Image Processing Reference
In-Depth Information
1.4.2
Computer interfaces
The basic computer
interface
needs to convert an analogue signal from a camera into a set
of digital numbers. The interface system is called a
framegrabber
since it grabs frames of
data from a
video sequence
, and is illustrated in Figure
1.11
. Note that intelligent cameras
which provide digital information do not need this particular interface, just one which
allows storage of their data. However, a conventional camera signal is
continuous
and is
transformed into
digital
(discrete) format using an Analogue to Digital (A/D) converter.
Flash converters
are usually used due to the high speed required for conversion (say 11
MHz that cannot be met by any other conversion technology). The video signal requires
conditioning prior to conversion; this includes DC restoration to ensure that the correct DC
level is attributed to the incoming video signal. Usually, 8-bit A/D converters are used; at
6 dB/bit, this gives 48 dB which just satisfies the CCIR stated
bandwidth
of approximately
45 dB. The output of the A/D converter is often fed to
look-up tables
(LUTs) which
implement designated conversion of the input data, but in hardware, rather than in software,
and this is very fast. The outputs of the A/D converter are then stored in computer memory.
This is now often arranged to be dual-ported memory that is shared by the computer and
the framegrabber (as such the framestore is
memory-mapped
): the framegrabber only takes
control of the image memory when it is acquiring, and storing, an image. Alternative
approaches can use Dynamic Memory Access (DMA) or, even, external memory, but
computer memory is now so cheap that such design techniques are rarely used.
Input
video
Signal
conditioning
Look-up
table
A/D converter
Image memory
Computer
interface
Control
Computer
Figure 1.11
A computer interface - the framegrabber
There are clearly many different ways to design framegrabber units, especially for
specialist systems. Note that the control circuitry has to determine exactly when image data
is to be sampled. This is controlled by synchronisation pulses that are supplied within the
video signal and can be extracted by a circuit known as a sync stripper (essentially a high
gain amplifier). The sync signals actually control the way video information is constructed.
Television pictures are constructed from a set of
lines
, those lines scanned by a camera. In
order to reduce requirements on transmission (and for viewing), the 625 lines (in the
PA L
system
) are transmitted in two
fields
, each of 312.5 lines, as illustrated in Figure
1.12
.
(There was a big debate between the computer producers who don't want interlacing, and
the television broadcasters who do.) If you look at a television, but not directly, the flicker
due to
interlacing
can be perceived. When you look at the television directly, persistence
in the human eye ensures that you do not see the
flicker
. These fields are called the
odd
and
