==============================================================================
FFTReal
Version 2.00, 2005/10/18
Fourier transformation (FFT, IFFT) library specialised for real data
Portable ISO C++
(c) Laurent de Soras <laurent.de.soras@club-internet.fr>
Object Pascal port (c) Frederic Vanmol <frederic@fruityloops.com>
==============================================================================
1. Legal
--------
This library is free software; you can redistribute it and/or
modify it under the terms of the GNU Lesser General Public
License as published by the Free Software Foundation; either
version 2.1 of the License, or (at your option) any later version.
This library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public
License along with this library; if not, write to the Free Software
Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
Check the file license.txt to get full information about the license.
2. Content
----------
FFTReal is a library to compute Discrete Fourier Transforms (DFT) with the
FFT algorithm (Fast Fourier Transform) on arrays of real numbers. It can
also compute the inverse transform.
You should find in this package a lot of files ; some of them are of interest:
- readme.txt: you are reading it
- FFTReal.h: FFT, length fixed at run-time
- FFTRealFixLen.h: FFT, length fixed at compile-time
- FFTReal.pas: Pascal implementation (working but not up-to-date)
- stopwatch directory
3. Using FFTReal
----------------
Important - if you were using older versions of FFTReal (up to 1.03), some
things have changed. FFTReal is now a template. Therefore use FFTReal<float>
or FFTReal<double> in your code depending on the application datatype. The
flt_t typedef has been removed.
You have two ways to use FFTReal. In the first way, the FFT has its length
fixed at run-time, when the object is instanciated. It means that you have
not to know the length when you write the code. This is the usual way of
proceeding.
3.1 FFTReal - Length fixed at run-time
--------------------------------------
Just instanciate one time a FFTReal object. Specify the data type you want
as template parameter (only floating point: float, double, long double or
custom type). The constructor precompute a lot of things, so it may be a bit
long. The parameter is the number of points used for the next FFTs. It must
be a power of 2:
#include "FFTReal.h"
...
long len = 1024;
...
FFTReal <float> fft_object (len); // 1024-point FFT object constructed.
Then you can use this object to compute as many FFTs and IFFTs as you want.
They will be computed very quickly because a lot of work has been done in the
object construction.
float x [1024];
float f [1024];
...
fft_object.do_fft (f, x); // x (real) --FFT---> f (complex)
...
fft_object.do_ifft (f, x); // f (complex) --IFFT--> x (real)
fft_object.rescale (x); // Post-scaling should be done after FFT+IFFT
...
x [] and f [] are floating point number arrays. x [] is the real number
sequence which we want to compute the FFT. f [] is the result, in the
"frequency" domain. f has the same number of elements as x [], but f []
elements are complex numbers. The routine uses some FFT properties to
optimize memory and to reduce calculations: the transformaton of a real
number sequence is a conjugate complex number sequence: F [k] = F [-k]*.
3.2 FFTRealFixLen - Length fixed at compile-time
------------------------------------------------
This class is significantly faster than the previous one, giving a speed
gain between 50 and 100 %. The template parameter is the base-2 logarithm of
the FFT length. The datatype is float; it can be changed by modifying the
DataType typedef in FFTRealFixLenParam.h. As FFTReal class, it supports
only floating-point types or equivalent.
To instanciate the object, just proceed as below:
#include "FFTRealFixLen.h"
...
FFTRealFixLen <10> fft_object; // 1024-point (2^10) FFT object constructed.
Use is similar as the one of FFTReal.
3.3 Data organisation
---------------------
Mathematically speaking, the formulas below show what does FFTReal:
do_fft() : f(k) = sum (p = 0, N-1, x(p) * exp (+j*2*pi*k*p/N))
do_ifft(): x(k) = sum (p = 0, N-1, f(p) * exp (-j*2*pi*k*p/N))
Where j is the square root of -1. The formulas differ only by the sign of
the exponential. When the sign is positive, the transform is called positive.
Common formulas for Fourier transform are negative for the direct tranform and
positive for the inverse one.
However in these formulas, f is an array of complex numbers and doesn't
correspound exactly to the f[] array taken as function parameter. The
following table shows how the f[] sequence is mapped onto the usable FFT
coefficients (called bins):
FFTReal output | Positive FFT equiv. | Negative FFT equiv.
---------------+-----------------------+-----------------------
f [0] | Real (bin 0) | Real (bin 0)
f [...] | Real (bin ...) | Real (bin ...)
f [length/2] | Real (bin length/2) | Real (bin length/2)
f [length/2+1] | Imag (bin 1) | -Imag (bin 1)
f [...] | Imag (bin ...) | -Imag (bin ...)
f [length-1] | Imag (bin length/2-1) | -Imag (bin length/2-1)
And FFT bins are distributed in f [] as above:
| | Positive FFT | Negative FFT
Bin | Real part | imaginary part | imaginary part
------------+----------------+-----------------+---------------
0 | f [0] | 0 | 0
1 | f [1] | f [length/2+1] | -f [length/2+1]
... | f [...], | f [...] | -f [...]
length/2-1 | f [length/2-1] | f [length-1] | -f [length-1]
length/2 | f [length/2] | 0 | 0
length/2+1 | f [length/2-1] | -f [length-1] | f [length-1]
... | f [...] | -f [...] | f [...]
length-1 | f [1] | -f [length/2+1] | f [length/2+1]
f [] coefficients have the same layout for FFT and IFFT functions. You may
notice that scaling must be done if you want to retrieve x after FFT and IFFT.
Actually, IFFT (FFT (x)) = x * length(x). This is a not a problem because
most of the applications don't care about absolute values. Thus, the operation
requires less calculation. If you want to use the FFT and IFFT to transform a
signal, you have to apply post- (or pre-) processing yourself. Multiplying
or dividing floating point numbers by a power of 2 doesn't generate extra
computation noise.
4. Compilation and testing
--------------------------
Drop the following files into your project or makefile:
Array.*
def.h
DynArray.*
FFTReal*.cpp
FFTReal*.h*
OscSinCos.*
Other files are for testing purpose only, do not include them if you just need
to use the library ; they are not needed to use FFTReal in your own programs.
FFTReal may be compiled in two versions: release and debug. Debug version
has checks that could slow down the code. Define NDEBUG to set the Release
mode. For example, the command line to compile the test bench on GCC would
look like:
Debug mode:
g++ -Wall -o fftreal_debug.exe *.cpp stopwatch/*.cpp
Release mode:
g++ -Wall -o fftreal_release.exe -DNDEBUG -O3 *.cpp stopwatch/*.cpp
It may be tricky to compile the test bench because the speed tests use the
stopwatch sub-library, which is not that cross-platform. If you encounter
any problem that you cannot easily fix whil
没有合适的资源?快使用搜索试试~ 我知道了~
Qt-5.15.5+vs2019+32位+windows版
共2000个文件
h:6255个
qml:1777个
png:1675个
需积分: 19 3 下载量 84 浏览量
2022-09-19
09:10:22
上传
评论
收藏 468.17MB ZIP 举报
温馨提示
Qt-5.15.5+vs2019+32位+webengine+https集成 可以正常使用,如果需要https,需要将openssl的两个dll复制到项目编译下.
资源详情
资源评论
资源推荐
收起资源包目录
Qt-5.15.5+vs2019+32位+windows版 (2000个子文件)
torrentclient.cpp 53KB
xform.cpp 41KB
pictureflow.cpp 40KB
volumetric.cpp 32KB
textedit.cpp 32KB
engine.cpp 27KB
mainwindow.cpp 27KB
main.cpp 26KB
colorswatch.cpp 25KB
pathstroke.cpp 25KB
weatherinfo.cpp 24KB
scriptdebugger.cpp 24KB
context2d.cpp 23KB
gradients.cpp 22KB
spreadsheet.cpp 22KB
mainwindow.cpp 22KB
peerwireclient.cpp 22KB
pathdeform.cpp 21KB
browserwindow.cpp 21KB
dialog.cpp 21KB
arthurstyle.cpp 20KB
composition.cpp 20KB
main.cpp 20KB
mainwindow.cpp 19KB
environment.cpp 19KB
window.cpp 19KB
regularexpressiondialog.cpp 19KB
glwidget.cpp 18KB
player.cpp 18KB
mainwindow.cpp 18KB
pieview.cpp 18KB
xmlconverter.cpp 18KB
widgetgallery.cpp 17KB
mainwindow.cpp 17KB
mainwindow.cpp 17KB
appmodel.cpp 17KB
mainwidget.cpp 16KB
glwindow.cpp 16KB
graphmodifier.cpp 16KB
main.cpp 16KB
classwizard.cpp 16KB
mainwindow.cpp 16KB
musicplayer.cpp 16KB
mainwidget.cpp 16KB
mainwindow.cpp 16KB
window_multithreaded.cpp 15KB
waveform.cpp 15KB
pingpong.cpp 15KB
themewidget.cpp 15KB
widget.cpp 15KB
filemanager.cpp 15KB
remoteselector.cpp 15KB
hoverpoints.cpp 15KB
variantdelegate.cpp 14KB
addressbook.cpp 14KB
mainwindow.cpp 14KB
d3d11squircle.cpp 14KB
padnavigator.cpp 14KB
camera.cpp 14KB
mainwindow.cpp 14KB
audiorecorder.cpp 14KB
xorblender.cpp 14KB
imageviewer.cpp 13KB
filetree.cpp 13KB
customitemgraph.cpp 13KB
mainwindow.cpp 13KB
main.cpp 13KB
addressbook.cpp 13KB
mainwindow.cpp 13KB
toolbar.cpp 13KB
mainwindow.cpp 13KB
window_singlethreaded.cpp 13KB
httpwindow.cpp 13KB
main.cpp 13KB
audioinput.cpp 12KB
cborconverter.cpp 12KB
arthurwidgets.cpp 12KB
document.cpp 12KB
mainwindow.cpp 12KB
qtlogo.cpp 12KB
device.cpp 12KB
licensewizard.cpp 12KB
browser.cpp 12KB
tabletcanvas.cpp 12KB
calculator.cpp 12KB
norwegianwoodstyle.cpp 12KB
main.cpp 12KB
mainwindow.cpp 12KB
d3d12renderer.cpp 12KB
audiodevices.cpp 12KB
histogramwidget.cpp 12KB
window.cpp 12KB
states.cpp 12KB
glwidget.cpp 11KB
hellowindow.cpp 11KB
mainwindow.cpp 11KB
devicehandler.cpp 11KB
window.cpp 11KB
main.cpp 11KB
encodingdialog.cpp 11KB
共 2000 条
- 1
- 2
- 3
- 4
- 5
- 6
- 20
三奇日月星
- 粉丝: 0
- 资源: 3
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功
评论0