Unmasked Janus

In the following URL;

http://www.unmannedspaceflight.com/index.php?act=Attach&type=post&id=1980

the left side is the original image of Janus taken by Cassini probe and
the right
side is the processed image deconvoluted by Focus Corrector.

For detail of Focus Corrector, visit;

http://139.134.5.123/tiddler2/c22508/focus.htm

tontoko skrev:
> In the following URL;
>
> http://www.unmannedspaceflight.com/index.php?act=Attach&type=post&id=1980
>
> the left side is the original image of Janus taken by Cassini probe and
> the right
> side is the processed image deconvoluted by Focus Corrector.
>
> For detail of Focus Corrector, visit;
>
> http://139.134.5.123/tiddler2/c22508/focus.htm

Seems to me that this improvement is due to floating point
arithmetics? The weights 1 and 9 in the integer mask operator
are substituted for 1/9 and 1, which reduces overflow and
quantization problems? 

Rune

Rune Allnor wrote:
> tontoko skrev:
> 
>>In the following URL;
>>
>>http://www.unmannedspaceflight.com/index.php?act=Attach&type=post&id=1980
>>
>>the left side is the original image of Janus taken by Cassini probe and
>>the right
>>side is the processed image deconvoluted by Focus Corrector.
>>
>>For detail of Focus Corrector, visit;
>>
>>http://139.134.5.123/tiddler2/c22508/focus.htm
> 
> 
> Seems to me that this improvement is due to floating point
> arithmetics? The weights 1 and 9 in the integer mask operator
> are substituted for 1/9 and 1, which reduces overflow and
> quantization problems? 
> 
This isn't really focus correction anyway. Its just a crude sharpener. 
Real focus correction means estimating the blurring function - e.g. 
using maximum entropy analysis - and applying the inverse of that to the 
image. The processed Hubble picture of Jupiter's spot, for example, 
gives a simplistic appearance of gerater sharpness. However, it has 
really just peaked everything, creating bogus structure out of nothing.

Steve

Steve Underwood wrote:
> Rune Allnor wrote:
>> tontoko skrev:
>>
>>> In the following URL;
>>>
>>> http://www.unmannedspaceflight.com/index.php?act=Attach&type=post&id=1980 
>>>
>>>
>>> the left side is the original image of Janus taken by Cassini probe and
>>> the right
>>> side is the processed image deconvoluted by Focus Corrector.
>>>
>>> For detail of Focus Corrector, visit;
>>>
>>> http://139.134.5.123/tiddler2/c22508/focus.htm
>>
>>
>> Seems to me that this improvement is due to floating point
>> arithmetics? The weights 1 and 9 in the integer mask operator
>> are substituted for 1/9 and 1, which reduces overflow and
>> quantization problems?
> This isn't really focus correction anyway. Its just a crude sharpener. 
> Real focus correction means estimating the blurring function - e.g. 
> using maximum entropy analysis - and applying the inverse of that to the 
> image. The processed Hubble picture of Jupiter's spot, for example, 
> gives a simplistic appearance of gerater sharpness. However, it has 
> really just peaked everything, creating bogus structure out of nothing.

Moreover, most blurring functions are circular, so the functions to 
properly deconvolve the blur must be circular too. The closest I've seen 
to making something from nothing is deconvolving a diffraction-limited 
Airy disk, including part of the first and second rings. Of course, 
there's nothing to be done for the nulls between the rings, but then 
there's nothing that needs to be done there.

Tontoko is either ignorant or a charlatan. Too bad. (If the latter, 
posting here was an avoidable mistake.)

Jerry
-- 
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������

On Sat, 09 Dec 2006 11:03:51 -0500, Jerry Avins <jya@ieee.org> wrote:

>Steve Underwood wrote:
>> Rune Allnor wrote:
>>> tontoko skrev:
>>>
>>>> In the following URL;
>>>>
>>>> http://www.unmannedspaceflight.com/index.php?act=Attach&type=post&id=1980 
>>>>
>>>>
>>>> the left side is the original image of Janus taken by Cassini probe and
>>>> the right
>>>> side is the processed image deconvoluted by Focus Corrector.
>>>>
>>>> For detail of Focus Corrector, visit;
>>>>
>>>> http://139.134.5.123/tiddler2/c22508/focus.htm
>>>
>>>
>>> Seems to me that this improvement is due to floating point
>>> arithmetics? The weights 1 and 9 in the integer mask operator
>>> are substituted for 1/9 and 1, which reduces overflow and
>>> quantization problems?
>> This isn't really focus correction anyway. Its just a crude sharpener. 
>> Real focus correction means estimating the blurring function - e.g. 
>> using maximum entropy analysis - and applying the inverse of that to the 
>> image. The processed Hubble picture of Jupiter's spot, for example, 
>> gives a simplistic appearance of gerater sharpness. However, it has 
>> really just peaked everything, creating bogus structure out of nothing.
>
>Moreover, most blurring functions are circular, so the functions to 
>properly deconvolve the blur must be circular too. The closest I've seen 
>to making something from nothing is deconvolving a diffraction-limited 
>Airy disk, including part of the first and second rings. Of course, 
>there's nothing to be done for the nulls between the rings, but then 
>there's nothing that needs to be done there.
>
>Tontoko is either ignorant or a charlatan. Too bad. (If the latter, 
>posting here was an avoidable mistake.)
>
>Jerry

Well, he started an interesting discussion, if nothing else!   That's
worthwhile...  ;)

Eric Jacobsen
Minister of Algorithms, Intel Corp.
My opinions may not be Intel's opinions.
http://www.ericjacobsen.org

Intuitively there seems to be no difference between usual convolution
method and my method except the effect caused from the treatment of
floating point, however actual calculation shows definite difference
between them.

Rune Allnor wrote:
> tontoko skrev:
> > In the following URL;
> >
> > http://www.unmannedspaceflight.com/index.php?act=Attach&type=post&id=1980
> >
> > the left side is the original image of Janus taken by Cassini probe and
> > the right
> > side is the processed image deconvoluted by Focus Corrector.
> >
> > For detail of Focus Corrector, visit;
> >
> > http://139.134.5.123/tiddler2/c22508/focus.htm
>
> Seems to me that this improvement is due to floating point
> arithmetics? The weights 1 and 9 in the integer mask operator
> are substituted for 1/9 and 1, which reduces overflow and
> quantization problems? 
> 
> Rune

tontoko skrev:
> Intuitively there seems to be no difference between usual convolution
> method and my method except the effect caused from the treatment of
> floating point, however actual calculation shows definite difference
> between them.

So I am right in my interpretation that this method "only" relies
on floating point implementations of operators that usually are
implemented with fixed-point arithmetics?

These originals you post, are these the "true" originals or are they
compressed or smaller versions than what you process?
Would it be possible for you to post the full-size, uncompressed
image you start from somewhere?

Rune

Rune Allnor wrote:
> tontoko skrev:
> > Intuitively there seems to be no difference between usual convolution
> > method and my method except the effect caused from the treatment of
> > floating point, however actual calculation shows definite difference
> > between them.
>
> So I am right in my interpretation that this method "only" relies
> on floating point implementations of operators that usually are
> implemented with fixed-point arithmetics?
>
> These originals you post, are these the "true" originals or are they
> compressed or smaller versions than what you process?
> Would it be possible for you to post the full-size, uncompressed
> image you start from somewhere?
> 
> Rune

Sorry, for my previous post seems not to have been uploaded correctly.

The original image was a jpg file, and I converted it to bmp image for
Focus Corrector without reducing size (but just used the part of it.)
Regarding the deconvolution of the whole image of Tempel 1, please
refer to the following thread.

http://groups.google.com/group/comp.dsp/browse_thread/thread/cf864b4a3a8454ed/ad2b99168331382c?lnk=st&q=tontoko&rnum=1&hl=en#ad2b99168331382c

Rune Allnor wrote:
> tontoko skrev:
> > Intuitively there seems to be no difference between usual convolution
> > method and my method except the effect caused from the treatment of
> > floating point, however actual calculation shows definite difference
> > between them.
>
> So I am right in my interpretation that this method "only" relies
> on floating point implementations of operators that usually are
> implemented with fixed-point arithmetics?
>
> These originals you post, are these the "true" originals or are they
> compressed or smaller versions than what you process?
> Would it be possible for you to post the full-size, uncompressed
> image you start from somewhere?
> 
> Rune