Well , I don't wanna boast about my newest & so to say the biggest achievement of life so far, but I'm so glad I made it. I waited 4 straight years to earn this right and I doubt even god knows how much effort I put into it. Gosh… I never thought I’d have to kill myself for the Curtin degree like I did. SLIIT degree program was pretty easy going, apart from the final year research project. Anyway, I had a much needed break of couple of months after a 10 month marathon run this year. The convocation for SLIIT is due to be held on 21st Feb next year where as the Curtin one too could well be on the same day though it’s not been confirmed yet. I’m in the mood of doing an Msc. also, but I’ve got a lot to do before that. Now, don’t bother asking what they are …(hihi lolz…………)
Sunday, December 23, 2007
Subscribe to:
Post Comments (Atom)
3 comments:
Wish you all the best man..! and wishing you bright future!
hi. i'm a 4th year student at SLIIT. Gayan ayya gave me ur blog site, becoz he said u can help with the project we are doing for 4th year. we're doing a project that has to do with image processing. we need to compare images. one image is in a database. when the user uploads an image of the person he wants details about, we have to search the database and return the details of the individual. can this be done using MATLAB? can u help us??? thanks
Hi dulani, Sorry I just saw ur comment.. Anywayz, it seems to me that u are doing a facial recognition project.. well , we did a similar project last year for human recognition with iris images. Basically the concept is prety much the same. As we did, we first isolated the iris area of the eye (isolation), segmented the iris area to small parts(segmenation), extracted the unique features and stored them according to a predefined format (encoding) in a template. When a user image is inserted, that image goes through the same process again and is matched with the existing templates (matching). Those are the 4 processes that need to be implemented in a biometric system. We made use of two mechanisms for the matching process. One was called hamming distance calculation and other one was a neural network (NN).
I suggest that it would be better if u could go for a NN as MATLAB itself has a very high level neural network toolbar. there are few types of NNs such as linear,radial etc. , but as i feel, linear type would be sufficient for ur requirement.. If u r doing IML or AI in this semester, probably you can get a good idea about how an NN works.. It's no big deal even if you don't do them. You have to have few images (take @ least 5 images) as the training set which are used to train the NN . then have some for the testing set which can be used to test the neural network...
we used a database only to store the userinfo, and the features we extracted were stored in .mat files(a MATLAB supported format) as it's easy to compare the templates with MATLAB. Of course , u can store them anyway u want. no probs.
Let me give u few hints before u move on with ur proj.
1.Read the existing literature. i.e do bit of googling and find out how existing facial recognition systems(implemented using MATLAB)work.
2.Focus more on ways to extract unique features from the face like its shape, wrinkles, etc..
3.Try to use 8bit grayscale images as they can speed up the process dramatically.
4.Meet Dr. Chandimal Jayawardene with a prior appointment via email (chandimal@sliit.lk) coz he's an expert on MATLAB and AI & NN. Make sure you read the relevant literature and have some idea about what your project really is, before meeting him.
Hope these information will be helpful to you somewhat...
good luck !!!!!!!
Post a Comment