Databases Reference
In-Depth Information
Y
U
V
F I GU R E 19 . 6
Recommendation 601 4:2:2 sampling format.
as a triple of integers, with the first integer corresponding to the sampling of the luminance
component and the remaining two corresponding to the chrominance components. Thus, 4:4:4
sampling means that all components were sampled at 13.5MHz. The most popular sampling
format is the 4:2:2 format, in which the luminance signal is sampled at 13.5MHz, while the
lower-bandwidth chrominance signals are sampled at 6.75MHz. If we ignore the samples of
the portion of the signal that do not correspond to active video, the sampling rate translates to
720 samples per line for the luminance signal and 360 samples per line for the chrominance
signal. The sampling format is shown in Figure
19.6
. The luminance component of the digital
video signal is also denoted by
Y
, while the chrominance components are denoted by
U
and
V
. The sampled analog values are converted to digital values as follows. The sampled values
of
YC
b
C
r
are normalized so that the sampled
Y
values,
Y
s
, take on values between 0 and 1,
and the sampled chrominance values,
C
rs
and
C
bs
, take on values between
−
2
and
2
. These
normalized values are converted to 8-bit numbers according to the transformations
Y
=
219
Y
s
+
16
(5)
U
=
224
C
bs
+
128
(6)
V
=
224
C
rs
+
128
(7)
Thus, the
Y
component takes on values between 16 and 235, and the
U
and
V
components
take on values between 16 and 240.
An example of the
Y
component of a CCIR 601 frame is shown in Figure
19.7
.Inthe
top image we show the fields separately, while in the bottom image the fields have been
interlaced. Notice that in the interlaced image the smaller figure looks blurred. This is because
the individual moved in the sixtieth of a second between the two fields. (This is also proof—if
any was needed—that a three-year-old cannot remain still, even for a sixtieth of a second!)
The
YUV
data can also be arranged in other formats. In the Common Interchange Format
(CIF), which is used for videoconferencing, the luminance of the image is represented by an
array of 288
352 pixels, and the two chrominance signals are represented by two arrays
consisting of 144
×
176 pixels. In the QCIF (Quarter CIF) format, we have half the number
of pixels in both the rows and columns.
The MPEG-1 algorithm, which was developed for encoding video at rates up to 1.5 Mbits
per second, uses a different subsampling of the CCIR 601 format to obtain the MPEG-SIF
format, where SIF stands for Source Input Format. Starting from a 4:2:2, 480-line CCIR 601
×
