On 15 Giu, 16:32, "Peter Bone" <peterb...@hotmail.com> wrote: > > Thank's a lot Pete, > > well my project is to make facedetectionfrom images, > > I can suppose that the face orientation is fixed, I suppose > > that the axes which describe the vertical semimajor axe is longer > > than > > horizontal axe. > > I'm tryin to modify that implementation of Hough Transform > > > <http://basic-eng.blogspot.com/2006/01/hough-transform-for-circle-dete...> > > > but i can't understand how modify it. > > I understand that i fix for any iteration any radius value and > > horizontal axe. > > > If you have some code example or pseudo-code algorithm and you can > > send me it > > i'm really glad to You. > > > Excuse for my english. > > > Thank you > > EoD > > Here's my code for circledetection > <http://www.mathworks.com/matlabcentral/fileexchange/loadFile.do?objec...> > > It's quite complex but highly optimized. You may be better off using > one of the simpler but slower implementations in order to convert it > forellipsedetection. > The basic idea is to draw a circle/ellipsearound each edge point in > a voting space. So you just need to use the equation of anellipse > instead of the equation of a circle. I guess the eccentricities will > change slightly so you'll have to repeat yourdetectionover a range > of different eccentricities and sizes. I don't have any code for > this. > Is a face really close enough to anellipseto be detected? What > about hair, etc? > > Pete
Hi all i've released one algorithm for ellipse detection this is not optimized, anyone want to take a look of my code to give me some advice to optimize my code?
Thanks all for your supports and helpful advices :D IG