sift::alg Namespace Reference

Functions
const vigra::MultiArray< 2, f32_t >	convolveWithGauss (const vigra::MultiArray< 2, f32_t > &img, f32_t sigma)
const vigra::MultiArray< 2, f32_t >	reduceToNextLevel (const vigra::MultiArray< 2, f32_t > &img, f32_t sigma)
const vigra::MultiArray< 2, f32_t >	increaseToNextLevel (const vigra::MultiArray< 2, f32_t > &img, f32_t sigma)
const vigra::MultiArray< 2, f32_t >	dog (const vigra::MultiArray< 2, f32_t > &lower, const vigra::MultiArray< 2, f32_t > &higher)
const vigra::Matrix< f32_t >	foDerivative (const std::array< vigra::MultiArray< 2, f32_t >, 3 > &img, const Point< u16_t, u16_t > &p)
const vigra::Matrix< f32_t >	soDerivative (const std::array< vigra::MultiArray< 2, f32_t >, 3 > &img, const Point< u16_t, u16_t > &p)
f32_t	gradientMagnitude (const vigra::MultiArray< 2, f32_t > &img, const Point< u16_t, u16_t > &p)
f32_t	gradientOrientation (const vigra::MultiArray< 2, f32_t > &img, const Point< u16_t, u16_t > &p)
const std::array< f32_t, 36 >	orientationHistogram36 (const vigra::MultiArray< 2, f32_t > &orientations, const vigra::MultiArray< 2, f32_t > &magnitudes, const vigra::MultiArray< 2, f32_t > &current_gauss)
const std::vector< f32_t >	orientationHistogram8 (const vigra::MultiArray< 2, f32_t > &orientations, const vigra::MultiArray< 2, f32_t > &magnitudes, const vigra::MultiArray< 2, f32_t > &current_gauss)
f32_t	vertexParabola (const Point< u16_t, f32_t > &ln, const Point< u16_t, f32_t > &peak, const Point< u16_t, f32_t > &rn)
std::array< Point< f32_t, f32_t >, 4 >	rotateShape (const Point< u16_t, u16_t > &center, f32_t angle, const u16_t width, const u16_t height)
void	normalizeVector (std::vector< f32_t > &vec)

Function Documentation

const vigra::MultiArray< 2, f32_t > sift::alg::convolveWithGauss	(	const vigra::MultiArray< 2, f32_t > &	,
		f32_t
	)

Convolves a given image with gaussian with a given sigma

Parameters

input	the input image which will be convolved
sigma	the standard deviation for the gaussian

Returns

blured image

const vigra::MultiArray< 2, f32_t > sift::alg::dog	(	const vigra::MultiArray< 2, f32_t > &	,
		const vigra::MultiArray< 2, f32_t > &
	)

Calculates the Difference of Gaussian, which is the differnce between 2 images which were convolved with gaussian under usage of a constant K

Parameters

lower	the image which lies lower in an octave
higher	the image which lies higher in an octave

Returns

the difference of gaussian image, which contains our interest points

const vigra::Matrix< f32_t > sift::alg::foDerivative	(	const std::array< vigra::MultiArray< 2, f32_t >, 3 > &	,
		const Point< u16_t, u16_t > &
	)

Calculates the first order derivative of the image, at the coordinates

Parameters

img	the image of which the first derivative is taken.
p	the point at which the derivative is taken

Returns

the derivative as a vector (dx, dy, ds)

f32_t sift::alg::gradientMagnitude	(	const vigra::MultiArray< 2, f32_t > &	,
		const Point< u16_t, u16_t > &
	)

Calculates the gradient magnitude of the given image at the given position

Parameters

img	the given img
p	the current point

Returns

the gradient magnitude value

f32_t sift::alg::gradientOrientation	(	const vigra::MultiArray< 2, f32_t > &	,
		const Point< u16_t, u16_t > &
	)

Calculates the gradient orientation of the given image at the given position

Parameters

img	the given img
p	the current point

Returns

the gradient orientation value

const vigra::MultiArray< 2, f32_t > sift::alg::increaseToNextLevel	(	const vigra::MultiArray< 2, f32_t > &	,
		f32_t
	)

Resamples an image by 2

Parameters

in	the input image

Returns

the output image

void sift::alg::normalizeVector

(

std::vector< f32_t > &

)

Normalizes a vector

Parameters

vec	the vector to be normalized

const std::array< f32_t, 36 > sift::alg::orientationHistogram36	(	const vigra::MultiArray< 2, f32_t > &	,
		const vigra::MultiArray< 2, f32_t > &	,
		const vigra::MultiArray< 2, f32_t > &
	)

Creates an orientation Histogram of a given img and his corresponding orientations and magnitudes. Places values in bins of size 10. So the resulting histogram has 36 elements.

Parameters

orientations	The img of which the histogram is taken from. Needs to be computed by gradient orientations before
magnitudes	The img of which the bins of the histogram will be weighted. Need to be precomputed by gradient magnitude
img	the given img

Returns

histogram with 36 bins which are weighted by magnitudes and gaussian

const std::vector< f32_t > sift::alg::orientationHistogram8	(	const vigra::MultiArray< 2, f32_t > &	,
		const vigra::MultiArray< 2, f32_t > &	,
		const vigra::MultiArray< 2, f32_t > &
	)

Creates an orientation Histogram of a given img and his corresponding orientations and magnitudes. Places values in bins of size 45. So the resulting histogram has 8 elements.

Parameters

orientations	The img of which the histogram is taken from. Needs to be computed by gradient orientations before
magnitudes	The img of which the bins of the histogram will be weighted. Need to be precomputed by gradient magnitude
img	the given img

Returns

histogram with 8 bins which are weighted by magnitudes and gaussian

const vigra::MultiArray< 2, f32_t > sift::alg::reduceToNextLevel	(	const vigra::MultiArray< 2, f32_t > &	,
		f32_t
	)

Resamples an image by 0.5

Parameters

img	the input image

Returns

the output image

std::array< Point< f32_t, f32_t >, 4 > sift::alg::rotateShape	(	const Point< u16_t, u16_t > &	,
		f32_t	,
		const u16_t	,
		const u16_t
	)

Rotates a given shape by a given degree clockwise

Parameters

center	the center point of the shape
angle	by which angle the shape should be rotated
width	the width of the shape
height	the height of the shape

Returns

array with 2 elements. First element represents upper left corner of the shape and bottom right is represented by the second argument

const vigra::Matrix< f32_t > sift::alg::soDerivative	(	const std::array< vigra::MultiArray< 2, f32_t >, 3 > &	,
		const Point< u16_t, u16_t > &
	)

Calculates the second order derivative of the image, at the coordinates

Parameters

img	the image of which the second derivative is taken.
p	the point at which the derivative is taken

Returns

the derivative as a matrix (dxx, dxy, dxs) (dyx, dyy, dys) (dsx, dsy, dss)

f32_t sift::alg::vertexParabola	(	const Point< u16_t, f32_t > &	,
		const Point< u16_t, f32_t > &	,
		const Point< u16_t, f32_t > &
	)

Calculates the vertex of a parabola, by taking a max value and its 2 neigbours

Parameters

ln	the left neighbor of the peak
peak	the peak value
rn	the right neighbor of the peak

Returns

the vertex value