求malab代码,对数据进行离散离散时间傅里叶变换换.

实验三数字图像的离散傅里叶变换_百度文库
两大类热门资源免费畅读
续费一年阅读会员,立省24元!
实验三数字图像的离散傅里叶变换
阅读已结束,下载文档到电脑
想免费下载本文?
定制HR最喜欢的简历
下载文档到电脑,方便使用
还剩7页未读,继续阅读
定制HR最喜欢的简历
你可能喜欢查看: 9042|回复: 5|关注: 0
一组离散数据FFT变换
<h1 style="color:# 麦片财富积分
新手, 积分 8, 距离下一级还需 42 积分
关注者: 5
最近使用一种磁阻传感器采集得到一组离散数据,这组离散数据用Excel表存储,现在打算用快速傅里叶变换对采集到的离散数据进行分析,并绘制幅值--频率特性曲线,但是这组离散数据没有现有的公式去表示,因此不知道如何入手,仅仅几个数据,如何确定采用点数?如何确定采样频率?自己写了一些代码,但是感觉不对,因此,我把代码上传,一方面,希望老师能够指出我不对的地方,另一方面,希望老师能够给出一些指导;谢谢!PS:Matlab版本是32位R2010b,操作系统是32为Win7系统%处理顺序:
%第一步:安全线信号时间序列&&
%第二步:加Hanning窗(此处Hanning的宽度需要自己调整,以求达到最佳的宽度)本程序中
设置窗的宽度大小为40
%第三步:FFT
%第四步:得出频率--幅值序列
%第五步:求最大幅值对应的特征频率
%关闭所有窗口并清理内存并清除命令窗口的命令
%读取数据
x=xlsread('5反上下2.xls','G2:G401');%读取原始数据存入x中
y=x';
%显示未滤波的离散数据的波形
figure('NumberTitle','off','Name','5元原始未滤波数据波形');
%显示原始数据波形
plot(x,'r-');%红色,实线
xlabel('Time1');%标记横坐标为:Time
ylabel('Amplitude');%标记纵坐标为:Amplitude
title('5元安全线原始时域波形');%绘图添加标题
%给当前图形添加网格
plot(y,'g-');%红色,实线
xlabel('Time');%标记横坐标为:Time
ylabel('Amplitude');%标记纵坐标为:Amplitude
plot(y,'g-');%显示平滑滤波后的波形
title('5元安全线信号滤波后时域波形');%绘图添加标题
%给当前图形添加网格
%手动生成汉宁窗
%加汉宁窗
N_H = 40;%汉宁窗的宽度设置为40
han = hanning(N_H);%汉宁窗生成函数
%显示手动生成的窗函数的波形
figure('NumberTitle','off','Name','窗宽为40的窗函数波形');
plot(han,'b-');
xlabel('窗宽');%标记横坐标
ylabel('窗幅值');%标记纵坐标
axis([1,N_H,0,1]);%设置坐标轴的最小和最大值
title('Hanning Window');%绘图设置标题
%给当前图形添加网格
%傅里叶分析
N = 512;%采样点数?怎么选取
fs = 1000;%采样频率为12.5kHz?怎么选取?有何标准
%短段波形与Hanning窗拟合,此处采用for循环来实现,仅仅是尝试
%步长40的参量
t = 0;%用于40为一步的跳转
for i = 1:400
& &if mod(i,40)==0
& && & t=t+1;
& && & y(1,40*(t-1)+1:i)=y(1,40*(t-1)+1:i).*han';%循环拟合,最后生成新的拟合向

& &end
end
%对信号进行FFT变换
y_fft = fft(y,N);
% mag = abs(conj(y_fft)*2/N);
Pyy2=y_fft.*conj(y_fft)/512;
f2=fs*(0:256)/512;
plot(f2,Pyy2(1:257),'k-');%黑色
xlabel('频率');%标记横坐标
ylabel('幅值');%标记纵坐标
title('5元单边频率--幅值波形');%绘图设置标题
%给当前图形添加网格
%将双边频谱转换为单边频谱
A=sort(Pyy2(1:257),'descend');
% %首先,求最大幅值
B=A(1,1);
% %求最大幅值处对应的频率,即最大特征频率
fre_max=find(Pyy2(1:257)==B);
% %频率转换
fre_max=round(fre_max*fs/N);复制代码
10:48 上传
点击文件名下载附件
68.5 KB, 下载次数: 6085
离散数据文件
<h1 style="color:# 麦片财富积分
顶一个 我也在处理相同的问题 但是毫无思路怎么处理 求大神帮助
<h1 style="color:# 麦片财富积分
楼主应该已经解决了问题了吧,我现在也遇到这样的问题,对一个采集的离散信号进行傅里叶变换求频率幅值能不能给发个合理的程序
<h1 style="color:# 麦片财富积分
楼主,请问找到对应方法了吗
<h1 style="color:# 麦片财富积分
楼主解决问题了吗
<h1 style="color:# 麦片财富积分
站长推荐 /2
基于模型设计的需求管理和验证
MATLAB中文论坛是全球最大的 MATLAB & Simulink 中文社区。用户免费注册会员后,即可下载代码,讨论问题,请教资深用户及结识书籍作者。立即注册加入我们吧!
MATLAB官方社交平台
MATLAB中文论坛微社区离散傅里叶变换 - CSDN博客
离散傅里叶变换
离散傅里叶变换(Discrete Fourier Transform,缩写为DFT),是在和上都呈离散的形式,将信号的时域采样变换为其的频域采样。在形式上,变换两端(时域和频域上)的序列是有限长的,而实际上这两组序列都应当被认为是的主&#20540;序列。即使对有限长的离散信号作DFT,也应当将其看作其周期延拓的变换。在实际应用中通常采用计算DFT。
下面给出离散傅里叶变换的变换对:
对于N点序列,它的离散傅里叶变换(DFT)为
其中e 是的,i
是。通常以符号表示这一变换,即
离散傅里叶变换的逆变换(IDFT)为:
可以记为:
实际上,DFT和IDFT变换式中和式前面的归一化系数并不重要。在上面的定义中,DFT和IDFT前的系数分别为1 和1/N。有时会将这两个系数都改成。
连续时间信号x(t) 以及对应的都是连续函数。由于数字系统只能处理有限长的,因此必须将x和都离散化,并且建立对应的傅里叶变换。
假设x(t)时限于[0, L],再通过时域采样将x(t)离散化,就可以得到有限长离散信号,记为xdiscrete(t)。设采样周期为T,则时域采样点数N=L/T。
它的傅里叶变换为
这就是x(t)在时域采样后的连续傅里叶变换,也就是,它在频域依然是连续的。
下面将频域信号转化为有限长离散信号。与对时域信号的处理类&#20284;,假设频域信号是带限的,再经过离散化,即可得到有限长离散信号。依据,时域采样若要能完全重建原信号,频域信号应当带限于(0,1/T)。由于时域信号时限于[0,
L],由采样定理以及时频对偶的关系,频域的采样间隔应为1/L。故,频域采样点数为:
即频域采样的点数和时域采样同为N,频域采样点为 在DTFT频域上采样:
令T=1,将其归一化,就得到前面定义的离散傅里叶变换。因此,DFT就是先将信号在时域离散化,求其连续傅里叶变换后,再在频域离散化的结果。
下面考察离散傅里叶变换与连续傅里叶变换(CT,Continuous Fourier Transform)的关系。连续傅里叶变换
的采样为:
将这个积分以黎曼和的形式近&#20284;,有:
这一结论指出离散傅里叶变换确实是连续傅里叶变换在某种意义上的近&#20284;。不过必须注意以下两点:
时域采样必须满足采样定理离散傅里叶变换的处理对象是时限的
为此,通常对连续信号的时域采样再做一次加窗处理(Windowing),这样就得到带限的有限长离散信号。
离散时间傅里叶变换(DTFT)是在时域上对的采样。DFT则是在频域上对DTFT的均匀采样。离散信号x[n](n=0,...,N-1)的DTFT为:
对在离散的频点上采样
即为x 的DFT。由于DTFT在频域是周期的,所以在DTFT频域上的均匀采样也应是周期的。实际上是这个周期序列的主&#20540;序列。
N 点序列的DFT只能在有限的N个频点上观察频谱,这相当于从的缝隙中观察景色,对于了解信号在整个频域上的特性是不够的。为了观察到其他频率的信息,需要对原信号x[n]做一些处理,以便在不同的频点上采样。
将原来在DTFT频域上的采样点数增加到M 点,这样采样点位置变为。则对应的DFT成为
若在序列x[n] 之后补上M-N个零,设为x'[n],则上式变为
因此将x[n]补零再做DFT就可以得到x[n]的DTFT在其他频率上的&#20540;,相当于移动了栅栏,因而能够从其他位置进行观察。
N 点DFT的频谱分辨率是2π / N。一节指出可以通过补零观察到更多的频点,但是这并不意味着补零能够提高真正的频谱分辨率。这是因为x[n]
实际上是x(t) 采样的主&#20540;序列,而将x[n]补零得到的x'[n] 周期延拓之后与原来的序列并不相同,也不是x(t) 的采样。因此与是不同离散信号的频谱。对于补零至M点的x'的DFT,只能说它的分辨率2π
/ M仅具有计算上的意义,并不是真正的、物理意义上的频谱。频谱分辨率的提高只能在满足采样定理的条件下增加时域采样长度来实现。
周期为N的离散信号构成一个N 维。在这一空间上两个信号x
和y 的定义为
下面给出上的一组:
将信号x 在这组正交基上分解,得
此即为离散傅里叶变换。又
此即为离散傅里叶变换的逆变换。
由此可知,在正交分解的角度上说,离散周期信号x的离散傅里叶变换实质上是x在正交基{ek}上的分量。而从的角度上说,{ek}是的,则是对应的。
根据,离散信号x与y的对偶于频域上x与y离散傅里叶变换的乘积。下面揭示DFT与圆周卷积的内在关系。
对长为N的离散信号与,如果要计算它们的卷积,就必须定义与在0≤n&N
之外的&#20540;。如果将与作周期为N的延拓,则有
如此,周期为N的圆周卷积:
卷积的结果仍然是以N为周期的离散信号。
前面指出,ek是DFT的特征矢量,实际上它也是圆周卷积的特征矢量。定义x与y的圆周卷积算子:
则ek与y的圆周卷积为
显然,y的DFT
就是圆周卷积的特征&#20540;。
离散傅里叶变换是可逆的
其中C表示。即,任意N-维复矢量都存在DFT和IDFT,而且其DFT和IDFT也是N-维复矢量。
矢量组exp(2πi kn/N)是N-维复数空间上的一组正交基:
其中 δkn是。
时域信号序列xn的相位移动exp(2πinm /
N)(其中m为整数)在频域反映为“循环移位”,即:频域信号序列Xk变为,其中下标是指将k-m
对N 。类&#20284;的,由对偶性,时域信号序列的循环移位对应于频域信号序列的相移:
若 则 且有
上文中一节已经证明,离散序列的傅里叶变换是周期的。有限长序列xn的离散傅里叶变换Xk可以被看作是它的周期延拓序列的离散时间傅里叶变换。由时频对偶性可知Xk也可以被看作它的周期延拓的主&#20540;。
上一节的移位定理隐含着DFT的周期性,因为DFT的幅度 | Xk | 不受输入信号的循环移位的影响,因为时移在频域对偶于相移,循环移位仅仅使DFT的相位产生平移。周期性的边界条件在DFT的许多应用中有重要作用。解时,就假设边界条件是满足周期性的,这是一个很有用的性质(参见)。
如果 Xk 和 Yk 分别是 xn 和
yn 的离散傅立叶变换,那么就有 :
其中星号表示复共扼。 是普朗歇尔定理的一个特例:
DFT在诸多多领域中有着重要应用,下面仅是颉取的几个例子。需要指出的是,所有DFT的实际应用都依赖于计算离散傅里叶变换及其逆变换的快速算法,即。
快速傅里叶变换(即FFT)是计算离散傅里叶变换及其逆变换的快速算法。按照DFT的定义计算一个长为n的序列的DFT需要的计算复杂度达到了,而同样长度FFT的计算复杂度仅为。由于DFT的逆变换可以由DFT表示,所以DFT逆变换的计算同样可以由FFT完成。FFT算法的提出,使DFT得到了广泛的实际应用。
前面指出,。因此可以对连续信号x(t)均匀采样并截断以得到有限长的离散序列,对这一序列作离散傅里叶变换,可以分析连续信号x(t)频谱的性质。前面还提到DFT应用于频谱分析需要注意的两个问题:即采样可能导致信号混叠和截断信号引起的频谱泄漏。可以通过选择适当的采样频率(见)消减。选择适当的序列长度并加窗可以抑制频谱泄漏。
由于人类感官的分辨能力存在极限,因此很多有损压缩算法利用这一点将语音、音频、图像、视频等信号的高频部分除去。高频信号对应于信号的细节,滤除高频信号可以在人类感官可以接受的范围内获得很高的压缩比。这一去除高频分量的处理就是通过离散傅里叶变换完成的。将时域或空域的信号转换到频域,仅储存或传输较低频率上的系数,在解压缩端采用逆变换即可重建信号。
离散傅里叶变换及其多维形式在偏微分方程的求解中也有应用。此时DFT被看作的近&#20284;。傅里叶级数将函数在复指数einx上展开,这正是微分算子的特征方程:d/dx einx
= in einx。因此,通过傅里叶级数的形式,线性常微分方程被转换为代数方程,而后者是很容易求解的。此时得到的结果是偏微分方程解的级数表示,只要通过DFT逆变换即可得到其一般表示。这种方法被称作谱方法或级数解法。
目前长整数或最快速的算法是基于离散傅里叶变换的。由于整数(或多项式)乘法是逐位(或逐项)乘累加的形式,因此整数(或多项式)乘积的数字(或系数)可以用乘数数字(或乘式系数)的表示。利用,只要将数字(或系数)序列通过离散傅里叶变换变到频域,就可以将逐个乘累加的卷积变为对位的乘法,从而减少计算量,再以一次逆变换便可以得到乘法结果。需要注意整数乘法还有的问题。下面以多项式乘法为例说明这一应用:
假设需要计算多项式乘法c(x) = a(x) · b(x),这一乘积结果的各项系数的表达式实际上是线性卷积的形式。由于离散傅里叶变换对应于圆周卷积,因此需要将这两个乘式的系数按照次数升序排列,序列后“补零”,使它们的长度d 大于两式最高项次数的和:d & deg(a(x)) &#43; deg(b(x))。然后作圆周卷积:
其中c 就是c(x) 系数的矢量。由于圆周卷积的DFT是乘积,有:
利用快速傅里叶变换,这一算法的运算复杂度为 。
OFDM(正交频分复用)在宽带中有重要的应用。这种技术将带宽分为N个等间隔的子载波,可以证明这些子载波相互正交。尤其重要的是,OFDM调制可以由IDFT实现,而解调可以由DFT实现。OFDM还利用DFT的移位性质,在每个帧头部加上循环前缀(Cyclic
Prefix),使得只要信道延时小于循环前缀的长度,就能消除信道延时对传输的影响。
本文已收录于以下专栏:
相关文章推荐
本篇博客主要是举个实例来展示离散傅里叶变换的计算过程(因为计算机主要是处理离散数值)...
学习DIP第23天
离散傅里叶变换,数学推导
The result of the DFT is a complex number. This can be plotted either as a magnitued, as ...
using namespac...
他的最新文章
讲师:董岩
您举报文章:
举报原因:
原文地址:
原因补充:
(最多只允许输入30个字)离散傅里叶变换 The Discrete Fourier Transform (DFT) - CSDN博客
离散傅里叶变换 The Discrete Fourier Transform (DFT)
The Discrete Fourier Transform (DFT)
The Java code to calculate the basic Discrete Fourier Transform can be found&&in&tar&format.
See the&&for a link to the tar program.
When I read that a DFT is calculated over a set of points, I expected that this referred to a sliding window over a larger data set. This incorrect assumption resulted from the fact that I came to Fourier analysis
via wavelets, where a sliding window is used. In fact, the number of points refers to the data set over which the DFT is calculated. So if there are 64-points in the data, a 64-point DFT is calculated. The DFT algorithm is an N2&algorithm, so computation
time goes up rapidly with the number of points.
The DFT sums a set of cosine and sine functions scaled by the points in the data sample. I've tried to show this in the graphs below. These graphs have 17 data points (e.g., a 17-point DFT). The points are generated
from the function&sin(x). The&sin(x)&function and the data points are shown in red.
These graphs are displayed using&gnuplot&from data generated by the Java class&dft_graph&which can be down loaded&
The result of the DFT function, calculated over a range from 0..i&is a set of complex objects. The&mag()&method will return the magnitude of the complex value. The magnitude graph for&sin(x)&is
shown below. Only N/2 points are shown (in this case N is 17, so N/2 is 8). Only half of the points are graphed since the graph is symmetric and the next 8 points would mirror the first 8 points. The magnitude is adjusted to the same scale as the input signal,
using the equation
adjMag = (2 * mag)/N
Where N is the number of points in the DFT. Note that the magnitude of the&sin(x)&is 1.0. Since there is only one wave form, there should be only one magnitude. However, the peak magnitude
shown in this graph is slightly less than 1.0 and there are smaller sub-magnitudes. This is an example of what Lyons calls &DFT leakage&.
Two period sin(x)
Note that the largest magnitude line is at 2. The sample over which the DFT was calculated has two sine periods. If this were a sampled signal we could relate this to frequency. The magnitude graph below is from
a sample with four sine periods. As with the magnitude plot above, only N/2 magnitudes are displayed, since the other half of the magnitude plot is symmetric.
Four period sin(x)
The interpretation of magnitude plots gets more obscure as the signal gets more complex. The function
f(x) = sin(2x) &#43; sin(x) &#43; sin(x/2)
is shown below.
The signal in this graph is made up of the sum of 8-periods from&sin(2x), 4-periods from&sin(x)&and 2-periods from&sin(x/2). The graph below shows magnitudes that result from the DFT of
this signal (sorry, this version of the DFT graph is mislabeled).
Each of the large magnitude points above represents one of the frequency components of the signal. In absolute terms this frequency relates to the sample. The sample contains the sum of a two period sine wave, a
four period sine wave and an eight period sine wave. The relationship between the sample frequency and the sampled signal depends on the&sampling&frequency. By including more signal periods in the sample, the DFT frequency goes up. For example, if
the signal above is multiplied by four, we get the equation:
f(x) = sin(8x) &#43; sin(4x) &#43; sin(2x)
Keeping the sampling region constant (0..25) on the x-axis yields a sample with four times as many periods. The graph is shown below (this version of the DFT graph is mislabeled).
Graphing the magnitudes that result from the DFT of this sample also shows the mangnitude frequencies increased by a factor of four as well.
Since the frequency of the DFT result is related to the sample, if the sample is reduced, so that there are only have two periods in the sample, the DFT result will show a lower frequency. The graph below shows&f(x)
= sin(8x) &#43; sin(4x) &#43; sin(2x), however the sample region is 0 ... ~6.25 (e.g., 2Pi)
The DFT magnitude result is plotted below. Note that this result is the same as
f(x) = sin(2x) &#43; sin(x) &#43; sin(x/2).
Postscript
I attempted to regenerate some of the graphs that I've displayed here in my web pages that discuss time/frequency analysis using wavelets. I noticed that the equations that I used to label the plots were incorrect.
In particular where I should have written sin(8x) I wrote 8*sin(x). The discussion and the the plots of the sine functions are labeled correctly now, but the DFT plots are still incorrect. I hope that the reader will forgive this error.
Ian Kaplan, September 2001
Revised: April 2002
The Discrete Fourier Transform (DFT)
The Java code to calculate the basic Discrete Fourier Transform can be found&&in&tar&format.
See the&&for a link to the tar program.
When I read that a DFT is calculated over a set of points, I expected that this referred to a sliding window over a larger data set. This incorrect assumption resulted from the fact that I came to Fourier analysis
via wavelets, where a sliding window is used. In fact, the number of points refers to the data set over which the DFT is calculated. So if there are 64-points in the data, a 64-point DFT is calculated. The DFT algorithm is an N2&algorithm, so computation
time goes up rapidly with the number of points.
The DFT sums a set of cosine and sine functions scaled by the points in the data sample. I've tried to show this in the graphs below. These graphs have 17 data points (e.g., a 17-point DFT). The points are generated
from the function&sin(x). The&sin(x)&function and the data points are shown in red.
These graphs are displayed using&gnuplot&from data generated by the Java class&dft_graph&which can be down loaded&
The result of the DFT function, calculated over a range from 0..i&is a set of complex objects. The&mag()&method will return the magnitude of the complex value. The magnitude graph for&sin(x)&is
shown below. Only N/2 points are shown (in this case N is 17, so N/2 is 8). Only half of the points are graphed since the graph is symmetric and the next 8 points would mirror the first 8 points. The magnitude is adjusted to the same scale as the input signal,
using the equation
adjMag = (2 * mag)/N
Where N is the number of points in the DFT. Note that the magnitude of the&sin(x)&is 1.0. Since there is only one wave form, there should be only one magnitude. However, the peak magnitude
shown in this graph is slightly less than 1.0 and there are smaller sub-magnitudes. This is an example of what Lyons calls &DFT leakage&.
Two period sin(x)
Note that the largest magnitude line is at 2. The sample over which the DFT was calculated has two sine periods. If this were a sampled signal we could relate this to frequency. The magnitude graph below is from
a sample with four sine periods. As with the magnitude plot above, only N/2 magnitudes are displayed, since the other half of the magnitude plot is symmetric.
Four period sin(x)
The interpretation of magnitude plots gets more obscure as the signal gets more complex. The function
f(x) = sin(2x) &#43; sin(x) &#43; sin(x/2)
is shown below.
The signal in this graph is made up of the sum of 8-periods from&sin(2x), 4-periods from&sin(x)&and 2-periods from&sin(x/2). The graph below shows magnitudes that result from the DFT of
this signal (sorry, this version of the DFT graph is mislabeled).
Each of the large magnitude points above represents one of the frequency components of the signal. In absolute terms this frequency relates to the sample. The sample contains the sum of a two period sine wave, a
four period sine wave and an eight period sine wave. The relationship between the sample frequency and the sampled signal depends on the&sampling&frequency. By including more signal periods in the sample, the DFT frequency goes up. For example, if
the signal above is multiplied by four, we get the equation:
f(x) = sin(8x) &#43; sin(4x) &#43; sin(2x)
Keeping the sampling region constant (0..25) on the x-axis yields a sample with four times as many periods. The graph is shown below (this version of the DFT graph is mislabeled).
Graphing the magnitudes that result from the DFT of this sample also shows the mangnitude frequencies increased by a factor of four as well.
Since the frequency of the DFT result is related to the sample, if the sample is reduced, so that there are only have two periods in the sample, the DFT result will show a lower frequency. The graph below shows&f(x)
= sin(8x) &#43; sin(4x) &#43; sin(2x), however the sample region is 0 ... ~6.25 (e.g., 2Pi)
The DFT magnitude result is plotted below. Note that this result is the same as
f(x) = sin(2x) &#43; sin(x) &#43; sin(x/2).
Postscript
I attempted to regenerate some of the graphs that I've displayed here in my web pages that discuss time/frequency analysis using wavelets. I noticed that the equations that I used to label the plots were incorrect.
In particular where I should have written sin(8x) I wrote 8*sin(x). The discussion and the the plots of the sine functions are labeled correctly now, but the DFT plots are still incorrect. I hope that the reader will forgive this error.
Ian Kaplan, September 2001
Revised: April 2002
本文已收录于以下专栏:
相关文章推荐
/*******************************************************************************
** 程序名称:离散傅里叶变换(DFT...
傅丽叶变换(二)
——(java)算法实现
离散傅里叶变换
离散傅里叶变换使得数学方法与计算机技术建立了联系,这就为傅里叶变换这样一个数学工具在实用中开辟了一条宽阔的道路。因此,它不仅仅...
要:本实验开发了一个2-D FFT程序包,主要实现图像的二维快速傅里叶变换、图像傅里叶频谱计算及高斯低通滤波器。傅里叶变换是数字图像处理技术的基础,是将时域信号分解为不同频率的正弦和或余弦和...
冈萨雷斯版里面的解释非常形象:一个恰当的比喻是将傅里叶变换比作一个玻璃棱镜。棱镜是可以将光分解为不同颜色的物理仪器,每个成分的颜色由波长(或频率)来决定。傅里叶变换可以看作是数学上的棱镜,将函数基于频...
opencv中提供了傅里叶变换函数cvDFT,执行二维浮点数离散傅里叶变换的代码如下:
void fft2(const IplImage* src, IplImage* dst)
using namespac...
学习DIP第24天
转载请标明本文出处:http://blog.csdn.net/tonyshengtan,欢迎大家转载,发现博客被某些论坛转载后,图像无法正常显示,无法正常表达本人观点,对此表...
1.离散FFT简单介绍
FFT是一种DFT的高效算法,称为快速傅立叶变换(fast Fourier transform)。其原理比较复杂,我们可以不关其具体
细节,值得注意的是:二维FFT可以...
他的最新文章
讲师:董岩
您举报文章:
举报原因:
原文地址:
原因补充:
(最多只允许输入30个字)

我要回帖

更多关于 matlab离散傅里叶变换 的文章

 

随机推荐