"Shimmer Vision" Scopes See Better Using Heat

holy_calamity writes "New Scientist reports on a neat DARPA idea that uses the shimmer of heat haze to allow binoculars to see further. It works by exploiting the fact that some distortions from heat haze actually magnify objects behind them. The binoculars collect a series of frames when that is occurring to boost magnification by 3 times. The design goal is to be able to present one image a second, and to enable facial recognition at 90% accuracy at a distance of 1 km. The scopes could be on the battlefield inside of 3 years."

Comments

On the Battlefield

[Iamthecheese]

While military technology has been one of the primary leaders of general technology for thousands of years, it would be nice if there could be more non-military leaps.

Could this technique be used for general astronamy as well, making use of temporary increases in gravitational lensing? I know that gravitational lensing is being made use of, but I bet there are fluctuations that have, until now, been seen only as a limitation.

Re:On the Battlefield

[NeoSkink]

Clusters aren't the only things that lens. Anything with mass can do it. We've even observed planets lensing stars they're orbiting.

See for example: http://en.wikipedia.org/wiki/Image:Gravitational.Microlensing.Light.Curve.OGLE-2005-BLG-006.png

The scale on the x axis is days, just to give you an idea of the time scale some of these events happen on.

Re:On the Battlefield

[j. andrew rogers]

The advantage military technology has is that it is results oriented. In terms of capability they know where they are today and they can specify in considerable detail exactly where they want to be. In the case of organizations like DARPA, they give considerable latitude to the designers in how they solve these problems and the US DoD is relatively patient when it comes to ultimately getting the results they were looking for. There are qualities reflected here that are absent in many other sectors that have little to do with military research per se. In fact, these qualities are not intrinsic to military research at all, so I would say it reflects favorably on the R&D culture that the US DoD has fostered that so many interesting "blue sky" research projects get funded that more conservative private sector institutions would never consider.

There is still plenty of basic science and technology research that gets done outside of military research circles, but military research has the advantage that they are working toward a specific result or technology, even when working on "blue sky" projects. I suspect that focus on specific high-level results combined with wide latitude in design and patience in delivery breeds a very productive research environment relative to those with less critical or obvious goal structures.

huh?

[Anonymous Coward]

"The scopes could be on the battlefield inside of 3 years."

"and to enable facial recognition at 90% accuracy at a distance of 1 km."

Who cares. Marines have been killing folks at 1 km for a long time using scopes about the same size. Who needs facial recognition when you already know the target?

Used in Telescopes as well, I think...

[Iftekhar25]

There was an article on the BBC about a similar method like this this last year.

A new method of looking at stars in the sky through cloud cover; it actually takes several pictures, and combines the best parts of each picture to form one clear picture. Allows telescopes to increase their sharpness many fold. The professor in the news story actually gives an example of a heat haze, coincidentally enough (or not)!

But this looks like a step up from what's in that article. They're taking the best magnified parts of the picture.

1 FPS scope?

[supernova_hq]

The design goal is to be able to present one image a second, and to enable facial recognition at 90% accuracy at a distance of 1 km. The scopes could be on the battlefield inside of 3 years."

Nothin' like sniping a long-range moving target with a full second of lag!

IS already available?

[toQDuj]

I thought one of the first obvious things to implement is Image Stabilisation. I find that one of the biggest drawbacks of binoculars is that the image shakes so much at high "magnifications".

Re:IS already available?

[Anonymous Coward]

Optical image stabilization is an old hat, even $100 digital cameras have it these days!

Currently, even the best image stabilizers can stabilize at most just 4 stops worth (16x more light to sensor), and are rather useless if there's movement in the scene itself. This method might be able to go far further than that.

This would be in addition to traditional optical image stabilization methods, making the stabilization "smarter", picking best features from each captured image and combining them automatically to construct a superior, far more detailed composite.

I guess what makes this new is compensation for atmospheric effects, and integrating that with super-resolution type filtering. Super-resolution methods are pretty old, see Wikipedia on super-resolution. If you have good lighting, you can literally exceed sensor and/or optical resolution by using the information from invidual frame aliasing.

It's nice to see research in this field, it'll also help consumer applications like digital still and video cameras. Imagine a camera that will integrate best features out of 100s shots in low light while also eliminating further camera shake, all in one button press without you even knowing about it! A video camera could further record motion vectors and use them together with the composite image to calculate way superior, noise free video. In good light you could have video/still camera that will exceed its resolution. It's worth noticing that this type of algorithms require frequently uncompressed data, because typical compression methods destroy this subtle information. That's why it should happen in the imaging device itself.

Re:That explains it...

[n3tcat]

there are many types of post-processing filters that could theoretically be run on a live feed to make it easier to pick details out of it. hell there might even be different filters that would go better for watching birds as opposed to tracking isuzu trucks out in the middle of the desert. Being able to cycle through those would be another useful feature. or being able to digitally zoom an image by studying the image via the natural shaking of your hand and using the slight shifting of the image to determine depth and then push past the focal point of the closer objects. (I feel like I totally just pulled that sentence out of my ass)

Getting "lucky"?

[Mr Z]

This sounds an awful lot like the technology behind the "lucky" telescope". The basic idea, at least, is similar: Take the clearest images obtained over several samples and composite it into an image that otherwise couldn't be obtained given the distortion field.

This should work great for relatively stationary things. For moving objects, I imagine the effectiveness would be greatly diminished.

Thoughts?

Yes, 1 kilometer is within sniper range

[giafly]

The M-107 enables Army snipers to accurately engage personnel and material targets out to a distance of 1,500 to 2,000 meters respectively - M-107 Long Range Sniper Rifle

The muzzle velocity of such a rifle seems to be about 1 kilometer per second (M16 rifle), and also there's the one-per-second frame rate, so this scope seems best suited to assassinations, where your target is out in the open and stationary.

Re:That explains it...

[ElizabethGreene]

This reminds me of an article from several years back in ?popular mechanics? (I think) . There was a chap who "for fun" would hike to the top of a mountain adjacent to Area 51 and take pictures of the base and air traffic with an uber-telephoto lens. The rig was impressive to my untrained eye, and I have to assume it would make even the most hardened paparazzi jealous. From memory, he was ~6ish miles from the runway. The limitations on picture quality were from atmospheric distortion because of heat differentials rising off the desert, AKA the 'shimmer' from TFA.

Here you go mountain guy, this ones for you.