Re: Quality factor and bits per pixel in JPEG

On Mar 24, 10:45 am, "fulltime" <PiaoCh...@xxxxxxxxx> wrote:
On Mar 24, 2:30 am, Thomas Richter <t...@xxxxxxxxxxxxxxxxxxxxxxxxxx>
wrote:



fulltime <PiaoCh...@xxxxxxxxx> wrote:
You really just need to do it iteratively.
If you're doing this on many images you could plot a curve
of file size versus quality factor for a variety of images, and use that
to get you close to the final value based on the file size that
results from an initial quality factor. Otherwise, just try it with
various quality factors to get the file size you need.

This is exactly the point. JPEG-1 doesn't have a rate allocation,
and you really need to try to find a "quality factor" that gives
you the output rate. This quality factor is then used by an
"ad-hoc" formula to compute the quantization factors.

For the grayscale image i mentioned above, 8bits are needed to
represent each pixel, so am i correct to say its 8bpp?

Yes.

To attain say
0.5bpp, given 512 by 512 pixels, i will need the file size to be of
(512*512) / 2 bits??

Exactly. Divide by eight to get the number of bytes.

another qn is the file size we are talking abt here is the number of
bits tat we need to encode the JPEG image right?

Yes.

As a rule of thumb, a "sane" image compression scheme should approximately
improve the image quality (as in PSNR) by 6dB per bit (as in bpp) which
allows you at least to get an initial guess on how the graph should look
like. JPEG-1 isn't quite that sane (the slope is too low), but it's close.

Greetings,
Thomas

thks Peter and Thomas. I will proceed w my work and will post here
shld i encounter any problems. U guys are so willing to share your
knowledge :)


Hi gentlemen,

I have another 2 questions regarding quality factor.

to quantize a coefficient with a quality factor, Qs, it is =
Coefficient / (Quant Matrix Value * Qs)

1) i have been searching the net and people are talking about a
quality factor of 0 to 100 with 100 being the best quality. So for
quality of 100, does tat mean Qs = 1 in the eqn and Qs=0.5 for a
quality of 50?

2) how is this Qs going to reduce the file size, since if Qs is small,
it will require more bits to quantise the coefficient?

thks for solving my doubts
FT


.

Relevant Pages

  • Re: Image Quality
    ... adjustment for download speed/picture quality. ... > Zone Alarm Pro can block Images in IE ... > Scroll down to the Multimedia section and make sure the box>Show> Pictures> ...
    (microsoft.public.windows.inetexplorer.ie6.browser)
  • Re: Image Quality
    ... adjustment for download speed/picture quality. ... >> Zone Alarm Pro can block Images in IE ... >> Pictures Are Not Displayed on Web Sites in Internet Explorer ...
    (microsoft.public.windows.inetexplorer.ie6.browser)
  • Re: Image Quality
    ... I am using IE 6 and Windows XP home edition. ... > 'suboptimal quality' is. ... > Zone Alarm Pro can block Images in IE ... > or contains entries that are blocking access to the desired site. ...
    (microsoft.public.windows.inetexplorer.ie6.browser)
  • Re: TIFF vs JEPG
    ... At 133 MB TIFFs, ... this equates to 35 images to a DVD vs 60 images for JPEG. ... the DVD will last 20 years and be transcribed to ... If you are truly doing thousands you need to save high quality ...
    (comp.periphs.scanners)
  • Re: Quality factor and bits per pixel in JPEG
    ... Coefficient / ... This is just one possible way to setup the quantization matrix, ... quality factor of 0 to 100 with 100 being the best quality. ...
    (comp.compression)