[olug] Document Image Type Recognition

[Todd Christopher Hamilton]

I am not finished reading the ImageSeek web site but I had a thought:
I might be able to accomplish my goal by utilizing Imagemagick. The
Imageseek web site refers to using a "multi resolution wavelet
decomposition of the query and database images".  If I knew the
algorithm maybe I could just do it.  Maybe I use imagemagick to create
a series of decomposition images, perform an analysis of the image and
then use some mathematical equation to some up with a "looks like"
result.

On 11/3/06, Nick Veys <psylence519 at gmail.com> wrote:
> Maybe give this a try? http://www.imgseek.net/
>
> Otherwise, I imagine UNO has a few Neural Net students itching for
> final projects. :)
>
> On 11/3/06, Todd Christopher Hamilton <netarttodd at gmail.com> wrote:
> > I am looking for an application that can "look" at a scanned image and
> > determine what class/category/type the document is based on general
> > layout/look compared to other document images. For instance the system
> > looks at an image and says "bases on the document image templates I
> > have stored in a certain directory, this specific looks like a Bergan
> > Mercy Patient Discharge Summary" or "right this looks like a Methodist
> > Hospital History and Physical"
> >
> > It might be called like this:
> >   docrec --template_directory=/var/docrec/templates/
> > --confidence_threshold=80 image1.tiff
> >
> > It might return this:
> >   file_name: image1.tiff
> >   doc_type: Bergan Mercy Patient Discharge Summary
> >   confidence: 83.5
> >
> > Can anyone point me in the right direction?
> > Todd
> > _______________________________________________
> > OLUG mailing list
> > OLUG at olug.org
> > http://lists.olug.org/mailman/listinfo/olug
> >
> _______________________________________________
> OLUG mailing list
> OLUG at olug.org
> http://lists.olug.org/mailman/listinfo/olug
>


-- 
Todd Christopher Hamilton
(402) 660-2787