wavelet decompose

[Rainer M. Engel]

I'm searching for something like this but for ImageJ.
http://registry.gimp.org/node/11742

Thanks for your help.

Regards,
Rainer

--
Rainer M. Engel, Dipl. Digital Artist
scientific|Media GbR
Pichelsdorfer Str. 143
13595 Berlin

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Michael Schmid]

Hi Rainer,

there is an alternative to wavelets: Simply do a bandpass filter in the Fourier transform.

With Process>FFT>Bandpass filter (suppress stripes=none, all autoscaling off) I have tried decomposing an image into three bands, 0-3, 3-40 and 40-400000 wavelength bands.
(for a small image, 400000 is close to infinity; I think that the program does not accept infinity here).

Then I have added these three contributions. Apart from a constant offset (which means one should to do the calculations on float images), the result was very close to the original (0.3% of the maximum pixel value or so).

Michael
________________________________________________________________
On Nov 7, 2013, at 13:54, Rainer M. Engel wrote:

> I'm searching for something like this but for ImageJ.
> http://registry.gimp.org/node/11742
>
> Thanks for your help.
>
> Regards,
> Rainer
>
> --
> Rainer M. Engel, Dipl. Digital Artist
> scientific|Media GbR
> Pichelsdorfer Str. 143
> 13595 Berlin
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Rainer M. Engel]

Dear Michael,

thank you for this advice. I made a quick test myself and it works as
described. The only thing that would remain, is making it truely
lossless but it's very close, as you wrote.

Best regards,
Rainer


Am 07.11.2013 17:48, schrieb Michael Schmid:

> Hi Rainer,
>
> there is an alternative to wavelets: Simply do a bandpass filter in the Fourier transform.
>
> With Process>FFT>Bandpass filter (suppress stripes=none, all autoscaling off) I have tried decomposing an image into three bands, 0-3, 3-40 and 40-400000 wavelength bands.
> (for a small image, 400000 is close to infinity; I think that the program does not accept infinity here).
>
> Then I have added these three contributions. Apart from a constant offset (which means one should to do the calculations on float images), the result was very close to the original (0.3% of the maximum pixel value or so).
>
> Michael
> ________________________________________________________________
> On Nov 7, 2013, at 13:54, Rainer M. Engel wrote:
>
>> I'm searching for something like this but for ImageJ.
>> http://registry.gimp.org/node/11742
>>
>> Thanks for your help.
>>
>> Regards,
>> Rainer
>>
>> --
>> Rainer M. Engel, Dipl. Digital Artist
>> scientific|Media GbR
>> Pichelsdorfer Str. 143
>> 13595 Berlin
>>
>> --
>> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Adrian Daerr-2]

On 11/07/2013 05:48 PM, Michael Schmid wrote:
> Hi Rainer,
>
> there is an alternative to wavelets: Simply do a bandpass filter in the Fourier transform.

Hi,

Of course this is a real alternative only if Rainer wants to filter
exclusively in the frequency domain. The nice thing about wavelets is to
represent data resolved both in the spatial and in the frequency domain.
Of course one can window in the spatial domain before doing a Fourier
transform, and finally window in the frequency domain, and do it
properly so that the transformation can be inverted, but then one is
doing ... wavelets.
If Rainer is just interested in the frequency filtering aspect the FT
with bandpass is of course a good solution.

Adrian

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Robert Lockwood]

Some years ago I was able to use wavelets to achieve lossless compression
for our images.  But that was using a proprietary software at $400 per
licence one of which was required for each copy of my software that I
distributed! That turned out to be an unworkable solution.

We need lossless compression in order to do physics on our data.

I got really good and quick compression on 16 bit unsigned data, though, as
my files were reduced to 12% to about 25% of the original compared to the
approximately 50% for the TIFF files I use with ImageJ.  This is important
to us since we expect to be charged about $10 per MB to transmit our data
via a SATCOM service.

I have not checked the OpenJPG group lately but think that someone may have
a Java interface for their JPEG2000 C code.

Nate


On Fri, Nov 8, 2013 at 7:46 AM, Adrian Daerr <
[hidden email]> wrote:

> On 11/07/2013 05:48 PM, Michael Schmid wrote:
>
>> Hi Rainer,
>>
>> there is an alternative to wavelets: Simply do a bandpass filter in the
>> Fourier transform.
>>
>
> Hi,
>
> Of course this is a real alternative only if Rainer wants to filter
> exclusively in the frequency domain. The nice thing about wavelets is to
> represent data resolved both in the spatial and in the frequency domain. Of
> course one can window in the spatial domain before doing a Fourier
> transform, and finally window in the frequency domain, and do it properly
> so that the transformation can be inverted, but then one is doing ...
> wavelets.
> If Rainer is just interested in the frequency filtering aspect the FT with
> bandpass is of course a good solution.
>
> Adrian
>
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

--
When I was 12 I thought I would live forever.
So far, so good.

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html