Can you enhance it?

One of the staples of a Hollywood movie involving law enforcement or something along those lines is a scene involving a bunch of people crowded around a computer monitor, looking at some kind of image - for example, a picture of a car racing away into the distance.

Some guy tells the chap with the keyboard - "Can you enhance/zoom in on this?" - pointing at the license plate.

Now at this point, some weird stuff happens. What we would expect is that the guy will draw a little rectangle around the area and blow it up. Instead, you hear some furious typing sounds, and a large rectangle appears, and shrinks down until it fits perfectly over the thing the first dude pointed out. Then it expands, usually accompanied by more furious typing.

My question is - What the hell is the guy typing? How on earth did he tell this miracle application where the area of interest was? In 99% of the flicks I've seen, there's no sign of a grid or any other form of coordinate system. He isn't even moving it with the keyboard - it just sort of appears at some random point on the screen and then helpfully shrinks down to just the right spot. Weird...

Even more interesting is that no one ever seems to say "Sure I can zoom in, but it's no good, because the camera never captured that detail in the first place. You'd just see a weird blurry hodgepodge of colours." Seriously, I doubt that even a really good interpolation algorithm is going to be able to make sense of the little squiggles that constitute the contents of a license plate.

And of course, there's that thing where they zoom into some tiny reflective surface and see what was going on reflected there...

What kind of magic algorithm do these chaps use to enhance everything?

Oh, and the reason this blog post turned up is because Image Processing is the next exam on Thursday. Off I go now to grok the frequency domain and image enhancement using the Fourier transform.