is it possible to extract words from given text using camera focus ?

Is it your own idea, or did you saw some implementation of it? The idea sounds pretty questionable to me, but I think something similar might work. In all cases, this is still OCR.
—SA

It can be implemented, lots of apps do this already, there are apps that translate menus, signs etc by holding the video camera up to them and the text is translated into another language on top of the image in the viewfinder. There are also lots of augmented reality apps that can detect environmental aspects of a picture. I've no idea how they work though, it'll be a fairly specialised field of maths and image processing, neural networks etc. OCR and "augmented reality" are the fields you want to be studying though, and I'm sure doing this from scratch is going to be far from trivial.

Agree. The questionable part is "focus". If Sānyà Sani means camera auto-focus, one needs more refined control: moving in steps, allowing the focus to settle and getting the confirmation. That is, camera features have to be integrated with motion system and OCR. Alternatively, auto-focusing could be implemented as the part of the software application, not in camera. This is all not simple at all. But yes, it can be ultimately implemented; it's just a lot of problems to be solved. If focusing is required, it means that the uniform distance to the object is not maintained. In turn, it means that alignment and scaling becomes other problems.
—SA