next up previous
Next: Two-Dimensional Fourier Transform (2DFT) Up: fourier Previous: Fast Fourier Transform (FFT)

Fourier Transform 2 Real Functions with 1 DFT

First we recall the symmetry properties of the DFT. The DFT of $x[m]=x_r[m]+jx_i[m]$ is defined as
$\displaystyle X[n]$ $\textstyle =$ $\displaystyle \sum_{m=0}^{N-1} x[m]e^{-j2\pi mn/N}=\sum_{m=0}^{N-1} [x_r[m]+jx_i[m]]
[cos(2\pi mn/N)-j\;sin(2\pi mn/N)]$  
  $\textstyle =$ $\displaystyle \sum_{m=0}^{N-1} [x_r[m]cos(\frac{2\pi mn}{N})+x_i[m]sin(\frac{2\...
...
+j\sum_{m=0}^{N-1} [x_i[m]cos(\frac{2\pi mn}{N})-x_r[m]sin(\frac{2\pi mn}{N})]$  
  $\textstyle =$ $\displaystyle X_r[n]+jX_i[n]$  

where $X_r[n]$ and $X_i[n]$ are the real and imaginary part of the spectrum respectively. If $x[m]$ is real, i.e., $x_i[m] \equiv 0$, then we have

\begin{displaymath}\left\{ \begin{array}{c} X_r[-n]=X_r[n]  X_i[-n]=-X_i[n] \end{array} \right.
\end{displaymath}

or

\begin{displaymath}X[-n]=X_r[-n]+jX_i[-n]=X_r[n]-jX_i[n]=X^*[n] \end{displaymath}

If $x[m]$ is imaginary, i.e., $x_r[m] \equiv 0$, then we have

\begin{displaymath}\left\{ \begin{array}{c} X_r[-n]=-X_r[n]  X_i[-n]=X_i[n] \end{array} \right.
\end{displaymath}

or

\begin{displaymath}X[-n]=X_r[-n]+jX_i[-n]=-X_r[n]+jX_i[n]=-X^*[n] \end{displaymath}

Next we show how an arbitrary function $f(x)$ can be decomposed into the even and odd components $f_e(x)$ and $f_o(x)$:

\begin{displaymath}
\left\{ \begin{array}{c}
f_e(x)=(f(x)+f(-x))/2=f_e(-x)  f_o(x)=(f(x)-f(-x))/2=-f_o(-x) \end{array} \right.
\end{displaymath}

and

\begin{displaymath}f_e(x)+f_o(x)=f(x) \end{displaymath}

Now we are ready to show how to Fourier transform two real functions $x_1[m]$ and $x_2(m)$ to get their spectra $X_1[n]$ and $X_2[n]$ by one DFT.

  1. Define a complex function $x[m]$ by the two real functions:

    \begin{displaymath}x[m] \stackrel{\triangle}{=}x_1[m]+jx_2[m] \end{displaymath}

    Notice here that we impose $j$ on $x_2[m]$ to make it imaginary.
  2. Find the DFT of $x[m]$

    \begin{displaymath}DFT[x[m]]=X[n]=X_r[n]+jX_i[n] \end{displaymath}

  3. Separate $X[n]$ into $X_1[n]$ and $X_2[n]$, the spectra of $x_1[m]$ and $x_2[m]$, using the symmetry properties discussed previously. Note that $X[-n]=X[N-n]$ because $X[n]$ is a periodic function.


next up previous
Next: Two-Dimensional Fourier Transform (2DFT) Up: fourier Previous: Fast Fourier Transform (FFT)
Ruye Wang 2009-12-31