Unformatted text preview: l into the frequency domain, the IDFT is used
to map a signal in the frequency domain into time domain. Because, the DFT is an
orthogonal transformation, the inverse is computed using the Hermitian operator .
The cost of inverting an
matrix is proportional to
, the cost of multiplying a
matrix by a vector is proportional to
. We will further diminish the computation cost
of multiplying a matrix times a vector by using the FFT (Fast Fourier Transform). ¡4 X 4 4 4 3.1.2 Zero padding The DFT allows us to transform an points time series into frequency coefﬁcients
, where the index is associated to the discrete frequency , 45 0 4 0¢
4
0
$¢
4
BGEF4D) " " ) B )
0¢
5
44" #IA' § 7 ) 7 $£§ 7 3 1 § 60 0
$¢ 7 5
¤ the frequency axis is sampled every
radians. At this point it appears that
is controlled by the number of samples of the time series . Zero padding can be used to
, in this case we deﬁne a new time series that condecrease the frequency interval
sists of the original time series foll...
View
Full Document
 Winter '12
 Sacchi
 Digital Signal Processing, DFT

Click to edit the document details