[Choscura]

Just a few thoughts, I'm curious what people have to say about them.

1) enhancing resolution of an image using multiple still images to form a composite of higher resolution (ie: taking video footage and increasing resolution of moving elements)

it seems as though you could do this relatively quickly by examinging the way the pixels on the edge of a particular feature change as that feature moves about the frame- the most basic version of this would be to detect when a 'vertical' line (eg, the edge of a face) is not actually vertical, but is tilted slightly in or out or to the left or right.

The reason this seems possible is because the human mind already accomplishes this on a daily basis: whenever you watch TV (not hi-def TV, just normal TV), the screen's resolution is technically not great and the detail on the actors or objects on the screen is usually pretty bad, even from cable or DVD. Despite this, most of us know what, eg, Conan OBrien looks like in minute detail, because our brain combines the details of how the screen changes as he moves around on it.

2) enhancing from a single still image

This is more complex, but the basis of the idea is this: if you could take an image and extract the exact values of what the color value of each pixel should ultimately be, you could extrapolate from that and examine what the effect of the actual (starting) values have been modified to.

An easy way to see an example of this would be to open up photoshop, draw a black line anywhere on a white background, and zoom in: in between the space on the edge of the line: in between the black area and the white are gray pixels of varying intensity.

Thanks for your time, let me know what you think.

[Choscura]

Since I'm not sure how well I communicated the second idea for image enhancement, and since I've put a bit more thought into it (no idea if this has already been done), here's a picture to demonstrate a simple version of what I'm thinking of.