contrib/mul/mipa/mipa_orientation_histogram.txx File Reference

Functions to compute histogram of orientations (HOGs) More...

#include "mipa_orientation_histogram.h"
#include <vcl_cmath.h>
#include <vnl/vnl_math.h>
#include <vcl_cassert.h>

Go to the source code of this file.

Defines
#define	vaip_orientation_histogram_txx_
#define	MIPA_ORIENTATION_HISTOGRAM_INSTANTIATE(srcT, sumT)
Functions
template<class srcT , class sumT >
void	mipa_orientation_histogram (const vil_image_view< srcT > &src, vil_image_view< sumT > &hog_image, unsigned n_angles, unsigned cell_size, bool full360, bool bilin_interp)
	Generate an image containing histograms of oriented gradients (HOG).

---

Detailed Description

Functions to compute histogram of orientations (HOGs)

Author:

Tim Cootes

Definition in file mipa_orientation_histogram.txx.

---

Define Documentation

#define MIPA_ORIENTATION_HISTOGRAM_INSTANTIATE	(		srcT,
			sumT
	)

Value:

template void mipa_orientation_histogram(const vil_image_view<srcT >& src, \
                                         vil_image_view<sumT >& hog_image, \
                                         unsigned n_angles, \
                                         unsigned cell_size, \
                                         bool full360, \
                                         bool bilin_intrep)

Definition at line 122 of file mipa_orientation_histogram.txx.

#define vaip_orientation_histogram_txx_

Definition at line 2 of file mipa_orientation_histogram.txx.

---

Function Documentation

template<class srcT , class sumT >

void mipa_orientation_histogram	(	const vil_image_view< srcT > &	src,
		vil_image_view< sumT > &	hog_image,
		unsigned	n_angles,
		unsigned	cell_size,
		bool	full360,
		bool	bilin_interp
	)

Generate an image containing histograms of oriented gradients (HOG).

At each pixel in src, compute angle and quantise into n_angles. If full360, then angle range is 0-360, else it is 0-180. hog_image is set to have n_angles planes. hog_image(i,j,k) gives the weighted sum of pixels with angle k in cell (i,j), corresponding to the i,j-th cell_size square block.

The corner of cell(0,0) is at src(1,1), to ignore border pixels.

Number of cells (size of hog_image) chosen so every cell entirely within src. Thus hog_image.ni()=(src.ni()-2)/cell_size.

Definition at line 25 of file mipa_orientation_histogram.txx.