Help needed to impletement face recognition system

[Rassall Jubair]

Hello Everyone,

 

Im trying to do build up a complete face recognition system.It's my a research basis project,I have divided whole project into two major part.

 

part 1:(Detection part) Using OpenCV  face detect from webcam.(its already succesfully implemented in Microsoft Visual Studio with haarcascade algorithm).

 

part 2: (Recognition part) Now I want to develope a face recognition system with using this detection part.I want to do this way by following:

 

a. Webcam detect the face (Detection part @ already implemented)

b.Detecting face then goes to  compare state by previously store face model for recongnize. I think for this i have to use a database also PCA algorithm.

c.If detected face is recongnized/found into database face then program shows the name of this person in the detect area..

 

that's it...

 

i think this is already developed...

 

but i want to develope by myself..I m using Visual Studio C++,OpenCV..

 

So, Now the problem is, how i go forward by recognition part?i try to study PCA,OpenCV..so anyhelp/comments is most welcome..

 

The expert guys please suggest me  which way/procedure i can implement the recognition part(part 2).

 

 

Thanks

Rassall

[sarker asish]

Dear friend,
your project is almost same with us.we recognize face using OpenCV
technique.we have developed it using java.But the performance is not
so good. so we want to develop it using neural network technique. u
can go ahead.

thanks
asish

[David Bolme]

I have built a similar system called FaceL ( see http://pyvision.sourceforge.net/facel ).  I also have a list of other systems on that website. I think that PCA is a good way to start but you may find that it does not work very well for this application.  

Here are some tips for making PCA work well....  You will need to pay a lot of attention to how you preprocess the image before recognition.   Make sure that you do some sort of eye detection.  I think that the cascade classifier has eye detection cascades.  Use the eye coordinates to align the images.  Also make sure that you normalize the pixels values before PCA.  I also found that high pass filtering helps to remove much of the lighting variation.

Dave