RE: Misc Codec Questions....

Hi...

It makes sense now :) I don't want to introduce the non-linearities as i'd
rather stick to integer values. So i guess i'll either do what happens with
the Two-Six transform where the normalization for the xform and inverse
xform is accounted in the xform or i'll just change my zerotree algorithm to
account for the different significance levels.

Thanks!!!
	Sebastien St-Laurent

---

Hi Sebastien,

On Tue, 2002-05-21 at 16:43, Sebastien St-Laurent wrote:
> >> Hi.
> >>
> >> First of all, can someone confirm if the behavior i'm getting with
> >> the (2,2)
> >> wavelet is the same (i,e: No bit growth)?
> >
> >You can see this by looking on the equation for the lowpass
> >coefficient, it's simply a truncated (or rounded) weighted average of
> >the neighbor coefficient e.g. (x_i + x_{i+1})/2 for the (1,1) wavelet.
> >There is no bit growth.

There is no bit growth, ignoring normalisation constants, and ignoring
the fact that you introduce nonlinearity by rounding/truncating. You can
opt for bit growth if you need linearity. Don't forget your sqrt(2) and
sqrt(2)/2 normalisation factors.

> Which i guess is good since you don't need 16 bits to cary around so it's
> more memory efficient. I stil lhave to figure out why my SPHIT coder is
> doing so bad with this wavelet.

I would hazard a guess that perhaps you aren't taking the cumulative
normalisation factor into account when you are checking for
significance?

Keep well,
Ray
--
--
http://www.mythral.org/~ray/


--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to
'[email protected]'
containing only the word 'unsubscribe' in the body.  No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.


--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to '[email protected]'
containing only the word 'unsubscribe' in the body.  No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.