Quote:
Originally posted by mitheral
To a certain degree. But in the movies it's almost always some single frame, grainy, video survalince image that they blow up 10,000X and then read the guys wristwatch or something. The information they are supposedly sharpening only exists as a couple pixels in the original at best.
|
Very true, but by studying how those pixels change over dozens to hundreds of frames, a great deal of information can be extracted. Unfortunately, I can't offer any solid citations, but I remember seeing a demonstration on the Discovery Channel where they took images from a conveinance store security camera that recorded a murder-robbery, and extrapolated enough information to clarify his tattoo, which led to identification and conviction.
Let's say you have 1024 shades of gray, (just a guess on how many shades you could extract from a tape) and an 8x8 grid of pixels to work with for a feature. 640 bits of information. If you have the area on camera and moving for 4 seconds, thats around 8 kilobytes of infomation you have, or an 80*80 grid of 10 bit grey scale, which is significantly more detail. Now, i'm not saying you can necessarily get that much detail, but that is how much information you have to work with.
The theory, i believe, is that a detailed picture will light up different pixels in different ways as it moves up in a frame, as compared to left or right, or down, or as the angle of the surface changes, so that several low resolution representations can be processed into one higher resolution representation.
I'd like to say that it's entirely possible I'm wrong, because it's been years since I've seen this show, but as a graduating engineer, the theory seems sound to me. I'll send it up for peer review
.
Here's a relevant NASA press release:
http://www.jsc.nasa.gov/news/releases/2002/J02-81.html Best I can do right now
and another with some pictures:
http://science.nasa.gov/headlines/y2000/ast31aug_1.htm