Tuesday, January 24, 2006

Is it fake? Is it real? Mathematics and digital images.

There's a very interesting article in todays New York Times about scientists and image editing software. Many of the most striking images, such as astronomy photographs, must be digitally enhanced even to be intelligible. But in the Photoshop world, it is very easy to beautify digital images beyond this in a way that leaves no visible trace-- in short, to fudge. However, since Photoshop is essentially a collection of math algorithms, the manipulations-- rotations, rubber stamps, contrast gradients-- can be detected by their mathematical signatures. The editors of the Journal of Cell Biology have been able to detect these little fudges, such as cutting out a background band or selective contrast enhancement. Even as outright deception by Photoshop appears to be very rare, fudging is suprisingly common. My brand of science is very visually driven so a great deal of attention, benign and otherwise, is lavished on the images.

A tangentially related problem was covered in a recent Slashdot thread, which describes researchers using digital analysis to authenticate a Rembrandt. Jumping the link to ZDNET blog , a quote from Dan Rockmore says it all:

"The fact that you can put everything on the computer means that everything is numbers," Rockmore says. "As soon as everything is numbers, it makes perfect sense to ask mathematical questions about what the numbers represent." If he's right — if computers can distinguish between artists more accurately than connoisseurs can — the art world is in for some high-stakes corrections."

The ZDnet article is really fascinating, and shows how this method was used to identify four different artists' hands in a large Renaissance painting. ZDnet also cites some pushback by art historians- for example, a Rembrandt might have eight layers of paint, so that a surface analysis might miss underlying information. I also know that in Renaissance workshops the master might put the finishing touches on a face blocked in by an apprentice. This mixed effort should look very confusing at the mathematical level.

UPDATE: But the final picture is so compelling! (Can't decide if this better suits scientists or Rembrandt.)

No comments: