Page 113 - DSP Integrated Circuits

98 Chapter 3 Digital Signal Processing

It is possible to overlap the I/O operations, i.e., writing and reading data,
and computation of the FFT, thereby effectively extending the available time for
computing the FFT (see Problem 3.28). For the sake of simplicity we assume
that this possibility is not exploited. The available time for the FFT is therefore
0.372 ms.
The I/O processes will handle both the rearranging of data that is required to
compute the IFFT and the unscrambling of the data array in the last stage of the
FFT. As mentioned in section 3.16.4, the IFFT can be obtained by interchanging
the real and imaginary parts, when the data are both written into and read from
the memory. We can implement this scheme by changing the addressing of the
memories in the FFT processor or, alternatively, by interchanging the inputs to the
butterflies in the first stage and interchanging the outputs from the butterflies in
the last stage. The unscrambling can be accomplished by reversing the address
lines to the memory when reading data out of the FFT processor. Hence, both of
these tasks can be accomplished without any extra processing time.

3.18 IMAGE CODING

Efficient image coding techniques are
required in many applications. Figure
3.29 illustrates a typical image trans-
mission system for HDTV (high-defini-
tion TV) where the aim is to reduce the
number of bits to be transmitted over
the transmission channel. Modern cod-
ing techniques can compress images by
a factor of 10 to 50 without visibly
affecting the image quality.
Ordinary TV signals require a
transmission capacity of 216 Mbit/s of
which 166 Mbit/s are allocated for the
video signal. The remaining bits are
used for the sound signal and for syn-
chronization purposes. The required bit
rates are roughly five times larger for
HDTV. Efficient data compression tech-
niques therefore must be employed,
since the amount of transmitted infor-
mation is very large.
Other major applications of trans-
form coding techniques are to be found
in image storage and retrieval systems, Figure 3.29 Transform coding system
such as digital VCRs. Multimedia,
graphic arts, and desktop publishing
are examples of important new applications that emerge. Efficient high-speed com-
pression schemes are necessary in these applications to store and retrieve huge