DSPRelated.com
Forums

image subsampling problem???

Started by walala March 22, 2004
In article <40629864$0$263$4d4ebb8e@read.news.de.uu.net>,
Piergiorgio Sartor  <piergiorgio.sartor@nexgo.REMOVE.THIS.de> wrote:
>JPEG compression occurs after sampling, so it does >not partecipate to the anti aliasing process.
This is true if the aliasing folds all the way down to DC (and past). But if significant aliasing energy only folds into the sampled spectrum near the Nyquist frequency (due to some partial low-pass filtering), then it seems like some post filtering will help cut down the alias artifacts (as well as components of the image data). I think this may be roughly similar to some over-sampling A/D techniques. IMHO. YMMV. -- Ron Nicholson rhn AT nicholson DOT com http://www.nicholson.com/rhn/ #include <canonical.disclaimer> // only my own opinions, etc.

"Ronald H. Nicholson Jr." wrote:
> > In article <40629864$0$263$4d4ebb8e@read.news.de.uu.net>, > Piergiorgio Sartor <piergiorgio.sartor@nexgo.REMOVE.THIS.de> wrote: > >JPEG compression occurs after sampling, so it does > >not partecipate to the anti aliasing process. > > This is true if the aliasing folds all the way down to DC (and past).
It doesn't fold down to DC. In your previous post you mentioned the rectangular sampling - if you inspect the frequency response of same, you will find it has a zero at/near the sampling frequency (and of course multiples of Fs), so the signal is very clean at/near DC. Any aliasing artifacts are at high frequency. Most of this is not perceived by humans. At normal viewing distances and reproduction resolutions our eyeballs filter out most of the high frequency content. Jpeg eliminates much of the same frequency content, but not for the purpose of cleaning up the aliasing artifacts but simply because our eyes don't care and therefore its excess information. -jim
> But if significant aliasing energy only folds into the sampled spectrum > near the Nyquist frequency (due to some partial low-pass filtering), then > it seems like some post filtering will help cut down the alias artifacts > (as well as components of the image data). I think this may be roughly > similar to some over-sampling A/D techniques. > > IMHO. YMMV. > -- > Ron Nicholson rhn AT nicholson DOT com http://www.nicholson.com/rhn/ > #include <canonical.disclaimer> // only my own opinions, etc.
-----= Posted via Newsfeeds.Com, Uncensored Usenet News =----- http://www.newsfeeds.com - The #1 Newsgroup Service in the World! -----== Over 100,000 Newsgroups - 19 Different Servers! =-----
jim wrote:

> It doesn't fold down to DC. In your previous post you mentioned > the rectangular sampling - if you inspect the frequency response > of same, you will find it has a zero at/near the sampling > frequency (and of course multiples of Fs), so the signal is very > clean at/near DC. Any aliasing artifacts are at high frequency.
If I understand correctly, this means it is not possible to sample DC. bye, -- piergiorgio
Piergiorgio Sartor wrote:
> jim wrote: > >> It doesn't fold down to DC. In your previous post you mentioned >> the rectangular sampling - if you inspect the frequency response >> of same, you will find it has a zero at/near the sampling >> frequency (and of course multiples of Fs), so the signal is very >> clean at/near DC. Any aliasing artifacts are at high frequency. > > > If I understand correctly, this means it is not possible > to sample DC. > > bye,
How is that? Surely a digital camera can distinguish between featureless black and featureless white. What part of the theory suggests otherwise? Jerry -- Engineering is the art of making what you want from things you can get. &#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;
Jerry Avins wrote:

> How is that? Surely a digital camera can distinguish between featureless > black and featureless white. What part of the theory suggests otherwise?
I do not know, I do not understand the statement about a zero at multiples of Fs, since also DC is such a multiple. And also what happens if there is a picture, with vertical (or horizontal) lines at exactly 2Fs? bye,. -- piergiorgio
Piergiorgio Sartor wrote:

> Jerry Avins wrote: > >> How is that? Surely a digital camera can distinguish between featureless >> black and featureless white. What part of the theory suggests otherwise? > > > I do not know, I do not understand the statement > about a zero at multiples of Fs, since also DC is > such a multiple. > > And also what happens if there is a picture, with > vertical (or horizontal) lines at exactly 2Fs? > > bye,.
Consider the sinc. It has zeros at evenly spaced multiples _except_ at zero; there, it is unity. As for stripes at Fs/2 imaged onto rectangular pixels, it can come out as alternating black and white, uniform gray, or anything between, depending on how the image is aligned on the image. In other words, the modulation transfer function (MTF) can take any value from zero to one. Jerry -- Engineering is the art of making what you want from things you can get. &#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;
Jerry Avins wrote:

> Consider the sinc. It has zeros at evenly spaced multiples _except_ at > zero; there, it is unity.
Ah, OK, now it's clear.
> As for stripes at Fs/2 imaged onto rectangular pixels, it can come out > as alternating black and white, uniform gray, or anything between, > depending on how the image is aligned on the image. In other words, the > modulation transfer function (MTF) can take any value from zero to one.
Well, but this means there is aliasing at DC, or, at least, there is potential aliasing at DC. bye, -- piergiorgio
Jerry Avins wrote:

> Piergiorgio Sartor wrote:
>> If I understand correctly, this means it is not possible >> to sample DC.
> How is that? Surely a digital camera can distinguish between featureless > black and featureless white. What part of the theory suggests otherwise?
The autoexposure system. Well, I suppose between black and white, but not between dark gray and light gray. Audio systems tend not to be DC coupled, anyway, but many also have automatic level control. -- glen
Jerry Avins wrote:

> Piergiorgio Sartor wrote: > >> Jerry Avins wrote: >> >>> How is that? Surely a digital camera can distinguish between featureless >>> black and featureless white. What part of the theory suggests otherwise? >> >> >> >> I do not know, I do not understand the statement >> about a zero at multiples of Fs, since also DC is >> such a multiple. >> >> And also what happens if there is a picture, with >> vertical (or horizontal) lines at exactly 2Fs? >> >> bye,. > > > Consider the sinc. It has zeros at evenly spaced multiples _except_ at > zero; there, it is unity. > > As for stripes at Fs/2 imaged onto rectangular pixels, it can come out > as alternating black and white, uniform gray, or anything between, > depending on how the image is aligned on the image. In other words, the
Oops! Aligned on the sensor array, of course.
> modulation transfer function (MTF) can take any value from zero to one. > > Jerry
-- Engineering is the art of making what you want from things you can get. &#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;
Piergiorgio Sartor wrote:

> Jerry Avins wrote: > >> Consider the sinc. It has zeros at evenly spaced multiples _except_ at >> zero; there, it is unity. > > > Ah, OK, now it's clear. > >> As for stripes at Fs/2 imaged onto rectangular pixels, it can come out >> as alternating black and white, uniform gray, or anything between, >> depending on how the image is aligned on the image. In other words, the >> modulation transfer function (MTF) can take any value from zero to one. > > > Well, but this means there is aliasing at DC, or, at least, > there is potential aliasing at DC. > > bye,
More than potential aliasing. Only perfect alignment can avoid a seeming DC component. Aliasing at DC is not what I asked you about. A zero in the transfer function there predicted by theory is what puzzled me. You answered that by expressing your assumption that what is true for 1, 2, 3, ... is also true for 0. Sinc is a counterexample. Aliasing to DC is true in general for all systems where the MTF falls off at high frequencies i.e., all real systems. The uniform gray reminiscent of the heat death of the universe that all such imaging systems must show at high enough frequencies makes that self evident. Jerry -- Engineering is the art of making what you want from things you can get. &#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;