Help Chinese Surveillance Cameras See Through Smog

"Beijing has invested heavily to build up a nationwide surveillance network that lets police watch every major street and corner in main cities.

But with smoggy days becoming more frequent, the effectiveness of the system has been greatly compromised. Some fear terrorists may choose a smoggy day to launch attacks."

Chinese experts are working feverishly on a solution: "Our preliminary research shows that the smog particles are quite different from the small water droplets of fog in terms of optical properties. We need to heavily revise, if not completely rewrite, algorithms in some mathematical models. We also need to do lots of computer simulation and extensive field tests."


When I first read this story last night, I thought, "Whatever company answers the call, whether it works or not, is bound to make a killing." Also, I'll be interested to see the conclusion of their study.

Are you kidding? Two things.

1. Beijing's solution to their smog problem should not be better cameras, but to clean up their pollution.

2. I still question Beijing's definition of terrorist more than I do the US.

I know those can be considerd political orientated comments, but seriously, what are they going to film if they get special "smog" cameras but video of people keeling over in the streets.

"We need to heavily revise, if not completely rewrite, algorithms in some mathematical models."

I agree with Undisclosed. Focus on the root cause first, it will pay bigger dividends in the long run.

There's a significant body of research associated with enhancing visibility through a scattering medium, often involving coprocessing outputs from multiple cameras that are independently observing the same scene. Could be promising for integrators and manufacturers alike... .

"involving coprocessing outputs from multiple cameras that are independently observing the same scene"

Sounds very expensive!

...expensive... lucrative... same thing to some :)

Ha ha...great comment!

Pure research or applied research?

Can you recommend any material which gives a decent overview?


Microsoft has freely downloadable software called Image Composite Editor (ICE) that is said to integrate multiple video frames to provide sharp images, among other things. They give an interesting example of de-hazing Mount Rainier from 90 miles away.

What ever became of the Pentax PAIR technology? Seems ideal for this.

It would be helpful to know the size of the dropplet (scattering medium). Presumably it's bigger than water, which is typicaly 10 - 15 microns. That would I think explain why the article states conventional long wave IR imagers at 10 - 12 microns aren't effective. It would seem the imaging system would need to operate at wavelenghts larger than 15 microns. That's a tough one, as I don't think there is much unitl one hits the millimeter wave devices. I suppose for detection an active mmW should be fine, but can't really be considered an imager in the surveillance context.

Smog has always been there like in Mexico City, L.A. and a lot of other big cities. How about fog, that is a problem too with some U.S. cities no? But like H. Lassell said expensive...lucrative is the name of the game.

For example, "Seeing through biological tissues using the fly eye principle" is a seminal work showing how integration multiple images from many independent imaging channels can enable one to see general objects hidden behind scattering layers.

More recent efforts investigate mechanisms to see through tissue (a scattering medium) using a variety of mechanisms including microlens arrays. It seems that medical noninvasive imaging applications might be a little less cost sensitive than some surveillance applications can be.

Btw, there are a few existing surveillance products that claim to deal with fog, if not smog, similar to Hal's example of Microsoft's Image Composite Editor.

See Lyyn Griffin and IMINT Vidhance. Both work in real time and both are really expensive (thousands per channel).

PS, this whole article / story reads like an Onion skit.

Those are pretty interesting concepts. I'm going to download and play with that Microsoft ICE- that's waht so cool about discussion groups, no matter how much you know or think you know you learn about something new and possibly useful all the time.

My company, Pixon Imaging, sells video processors that are designed to see through fog. And yes, smog can act very differently than water droplets. Now I cannot comment of the work that the "Chinese experts" are working on since I am not privy to this information. However to perform a true solution to this problem in a practical, real-time solution is next to impossible. The individual properties of the particle affect their scattering and sbsorption properties, and these change. The particle size distribution affects the net optical properties, and these change. The angle of illumination affects the scattering properties, and these always change, and if there are multiple, changing sources of illumination, how can one keep up? What is needed is a practical, "good-enough" solution. In otherwords a compromise must be struck and one shouldn't bite off too big a piece of the puzzle. Simplifying assumption need to be made as we do in our products. Indeed, we have applied our algorithms to smoggy pictures and they help significantly even though they are not tunned specifically for smog (they are general purpose). Of course with 3 meter visibility, one will never see the mountains, but we generally claim that our products give useful video at 2 to 3x the distance one could see without using them.

Rick, do you have a video demo of your product dealing with smog?

Hi John. I don't have anything quite like the ones in the artcle. But we do have some with smoke, which is often one of the major components of smog. There are some still images on my company website. This is the same algorithm for fog, smoke, and murky water. You can also see the video of what we did for a smoggy day in San Diego.

San Diego doesn't have too much of a smog problem, but we were looking through about 5 miles of atmosphere. Note that while the dehaze step is probably the most important, the contrast enhancement steps are also significant and we currently have even better algorithms on our current products. I'll also scour my computer to see if I have anything else that would be more comparable to the conditions in China. As it happens, we have a Rep there right now and he is talking to some of the Chinese officials about this.

Rick, roughly how much does this cost? The videos seem similar to other image enhancing technologies we have looked at that cost thousands per camera.

To the extent that this costs in that range, I struggle to understand how even large users can justify this in widespread deployments.

Hi John. Yep. Unfortunately we are in the same price range. At quantity 100, we have models that are in the $1000 to $3000 range. Huge quantities, like the millions of units that China would need for their cities, would drop the price down as you could build an ASIC and be < $50. But short of that, the lowest price I could see would be in the $200-$500 range. And to do the things they really need on their smoggiest days, I think that is simply not technologically possible. As has already been said, the real solution is to clean up the smog.

Hi John,

We are at SecuQuest, do have the VisionClear technology including Smart De-mist, DWDR, Smart DNR and Smart ACE to enhance video feed realtime in the adverse conditions such as smoggy, foggy, rainny, snowy, sand storm, low lit...

We offer Video Enhancement Boxes with VisionClear technology, see specs of our VCE11, VCE12 and VCE13.

Our prices are very reasonable, retail at $229, $1,499 and $2,249 for VCE11, VCE12 and VCE13 respectively. We are the only company offering VCE box for HD quality video feed.

Thomas, what's the difference in functionality between the 3 offerings? It appears only the 13 model does HD and that's HD SDI, not HD IP. Is that correct?

Btw, you're website raises yellow flags to me because it's missing any contact information (e.g., no physical address), the about section is extremely vague (no real details of history, management, size, etc.) plus the FAQ is empty. It makes me wonder if this is a real company or just a re-seller of someone else's products.

The website does have a bit of the alibaba feel to it, particularly in its box and bullet sections.

On the other hand this "trackball" dome is different that any I've seen (not saying much).

I find myself strangely attracted to its sultry half-closed eye.

"trackball Dome"

Perhaps the membership at-large can say whether it is a one of-a-kind item or just another newbox repack??

Dibs, tho.


Hi John,

You are correct, the HD on VCE13 is SDI not IP. The main difference between VCE11 and VCE12 is our VisionClear techonology, we are using third generation on VCE12, and for VCE13 we offer HD-SDI capability.

We do have physical presence in US.

Since our company is strong in technology and IC with R&D group in Silicon Valley, we have applied the technology to surveillance industry to help enhance live video feed for existing camera in the adverse situations. As you see on our site, we do offer analog camera with VisionClear technology at very affordable prices, retail ranging from $80 to $160 depending on resolution and model, from 600TVL to 700TVL, dome to bullet...

We are working on IP cam with VisionClear technology which we are targeting in Q1 2014.

Saw this question pop up on my Quora feed this evening, maybe someone here has an answer? :)

What is China going to do with their security cameras since they're now useless because of the pollution?

Hikvision is promoting an 'Optical+ Algorithm' to solve the smog problem.

Code name: HazyViz