From: Alan B on 13 Jul 2010 13:17 "fraisa1985 Youssef" <meguebli(a)gmail.com> wrote in message <i18bfn$pnj$1(a)fred.mathworks.com>... > "Alan B" <monguin61REM(a)OVETHIS.yahoo.com> wrote in message <i186vb$u4$1(a)fred.mathworks.com>... > > "Matt J " <mattjacREMOVE(a)THISieee.spam> wrote in message <i181bc$ohg$1(a)fred.mathworks.com>... > > > "fraisa1985 Youssef" <meguebli(a)gmail.com> wrote in message <i18024$89e$1(a)fred.mathworks.com>... > > > > Hello, > > > > I'm working on an 3D face recognition problem. > > > > Each face is characterized with a 3D curve 3D, so i have to compare theses 3D curve. > > > > To be invariant of the origin point i would use Fourier descriptor to compare theses 3D curves. > > > ============== > > > > > > I doubt there will be many responses seeing as you've haven't described where in your problem you've gotten stuck. > > > > > > Nevertheless, I don't know how you'd define Fourier descriptors for 3D curves. I thought it possible only in 2D, so I'd be interested in any feedback you get. > > > > I'm not too familiar with Fourier descriptors, but what if you just found the descriptors for two pairs of your coordinates (ie, the x-y descriptors, plus the y-z descriptors)? Or the descriptors for the curvature and torsion of the curve, with respect to arc length? > Thank you for your answers, it good idea is to just found the descriptors for two pairs of my coordinates, have you an idea how i can implement him please ? > Also can you explain more the descriptors for the curvature and torsion of the curve, with respect to arc length? > Best regards Like I said, I'm not too familiar with Fourier descriptors, but I think fft() does all the work..? As far as curvature and torsion, this might be a good place to start: http://en.wikipedia.org/wiki/Frenet_frame |