an image using blocks

template<class T> int segmentLandmark (  const Image< T > &  src, const Point2D< int > &  seed, Image< byte > &  target, double &  activity, double &  standout, float &  area_percentage )

Segment landmark using an adaptive thresholding & region growing method. Returns an image that masks the landmark Parameters: activity  is a measure of mean activity in the landmark standout  is a measure of how much the landmark stands out from the rest of the image area  is the percentage area of landmark

template<class T> Image<byte> segmentObject (  const Image< T > &  src, const Point2D< int > &  center )

Segment object using an adaptive thresholding & region growing method. Returns an image that masks the object (see Itti et al, J Neuroimaging, 2001)

template<class T> Image<byte> segmentObjectClean (  const Image< T > &  src, const Point2D< int > &  seed, int  numdirs = 8 )

same as segmentObject, only with floodClean and different parameters

---