Re: Photo Image out of focus

[brian whatcott <inet@INTELLISYS.NET>]

I looked through undergrad notes more than a generation old,
without finding the photo image correction example that I wanted
to share. So I need to speak from memory, and I will be glad if
what I say can be corrected, or updated with more recent technique.

The approach discussed then, depended on detailed knowledge of the image
distortion introduced by the original image generating method.
 A 2-dimensional array is generated of digital values of the original
image. And a 2-dimensional array is produced
from (say) the distorted image of a point source through the same
defocussed camera.
 The two arrays are convoluted (I say 'convolved') together so as to
 produce an array which evaluates the intensity of a point source
from the spread representation of this point at all positions on
the image. You will recall that the Fourier transform of a point
is an array of infinite extent in two dimensions, so the computational
chore could formerly occupy a mainframe for more than trivial time;
in other words, a fast Pentium would be needed to handle the
computation now.

I probably glossed over the numerical massaging involved somewhat.
Applications to spy satellite imagery remediation are so obvious I
expect some relevant research is not on the open shelf.
 So the answer to the question, 'is there a purely optical method?'
is, 'I don't think so....'

Brian

At 17:18 6/19/01 -0400, you wrote:
> Hi Everyone,
>
> A parent of one of my students sent me the following question:
>
> At Lower School graduation, I took a great photo of a friend of mine
> and his son.  But it's somewhat out of focus.  I want to print it,
> but in focus.  My intuition tells me that all the various light
> rays hit the film -- but not quite in the right place.
> So it seems to me that all the data for a good
> print must be there on the emulsion.  And if I could only bend the
> light rays in the printing process (whether before or after it goes
> through the negative, I'm not sure), it should be possible to
> correct the error and make
> the print come out in focus.  (After all, they fixed the Hubble
> telescope, right?)
>
> I responded as follows:
>
> My first thought is that you are seeking an analog solution to a
> digital problem.
> You're probably right that all the necessary data is there, just
> not in the right place.  And your photography friends may also be
> right that there are no photographic techniques available at the
developer to move those data around.  The easy fluidity of data is
> a digital reality, not an
> analog one or we would have had it sooner, I imagine.
>
>
> Is anyone on this list a knowledgeable photographer who can resolve
> this question one way or the other.  I'm not sufficiently aware of
> the optics of the development process to answer directly.
> Is there an optical solution to the focus problem?
>
> Thanks for any help you can give.
>
> Jeff Weitz
> Physics Teacher
> Horace Mann School
> Riverdale, New York

brian whatcott <inet@intellisys.net>  Altus OK
                Eureka!