|
YUV Modes
The YUV modes have much in common with
techniques used in analogue broadcast television systems. These
modes are somewhat more complicated than the RGB modes, but they take
advantage of two simple physiological facts:
- Light receptors in the human eye are
more sensitive to light intensity (rod cells) than to colour (cone
cells).
- The eye is most sensitive to the green
part of the spectrum but is less sensitive to reds and
relatively insensitive to blue colours.
To take advantage of this, the largest
part of the
available time is devoted to light intensity information, called luminance, derived from the red, green and blue components,
which enables a
high resolution black-and-white image to be reconstructed. The colour information is
transmitted in two additional low resolution colour difference components, called chrominance
or chroma information, which effectively apply a colour wash to the black-and-white image.
The main drawback of YUV transmission
using SSB equipment is that mis-tuning of the signal causes a
significant hue shift (to green or magenta) instead of simply producing
a slightly lighter or darker image as with the RGB modes. Nevertheless,
the high stability of modern radio equipment together with the tuning
aids which can be incorporated in the decoding software mean that tuning
accuracy is no longer the problem it once was.
Encoding
The basic YUV SSTV system was
pioneered by Robot Research, Inc. of the USA. According to the technical
information given in the handbook for the Robot 1200C scan converter, the
luminance signal Y is computed as follows:
Y = 0.3R + 0.59G + 0.11B
The two chrominance signals are computed
by:
U = B -
Y and V = R - Y
Although a complete line still requires
three components the advantage lies in the fact that the U and V
components can be sent at lower resolution without significant loss of image
quality. This can be done in two ways. Firstly, the U and V
components can be sent faster than the Y components (less time per
pixel). Secondly, each of the U and V components can be averaged over two
lines.
The following table gives values calculated for different coloured pixels
according to the above formulae:
| WHITE |
BLACK |
| R = 255 |
Y = 255 |
R = 0 |
Y = 0 |
| G = 255 |
U = B-Y = 0 |
G = 0 |
U = B-Y = 0 |
| B = 255 |
V = R-Y = 0 |
B = 0 |
V = R-Y = 0 |
| YELLOW |
AQUA |
| R = 255 |
Y = 227 |
R = 0 |
Y = 179 |
| G = 255 |
U = B-Y = - 227 |
G = 255 |
U = B-Y = 76 |
| B = 0 |
V = R-Y = 28 |
B = 255 |
V = R-Y = - 179 |
It will therefore be seen that the luminance
component varies between 0 and 255. The Y signal is therefore transmitted in
the normal way with black corresponding to 1,500 Hz and white at 2,300
Hz. However, the chroma signals can be both positive and negative. In
order to allow for this, 1,900 Hz is taken to represent a value of zero,
with higher frequencies representing positive values and lower
frequencies representing negative values. The range of values
allocated to each of the three components is given in the following
table.
| Frequency
Hz |
Y |
U
(B-Y) |
V
(R-Y) |
| 2,300 |
255 |
227 |
179 |
| 1,900 |
127 |
0 |
0 |
| 1,500 |
0 |
- 227 |
- 179 |
Decoding
The Y, U and V values obtained from the
received line data can be written to a line buffer. Reconstructing
the colour pixels requires reading the corresponding Y, U and V values
from appropriate positions in the buffer and reconstructing the original
RGB values for each pixel. Calculating the red and blue values is
relatively simple:
R = V +
Y and B = U + Y
Retrieving the green values involves a
slightly more complex calculation:
G = Y - 0.5V - 0.186U
EXAMPLE: To
reconstruct an aqua colour pixel from the YUV values in the above table:
R = - 179 + 179 = 0
G = 179 - (0.5 x - 179) -
(0.186 x 76) = 179 + 90 - 14 = 255
B = 76 + 179 = 255
22 January 2008
|