chm++ This might be a little off topic, but in the area of fuzzy shape recognition, there's a technique called Fourier Descriptors. The technique turns a shape's boundary into a Fourier series; you can then compare the series to that of the desired shape. The "agreement" of two shapes is wonderfully intuitive: if two shapes agree only very roughly (hand-drawn sketch vs smooth lines), the low "frequencies" should have the same values. Two shapes that agree more precisely will have nearly the same spectra, even into the higher "frequencies"
I came across these a couple of years ago and thought it would be cool to write a PDL module for this idea, but never got around to it. I don't do any image analysis, so this would have been a bit off topic for me. Anybody interested in shape recognition should consider working on this. It would be really cool! David On Fri, Jun 27, 2014 at 5:18 PM, Chris Marshall <[email protected]> wrote: > Good luck. I've attached code for a sobel edge detection > routine that may be of use. > > --Chris > > On Fri, Jun 27, 2014 at 4:43 PM, Elżbieta Jakubska > <[email protected]> wrote: > > Thank you guys for your reply! I will try to face this:) > > Ela > > > > 27 cze 2014 22:22 "Craig DeForest" <[email protected]> > napisał(a): > > > >> Heh. Thanks for answering this, Chris. I missed it the first time > >> 'round. > >> > >> Elzbieta, you can find whether a particular pattern is present in an > image > >> by convolving the image with the pattern you're looking for. The > resulting > >> convolved image will have a spike everywhere that the pattern is > present. > >> One PDL function you can use to do that is convolveND(). > >> > >> > >> On Jun 27, 2014, at 6:24 AM, Chris Marshall <[email protected]> > >> wrote: > >> > >> > Perl and PDL can be used for image processing. > >> > Probably the easiest way to familiarize yourself > >> > with PDL capabilities is to install PDL and then > >> > look at the PDL Book. You can find this and other > >> > PDL information starting at the PDL home page: > >> > > >> > http://pdl.perl.org > >> > > >> > --Chris > >> > > >> > On Tue, Jun 24, 2014 at 3:03 AM, Elżbieta Jakubska > >> > <[email protected]> wrote: > >> >> I need to compare many images to pattern image (logo) and find out > >> >> which of > >> >> them match the pattern (shape is more meaningful than colour). Logo > has > >> >> the > >> >> same size, the same rotation etc on every picture, but backgrounds > are > >> >> different, including additional edges in interesting area. Till now I > >> >> was > >> >> doing this with Image Magick (edge detection + compare), but it is > slow > >> >> and > >> >> not always works correctly. > >> >> Is there a way to do this using PDL? Could anybody help me? > >> >> > >> >> Thank you very much! > >> >> > >> >> Ela > >> >> > >> >> _______________________________________________ > >> >> Perldl mailing list > >> >> [email protected] > >> >> http://mailman.jach.hawaii.edu/mailman/listinfo/perldl > >> >> > >> > > >> > _______________________________________________ > >> > Perldl mailing list > >> > [email protected] > >> > http://mailman.jach.hawaii.edu/mailman/listinfo/perldl > >> > > >> > > > > _______________________________________________ > Perldl mailing list > [email protected] > http://mailman.jach.hawaii.edu/mailman/listinfo/perldl > > -- "Debugging is twice as hard as writing the code in the first place. Therefore, if you write the code as cleverly as possible, you are, by definition, not smart enough to debug it." -- Brian Kernighan
_______________________________________________ Perldl mailing list [email protected] http://mailman.jach.hawaii.edu/mailman/listinfo/perldl
