extract exact match of image from dataset

edit

Can you describe more about your Input images , methods you tried & observed results?

If you have a single object that you want to detect (and not all object of that kind) you can use this tutorial for it. If you want to detect a type of object (like faces) you should use cascades (and then you need a db of images, negatives and positives, for training)

i am having real part of bracket of vehicle and I captured images of that from different views and angles(side,top,bottom,front,different angles).So whenever I put that bracket in front of the camera (suppose front view in front of camera) exact (i.e front view)image should be extracted from the set of images.so how to do this?

I am not sure what you mean by "bracket", but want you can do is using the template matching or features matching + homography

image I am having the same real part with me and its images from different views(images are captured by camera).

So, what you are trying to say is that your images have not the same small problems (like craters, etc)? And what you want to do is to detect if there is the same problem in more images?

no.Actually I am having same real part and data set (of images of same real part captured from different views like top view,bottom view etc ) so whenever i put real part in front of camera(in particular view) then my search engine should find the the exact image of that view to match.

oh, so you have a set of images and you want to detect what image is in the camera?

1. One way to go is to do a simple diff on images and pick the one with the less differences. (this may be improved with smoothing and histogram equalizations, maybe even CLAHE for doing contrast corrections)
2. Another one is to do the feature matching and select the best match as the one that has the most features matched.