Path: news.mathworks.com!not-for-mail
From: <HIDDEN>
Newsgroups: comp.soft-sys.matlab
Subject: Re: shape &  texture extraction from face image
Date: Fri, 10 Oct 2008 12:39:02 +0000 (UTC)
Organization: STFC Rutherford Appleton Laboratory
Lines: 19
Message-ID: <gcnid6$d9f$1@fred.mathworks.com>
References: <gcnf0d$7tn$1@fred.mathworks.com>
Reply-To: <HIDDEN>
NNTP-Posting-Host: webapp-05-blr.mathworks.com
Content-Type: text/plain; charset="ISO-8859-1"
Content-Transfer-Encoding: 8bit
X-Trace: fred.mathworks.com 1223642342 13615 172.30.248.35 (10 Oct 2008 12:39:02 GMT)
X-Complaints-To: news@mathworks.com
NNTP-Posting-Date: Fri, 10 Oct 2008 12:39:02 +0000 (UTC)
X-Newsreader: MATLAB Central Newsreader 968489
Xref: news.mathworks.com comp.soft-sys.matlab:494551


"Nive S" <xyz@yahoo.com> wrote in message <gcnf0d$7tn$1@fred.mathworks.com>...
> Hi
> I m doin project in age estimation in matlab. I m extracting features shape and texture from face image, can anyone have idea of doing this if so , please help me moreover i m new to this area and matlab.

Something I have never tried - so this will need some research on your part.

1) Switch your colour space to normalized Red, human skin is relatively racially invariant in this space.

2) Use a threshold value to isolate the face - you will need to experiment to find the optimum value, from this form a binary image (or mask) of the facial position, and extract a greyscale image of the original face.

3) Look up Law's texture analysis on the web, to undertake a texture estimation on the web.

4) Alternatively investigate the use of Wavelets, I would imagine that the 'wrinkles' on an old persons face would provide more information in the fine scale regions than would a smooth skin of a young person - of course you may get thrown by beards etc.

5) Investigate neural nets for processing the texture/wavelet data for age classification.

Hope that helps

Dave Robinson