Home > Archive > Compression > January 2008 > Range of k in JPEG-LS Golomb encoding
You are viewing an archived Text-only version of the thread.
To view this thread in it's original format and/or if you want to reply to
this thread please [click here]
| Author |
Range of k in JPEG-LS Golomb encoding
|
|
| LiloLilo 2008-01-08, 6:57 pm |
| Hi,
I am trying to understand use of Golomb-Rice Coding in JPEG-LS. As
described in these papers, http://www.hpl.hp.com/loco/, the value of k, used
to find the Golomb(k) of a prediction residual, is cumputed using the
following:
for (k=0; (N<<k)<A; k++);
So, I am trying to figure out the possible range of values k can assume, I
suppose starting of 0 to a low integer value. Anyone can help?
Thanks in advance.
| |
| Marco Al 2008-01-08, 6:57 pm |
| LiloLilo wrote:
> for (k=0; (N<<k)<A; k++);
>
> So, I am trying to figure out the possible range of values k can assume, I
> suppose starting of 0 to a low integer value. Anyone can help?
N is the number of previous samples in the current context and A is the
biased average of the absolute values of those samples. With 8 bit input
A can not be larger than 255*N, so k can't go higher than 7.
Marco
| |
| LiloLilo 2008-01-09, 6:57 pm |
| >> for (k=0; (N<<k)<A; k++);
> N is the number of previous samples in the current context and A is the
> biased average of the absolute values of those samples. With 8 bit input A
> can not be larger than 255*N, so k can't go higher than 7.
Thank you for help. So, only 8 Golomb codes are involved in JPEG-LS coding
for 8 bit channel images. Nice.
| |
| LiloLilo 2008-01-09, 6:57 pm |
| Is this method copyright protected?
"Marco Al" <m.f.al@versatel.nl> wrote in message
news:fm15tm$pbq$1@netlx020.civ.utwente.nl...
> LiloLilo wrote:
>
>
> N is the number of previous samples in the current context and A is the
> biased average of the absolute values of those samples. With 8 bit input A
> can not be larger than 255*N, so k can't go higher than 7.
>
> Marco
| |
| Marco Al 2008-01-09, 9:57 pm |
| LiloLilo wrote:
> Is this method copyright protected?
Hypothetically, yes there might be a patent which has this method as one
of it's claims. I'm not sure how valid that claim is since using the
average of absolute values for determining the parameter is very old
(the shorten paper is from 1992).
Marco
| |
| Thomas Richter 2008-01-10, 3:56 am |
| Marco Al schrieb:
> LiloLilo wrote:
>
> Hypothetically, yes there might be a patent which has this method as one
> of it's claims. I'm not sure how valid that claim is since using the
> average of absolute values for determining the parameter is very old
> (the shorten paper is from 1992).
The baseline technology of everything standardized from JPEG is always
licence-fee free. Which means that writing a licence request to, in this
case HP, will be enough to grant you a licence, with no additional
charges added. If you plan to use it outside of JPEG-LS, then this might
be a different question, but as long as you don't want to sell this as a
product, it is very unlikely to cause trouble in first place.
"Copyright free" is nothing - there is always a copyright, at least
owned by the authors. Copyright can be transfered, and you might get a
licence from the owners of the copyright.
So long,
Thomas
|
|
|
|
|