Creating a feature vector

I have calculated texture, color and shape features on an image.But those just add up to 12 features. I read that people extract 1000 features and such. Could someone please explain to me how do i increase the number of features? And then how do i save them to form a feature vector?

[Laurent]

It depends on which features you are interested in, for example mean, std, max intensity, min intensity, and so on. Obviously there are ways to extract 1000s of such features from images (or objects in these images), but it doesn't make much sense to do so without knowing the purpose.

Answer 1

[Image Analyst]

Nobody says you need to have thousands of features. In fact most times there aren't that many needed. You only need as many as you truly need, not more. Maybe you can get by with 1 or 2. Can you attach your image and say what you are trying to do with it?

[sidra]

@Image Analyst : I need to extract features for face recognition from a standard face image. And for face recognition/object recognition many features are used to describe the image right? So how do i increase the number of features? Should i take color/texture/shape features of each pixel rather than the whole image? And if yes, how do i do it?

And could you tell me how do i save the features to form a feature vector?

[Image Analyst]

I don't work in face recognition so you'll have to go here to see what kinds of feature vectors people are building: http://iris.usc.edu/Vision-Notes/bibliography/contentspeople.html#Face%20Recognition,%20Detection,%20Tracking,%20Gesture%20Recognition,%20Fingerprints,%20Biometrics You can make a feature vector simply by stringing together a bunch of numbers into a vector, where each number represents one of the measurements you have made, for example the skin color, the width of the mouth, the spacing between the eyes, etc.

[sidra]

@Image Analyst : Thank you, the link was really helpful.