Hi Prasad,

To create this data first you need to extract some features. You can sea bellow links to understand this process.

http://www.mathworks.com/help/vision/ref/detectsurffeatures.html

http://www.mathworks.com/discovery/face-recognition.html

https://web.engr.illinois.edu/~zzhu27/cs445/project/

and then save your data in matlab  by these codes:

eye=[A];

nose=[B];

lips=[C];

saving='data .mat'

save(saving,'eye','nose','lips')

best regards,

Thank you Mr. Antonios and Mr. Moslem,

when i am using MATLAB surf feature extraction function it returns too many points, but i need specific features from face such as eyes (for knowing wether eyes are opened or closed) and similarly to know mouth opened or closed. please tell me how to do this?

Usually SURF is used for recognition-based applications.

Regarding to your explanation, you want to know if the eye/mouth is close or open etc, then I would recommend you to not use SURF.

You can easily find the position of your region of interest by using the geometry of the face. An example can be found here:  http://www.mathworks.com/matlabcentral/fileexchange/25056-fast-eyetracking/content/getFeatures.m

I did't use that, but seems working fine.

However, in case if you really want to use SURF, then after extracting the region of interest, you can simply apply SURF on these regions. For instance the sub image contains eyes etc.

Remember, as the other people mentioned, SURF will give you a feature vector for every keypoint (or blob). Thus depending on how many blob is detected, the size of the feature vector might changes. You just can control the input of the SURF function not the position of the blob, hence if you are interested to have SURF feature of eyes, you may have better result by setting the sub image of eyes as the input of SURF. You also need to have dataset of open/close eyes in order to classify a new sample!

Hope it helps.