IgorCarron (December 31, 1969 at 6:59 pm)

By the way, it looks like the video has been seen more than 1000 times in less than two days. Where is this traffic coming from ?

IgorCarron (December 31, 1969 at 6:59 pm)

Current Coded Aperture Imaging does not use the theoretical framework called Compressed sensing featured in this video. It is only a matter of time before it does though. In short compressed sensing (CS) works when the image is sparse. When that is the case, CS brings new rules on how to design the "coded aperture" and more importantly relies on nonlinear reconstruction techniques that were barely available four years ago.

cdmmem1000 (December 31, 1969 at 6:59 pm)

This is an interesting extension of the range of techniques generally known as coded aperture imaging. These are often used in imaging applications where only very low resolution detectors are used. For example high-energy x-ray and gamma-ray imaging as well as far infrared imaging. Usually these employ a single coded mask and a multiple element (the low resolution) detector. By using a single pixel detector and multiple instances of the mask they achieved the same effect. Fascinating!

PaulyCracker (December 31, 1969 at 6:59 pm)

It's amazing that it can capture other wavelengths like infrared, UV and even low light! I've often though that the compression used today was a limiter. But this technology is amazing. I'd never have thought to use a micro mirror like that. Perhaps, they could even use it for video or night vision in the future!? I hope they keep up the good work, and also make it affordable! This is great!

brettero (December 31, 1969 at 6:59 pm)

It is actually more expansive then a multi pixel cam. It actually do the optical data acquisition using one optical unit (pixel), but then it needs to correlate it with many random patterns, generated by the DLP, which has a lot more then one pixel.The bigger the number of pixels on the DLP, the better maximum final resolution can be achieved. the data storage must hold all the integrated scrambled patterns (for post processing), and should have much more memory then conventional camera.