Chroma Decimation in Digital Video Signals

The chroma aliasing certainly occurs and can be seen for example as
false color fringes on high frequency details - eg closely spaced white
fence posts.

Chris
=========================
Chris Bore
BORES Signal Processing
www.bores.com

The chroma aliasing certainly occurs and can be seen for example as
false color fringes on high frequency details - eg closely spaced white
fence posts.

Chris
=========================
Chris Bore
BORES Signal Processing
www.bores.com

Randy Yates wrote:
> I've seen several cases where the chroma decimation, e.g.,
> when going from YCrCb 4:4:4 to 4:2:2, is done by simply
> throwing away every other sample. I have two questions
> regarding this operation:
>
>   1. If there were a 1-D signal, simply throwing away
>   every other sample without first filtering would cause
> (potentially)
>   aliasing. Why doesn't this happen in a chroma signal, or
>   why doesn't it matter if it does happen?

Often, the chroma data in the original content has already
been band limited to below 1/4th the sampling frequency.  For
instance, in many digital cameras, a color mosaic in front
of the sensor pixels only allows sampling color at half of
the spacial sampling rate.  Data converted from analog video
may already have had the chroma channels band limited to meet
the NTSC spec.

But if the data is from, say, a non-filtered synthesized
RGB source or other hires source, then there can be aliasing.


IMHO. YMMV.
-- 
rhn A.T nicholson d.O.t C-o-M

Hi Everyone,

Thanks very much for your replies. It is very useful to see a
variety of responses - gives you more "averaging" to get a more
accurate answer (much like delta-sigma processing!).

Chris, it's nice to see you come out of hiding and actually make
a post or two on comp.dsp!

Chris (et al.), is this chroma filtering problem what causes the
moire pattern you see on certain image inputs (vertical lines?).

--Randy

.

Randy Yates wrote:

   ...

> Chris (et al.), is this chroma filtering problem what causes the
> moire pattern you see on certain image inputs (vertical lines?).

What image source? With NTSC, it's the beat between luminance and the 
3.58 MHz color subcarrier. A comb filter can greatly reduce that effect.

Jerry
-- 
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������

Randy Yates wrote:

> Chris (et al.), is this chroma filtering problem what causes the
> moire pattern you see on certain image inputs (vertical lines?).

Uhm, I do not think that is the main reason.

I saw the moire as consequence, as Jerry says, of the
chroma separation filter, in case of composite signal.
Or, in case picture processing, where certain bandwidth
conditions were not respected.

bye,

-- 

piergiorgio

>Often, the chroma data in the original content has already
>been band limited to below 1/4th the sampling frequency.  For
>instance, in many digital cameras, a color mosaic in front
>of the sensor pixels only allows sampling color at half of
>the spacial sampling rate.

I think a color mosaic only samples, it does not band limit.

Chris

It can look like a moire pattern, but is more like color fringes around
the high-frequency detail. But of course a similar effect also arises
from analog interference as Jerry says. The classic moire pattern is
simple aliasing - the color version changes colors because the color is
aliased differently from the luminance and depends on the precise
sampling pattern (eg Bayer).

To see that it results from sampling, take an image with lots of
pixels, YUV444 or an RGB equivalent, and some high-frequncy detail
(preferably white), and subsample the chroma - without filtering.

Since Foveon offer sensors with coincident RGB pixles I expect they
have some explanations of this (since it justifies their sensor
patterns).


One of the Swiss universities (I think) has a good tutorial with images
on the web. Sorry, can't remember who.. :-(

Chris

Chris Bore wrote:

>>Often, the chroma data in the original content has already
>>been band limited to below 1/4th the sampling frequency.  For
>>instance, in many digital cameras, a color mosaic in front
>>of the sensor pixels only allows sampling color at half of
>>the spacial sampling rate.

> I think a color mosaic only samples, it does not band limit.

In all video camera systems, as far as I know, it is the
optical system that does band limiting.

Well, for scanned electron beam systems you might do some
with beam spot size, but mostly the optical system.

-- glen

Yes, that would be correct.

But I think it is hard to arrange optical band limiting that is
different for red and blue than it is for green, so it is hard to band
limit red and blue to half the bandwidth of green? This is what would
be necessary to cope with a Bayer mosaic sensor.

Chris