The murmur of the snarkmatrix…

August § The Common Test / 2016-02-16 21:04:46
Robin § Unforgotten / 2016-01-08 21:19:16
MsFitNZ § Towards A Theory of Secondary Literacy / 2015-11-03 21:23:21
Jon Schultz § Bless the toolmakers / 2015-05-04 18:39:56
Jon Schultz § Bless the toolmakers / 2015-05-04 16:32:50
Matt § A leaky rocketship / 2014-11-05 01:49:12
Greg Linch § A leaky rocketship / 2014-11-04 18:05:52
Robin § A leaky rocketship / 2014-11-04 05:11:02
P. Renaud § A leaky rocketship / 2014-11-04 04:13:09
Jay H § Matching cuts / 2014-10-02 02:41:13

Disguise detection

Here’s a puzzle for you. This is a picture of a person wearing one of those creepy super-detailed silicone face masks:

Now, if you point a camera at this guy and pipe the feed into a face-recognition algorithm, it will say, yep! That’s a face! But what if you don’t want it to? What if you want to be able to differentiate between real faces and fake ones? How would you do it? I mean, those masks are pretty good.

The solution — and code to implement it (!) — is right here.

(Don’t miss the fairly surreal YouTube video at the end. That is pure 2013 right there.)

One comment

Very nice! And it’s good that this implementation uses some domain specific knowledge to zero in on far infrared to detect thermal anomalies. But this also demonstrates that hyperspectral imaging is useful for these applications (and many more) in general: there are many other wavelengths where a mask would be easily distinguishable from a real face. In fact, a high (spectral) resolution imager should be able to quickly identify the true material of many potentially fake objects. Other applications out there include detecting plant health.

IMEC has recently started trying to put CMOS hyperspectral imaging chips on the shelves. I would love to see future iterations of Google Glass and similar products move towards these higher spectral resolution imagers. There’s a vast universe of information out there in the form of EM radiation that the paltry human 3-channel broad spectral absorption visual system is just missing. “Ok glass, infrared mode.” “Okay glass, tell my whether this plastic is recyclable.” Etc.

There is an even more general point here: CMOS imaging technology for digital cameras has produced a major surplus of pixels. No one needs a 100 megapixel camera, but we now have chips that do this. One application for all these extra pixels is increasing the spectral resolution as IMEC is doing. Another possibility is using extra pixels to differentiate the wavepath of impinging light, as the Lytro cameras do. Another possibility would be full polarization detection, which is another world of information that our visual system is blind to.

The snarkmatrix awaits you

Below, you can use basic HTML tags and/or Markdown syntax.