Woolz Image Processing  Version 1.7.5
AlgFourier

## Files

file  AlgFourier.c
Fast Fourier and Hartley transform functions.

## Functions

void AlgFourHart1D (double *data, int num, int step)
Computes the Hartley transform of the given one dimensional data, and does it in place. More...

AlgError AlgFourHart2D (double **data, int useBuf, int numX, int numY)
Computes the Hartley transform of the given two dimensional data, and does it in place. More...

void AlgFour1D (double *real, double *imag, int num, int step)
Computes the Fourier transform of the given one dimensional complex data, and does it in place. The transformed values data are scaled by a factor of $$\sqrt{n}$$. More...

void AlgFourInv1D (double *real, double *imag, int num, int step)
Computes the inverse Fourier transform of the given complex one dimensional data, and does it in place. The transformed values data are scaled by a factor of $$\sqrt{n}$$. More...

void AlgFourReal1D (double *real, int num, int step)
Computes the Fourier transform of the given one dimensional real data, and does it in place. Data are returned in the array of size $$N$$ as: The data are returned in the array of size N with the layout as shown in the table (with 2M = N, r = real, i = imaginary): |-------—| | r0 $$| | r1$$ | | .. | | r(M-1) | | rM | | i1 | | i2 | | ... | | i(M - 1) | where the real and imaginary components are indexed as in the arrays computed with AlgFour1D(). The transformed values data are scaled by a factor of $$\sqrt{n}$$. More...

void AlgFourRealInv1D (double *real, int num, int step)
Computes the inverse Fourier transform of the given one one dimensional real data, and does it in place. The data should layed out in the array as returned by AlgFourReal1D(). The transformed values data are scaled by a factor of $$\sqrt{n}$$. More...

AlgError AlgFour2D (double **real, double **imag, int useBuf, int numX, int numY)
Computes the Fourier transform of the given two dimensional complex data, and does it in place. The transformed values data are scaled by a factor of $$\sqrt{n_x} \sqrt{n_y}$$. More...

AlgError AlgFourInv2D (double **real, double **imag, int useBuf, int numX, int numY)
Computes the inverse Fourier transform of the given two dimensional complex data, and does it in place. The transformed values data are scaled by a factor of $$\sqrt{n_x} \sqrt{n_y}$$. More...

AlgError AlgFourReal2D (double **real, int useBuf, int numX, int numY)
Computes the Fourier transform of the given two dimensional real data, and does it in place. More...

AlgError AlgFourRealInv2D (double **real, int useBuf, int numX, int numY)
Computes the Fourier transform of the given two dimensional data which resulted from a transform using AlgFourReal2D(), and does it in place. The transformed values data are scaled by a factor of $$\sqrt{n_x} \sqrt{n_y}$$. More...

AlgError AlgFour3D (double ***real, double ***imag, int useBuf, int numX, int numY, int numZ)
Computes the Fourier transform of the given three dimensional complex data, and does it in place. The transformed values data are scaled by a factor of $$\sqrt{n_x} \sqrt{n_y} \sqrt{n_z}$$. More...

AlgError AlgFourInv3D (double ***real, double ***imag, int useBuf, int numX, int numY, int numZ)
Computes the inverse Fourier transform of the given three dimensional complex data, and does it in place. The transformed values data are scaled by a factor of $$\sqrt{n_x} \sqrt{n_y} \sqrt{n_z}$$. More...

AlgError AlgFourReal3D (double ***real, int useBuf, int numX, int numY, int numZ)
Computes the Fourier transform of the given three dimensional real data, and does it in place. The transformed values data are scaled by a factor of $$\sqrt{n_x} \sqrt{n_y} \sqrt{n_z}$$. More...

AlgError AlgFourRealInv3D (double ***real, int useBuf, int numX, int numY, int numZ)
Computes the Fourier transform of the given three dimensional data which resulted from a transform using AlgFourReal3D(), and does it in place. The transformed values data are scaled by a factor of $$\sqrt{n_x} \sqrt{n_y} \sqrt{n_z}$$. More...

## Function Documentation

 void AlgFourHart1D ( double * data, int num, int step )

Computes the Hartley transform of the given one dimensional data, and does it in place.

Returns
void
Parameters
 data Given data. num Number of data. step Offset in data elements between the data to be transformed.

References ALG_DBG, ALG_DBG_LVL_1, ALG_DBG_LVL_FN, and ALG_M_SQRT2.

Referenced by AlgFour1D(), AlgFourHart2D(), AlgFourInv1D(), AlgFourReal1D(), and AlgFourRealInv1D().

 AlgError AlgFourHart2D ( double ** data, int useBuf, int numX, int numY )

Computes the Hartley transform of the given two dimensional data, and does it in place.

Returns
Error code, may be set if buffers can not be allocated.
Parameters
 data Given data. useBuf Allocate private buffers to make columns contiguous. numX Number of data in each row. numY Number of data in each column.
 void AlgFour1D ( double * real, double * imag, int num, int step )

Computes the Fourier transform of the given one dimensional complex data, and does it in place. The transformed values data are scaled by a factor of $$\sqrt{n}$$.

Returns
void
Parameters
 real Given real data. imag Given imaginary data. num Number of data. step Offset in data elements between the data to be transformed.

References ALG_DBG, ALG_DBG_LVL_1, ALG_DBG_LVL_FN, and AlgFourHart1D().

Referenced by AlgFourRealInv3D().

 void AlgFourInv1D ( double * real, double * imag, int num, int step )

Computes the inverse Fourier transform of the given complex one dimensional data, and does it in place. The transformed values data are scaled by a factor of $$\sqrt{n}$$.

Returns
void
Parameters
 real Given real data. imag Given imaginary data. num Number of data. step Offset in data elements between the data to be transformed.

References ALG_DBG, ALG_DBG_LVL_1, ALG_DBG_LVL_FN, and AlgFourHart1D().

Referenced by AlgFourRealInv3D().

 void AlgFourReal1D ( double * real, int num, int step )

Computes the Fourier transform of the given one dimensional real data, and does it in place. Data are returned in the array of size $$N$$ as: The data are returned in the array of size N with the layout as shown in the table (with 2M = N, r = real, i = imaginary): |-------—| | r0 $$| | r1$$ | | .. | | r(M-1) | | rM | | i1 | | i2 | | ... | | i(M - 1) | where the real and imaginary components are indexed as in the arrays computed with AlgFour1D(). The transformed values data are scaled by a factor of $$\sqrt{n}$$.

Returns
void
Parameters
 real Given real data. num Number of data (N). step Offset in data elements between the data to be transformed.

References ALG_DBG, ALG_DBG_LVL_1, ALG_DBG_LVL_FN, and AlgFourHart1D().

Referenced by AlgFourRealInv3D().

 void AlgFourRealInv1D ( double * real, int num, int step )

Computes the inverse Fourier transform of the given one one dimensional real data, and does it in place. The data should layed out in the array as returned by AlgFourReal1D(). The transformed values data are scaled by a factor of $$\sqrt{n}$$.

Returns
void
Parameters
 real Given real/complex data. num Number of data. step Offset in data elements between the data to be transformed.

References ALG_DBG, ALG_DBG_LVL_1, ALG_DBG_LVL_FN, and AlgFourHart1D().

Referenced by AlgFourRealInv3D().

 AlgError AlgFour2D ( double ** real, double ** imag, int useBuf, int numX, int numY )

Computes the Fourier transform of the given two dimensional complex data, and does it in place. The transformed values data are scaled by a factor of $$\sqrt{n_x} \sqrt{n_y}$$.

Returns
Error code, may be set if buffers can not be allocated.
Parameters
 real Given real data. imag Given imaginary data. useBuf Allocate private buffers to make columns contiguous. numX Number of data in each row. numY Number of data in each column.

Referenced by WlzFourierTransformObj().

 AlgError AlgFourInv2D ( double ** real, double ** imag, int useBuf, int numX, int numY )

Computes the inverse Fourier transform of the given two dimensional complex data, and does it in place. The transformed values data are scaled by a factor of $$\sqrt{n_x} \sqrt{n_y}$$.

Returns
Error code, may be set if buffers can not be allocated.
Parameters
 real Given real data. imag Given imaginary data. useBuf Allocate private buffers to make columns contiguous. numX Number of data in each row. numY Number of data in each column.

Referenced by WlzFourierTransformObj().

 AlgError AlgFourReal2D ( double ** real, int useBuf, int numX, int numY )

Computes the Fourier transform of the given two dimensional real data, and does it in place.

Returns
Error code, may be set if buffers can not be allocated.
    The layout of the array is similar to that in AlgFourReal1D()
and as shown in the table (with 2M = N, r = real,
i = imaginary):

-------|--------|---|------------|-------|--------|---|-----------
r00    |r01     |...|r0(M-1)     |r0M    |i01     |...|i0(M-1)
...    |...     |...|...         |...    |...     |...|...
r(M-1)0|r(M-1)1 |...|r(M-1)(M-1) |r(M-1)M|i(M-1)1 |...|i(M-1)(M-1)
rM0    |rM1     |...|rM(M-1)     |rMM    |iM1     |...|iM(M-1)
i10    |r(M+1)1 |...|r(M+1)(M-1) |i1M    |i(M+1)1 |...|i(M+1)(M-1)
...    |...     |...|...         |...    |...     |...|...
i(M-1)0|r(2M-1)1|...|r(2M-1)(M-1)|i(M-1)M|i(2M-1)1|...|i(2M-1)(M-1)

Using contiguous buffers has a large effect for
data larger than a CPU's fastest cache and little
cost for smaller data arrays.
The times below were measured for on a Lenovo T430s
with a i7-2640M with 4kB cache. Data sizes are the
number of double values. These times include buffer
allocation but not array allocation or file I/O
and are the mean of 5 runs using elapsed (wall
clock) time.
data sz | tm (no buf, 1 thr) | tm (buf, 1 thr) | tm (buf, 2 thr)
--------|--------------------|-----------------|-------------
256^2 |     3ms            |    3ms          |    2ms
512^2 |    12ms            |    6ms          |    4ms
1024^2 |   120ms            |   44ms          |   35ms
2048^2 |   600ms            |  170ms          |  110ms
4096^2 |  3300ms            |  764ms          |  460ms
8192^2 | 17000ms            | 3600ms          | 2100ms
The transformed values data are scaled by a factor
of \form#628.

Parameters
 real Given real data. useBuf Allocate private buffers to make columns contiguous. numX Number of data in each row. numY Number of data in each column.

Referenced by AlgAutoCorrelate2D(), AlgCrossCorrelate2D(), and WlzFourierTransformObj().

 AlgError AlgFourRealInv2D ( double ** real, int useBuf, int numX, int numY )

Computes the Fourier transform of the given two dimensional data which resulted from a transform using AlgFourReal2D(), and does it in place. The transformed values data are scaled by a factor of $$\sqrt{n_x} \sqrt{n_y}$$.

Returns
Error code, may be set if buffers can not be allocated.
Parameters
 real Given real/complex data. useBuf Allocate private buffers to make columns contiguous. numX Number of data in each row. numY Number of data in each column.

Referenced by AlgAutoCorrelate2D(), AlgCrossCorrelate2D(), and WlzFourierTransformObj().

 AlgError AlgFour3D ( double *** real, double *** imag, int useBuf, int numX, int numY, int numZ )

Computes the Fourier transform of the given three dimensional complex data, and does it in place. The transformed values data are scaled by a factor of $$\sqrt{n_x} \sqrt{n_y} \sqrt{n_z}$$.

Returns
Error code, may be set if buffers can not be allocated.
    Using contiguous buffers has a large effect for
data larger than a CPU's fastest cache and little
cost for smaller data arrays.

Parameters
 real Given real data. imag Given imaginary data. useBuf Allocate private buffers to make columns contiguous. numX Number of data in each row. numY Number of data in each column. numZ Number of data in each plane.

Referenced by WlzFourierTransformObj().

 AlgError AlgFourInv3D ( double *** real, double *** imag, int useBuf, int numX, int numY, int numZ )

Computes the inverse Fourier transform of the given three dimensional complex data, and does it in place. The transformed values data are scaled by a factor of $$\sqrt{n_x} \sqrt{n_y} \sqrt{n_z}$$.

Returns
Error code, may be set if buffers can not be allocated.
Parameters
 real Given real data. imag Given imaginary data. useBuf Allocate private buffers to make columns contiguous. numX Number of data in each row. numY Number of data in each column. numZ Number of data in each plane.

Referenced by WlzFourierTransformObj().

 AlgError AlgFourReal3D ( double *** real, int useBuf, int numX, int numY, int numZ )

Computes the Fourier transform of the given three dimensional real data, and does it in place. The transformed values data are scaled by a factor of $$\sqrt{n_x} \sqrt{n_y} \sqrt{n_z}$$.

Returns
Error code, may be set if buffers can not be allocated.
    Using contiguous buffers has a large effect for
data larger than a CPU's fastest cache and little

Parameters
 real Given real data. useBuf Allocate private buffers to make columns contiguous. numX Number of data in each row. numY Number of data in each column. numZ Number of data in each plane.

Referenced by WlzFourierTransformObj().

 AlgError AlgFourRealInv3D ( double *** real, int useBuf, int numX, int numY, int numZ )

Computes the Fourier transform of the given three dimensional data which resulted from a transform using AlgFourReal3D(), and does it in place. The transformed values data are scaled by a factor of $$\sqrt{n_x} \sqrt{n_y} \sqrt{n_z}$$.

Returns
Error code, may be set if buffers can not be allocated.
Parameters
 real Given real/complex data. useBuf Allocate private buffers to make columns contiguous. numX Number of data in each row. numY Number of data in each column. numZ Number of data in each plane.

Referenced by WlzFourierTransformObj().