Re: zigzag scan: Image processing function

------=_Part_21198_1222623.1160225406791
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline

Hi again!

I've done the following.


1. Changed the "argument check" to nargin !=3D 1. I was thinking about
   putting the whole function in the else part of this if clause. I
   know that the error statements exits the function so I guess one
   might as well keep it the way it is. I don't know what the standard
   way of doing these things within the octave community is.

2. It's perfectly fine to implement the new ordering with mtrx=3Dmtrx'
   but in order to show off I did it by changing odd and even operations
   around instead. Since both your and my zigzag behave the same way
   I also renamed the new function to zagzig.

3. Cleaned the square case a bit by replacing n=3Dsize(mtrx) with
   n=3Dlength(mtrx).

One thing I thought of... If you have lots of matrices A1, A2, A3 and
they all are of the same dimensions, then zagzig can be used "once and
for all" with massive CPU savings. First make a "index mapping vector"
like this:

n=3Dsize(A1);
i=3Dzagzig(reshape(1:prod(n),n(1),n(2)));

Thereafter the zagzigs are given by

A1(i), A2(i) and A3(i).

I suspect that this is about an order of magnitude faster than doing
zagzig(A) for each matrix individually.

The perfect way of doing zagzig would be finding out a quick way of
generating the i vector without for-loops and if-statements all
together. I suspect that this can be done easily but I can't figure
out how.

Code attached.

Cheers
Fredrik

On 10/7/06, Muthiah Annamalai <[email protected]> wrote:
> Hello Fredrik,
> I think your code does zigzag on a transposed matrix than the one I am
> speaking about (I have in my zigzag function).
>
> Your code does a zigzag of matrix =3D reshape(1:9,3,3) as follows,
> ##   1   4- 7
> ##   | /  /
> ##   2   5  8
> ##     /   /|
> ##   3 - 6  9
> but I would expect the output to be as [1 4 2 3 5 7 8 6 9]; instead.
> This is the expected behaviour of a DCT frequency coeffients walk-order.
>
> I have also added some documentation to your code. Please see that,
> and also adjust the code.
>
> One way  I did adjust your code was to add one extra line
>
>  mtrx=3Dmtrx'
>
> before the if statement. Your code however is not modified, as I figured
> you may have another way to do the same without a transpose.
>
> -Muthu
>
>
> On Fri, 2006-10-06 at 16:40 +0200, Fredrik B=FClow wrote:
> > Hi Muthiah!
> >
> > I took the liberty of writing my own quicker version of zigzag.  I've
> > also added support for mxn matrices (where m!=3Dn).
> >
> > Code is attached.
> >
> > Cheers
> > Fredrik B=FClow
> >
> > On 10/5/06, Muthiah Annamalai <[email protected]> wrote:
> > >
> > >  Hello there,
> > >  I have a function zigzag() which walks-off a (square)matrix and read=
s its
> > > elements in a zigzag fashion. It is useful for image processing.
> > >
> > >  Example:
> > >  octave  --eval
> > > "x=3D[1:4]'*[1:4];eval('x');tic;printf('%d\n',zigzagscan(x));toc()"
> > > -q
> > >  x =3D
> > >
> > >      1    2    3    4
> > >      2    4    6    8
> > >      3    6    9   12
> > >      4    8   12   16
> > >
> > >  1
> > >  2
> > >  2
> > >  3
> > >  4
> > >  3
> > >  4
> > >  6
> > >  6
> > >  4
> > >  8
> > >  9
> > >  8
> > >  12
> > >  12
> > >  16
> > >  Elapsed time is 0.177223 seconds.
> > >
> > >
> > >  However I dont know where I should commit it to in Octave-forge.
> > >  Someone please suggest the right place.
> > >
> > >  Code is attached.
> > >
> > >  Cheers
> > >  Muthu
> > >
> > >
> > > _______________________________________________
> > > Octave-sources mailing list
> > > [email protected]
> > > https://www.cae.wisc.edu/mailman/listinfo/octave-sources
> > >
> > >
> > >
> > >
>
>
>

------=_Part_21198_1222623.1160225406791
Content-Type: text/x-objcsrc; name=zagzig.m; charset=ANSI_X3.4-1968
Content-Transfer-Encoding: base64
X-Attachment-Id: f_et007xs2
Content-Disposition: attachment; filename="zagzig.m"
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==
------=_Part_21198_1222623.1160225406791
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
Octave-sources mailing list
[email protected]
https://www.cae.wisc.edu/mailman/listinfo/octave-sources

------=_Part_21198_1222623.1160225406791--