Friday, November 20, 2009

Why I hate LUX ratings...selecting a good night vision camera



If you read this, and find it useful or not, please leave a comment below. Thank you.

In the mounds of junk out there in the surveillance world, it seem there is an over-hyped specification that many use to choose their next security camera. I'm talking LUX. Now if you're reading this you've probably already seen those cameras claims: "Low Lux", "Zero Lux", "0.0001 Lux!", and so on. Quote me here people..."Lux means nothing." Sure, by definition the lux rating of a camera should give you an idea of how well it performs in low light. The problem with this is that each camera manufacturer can cheat in a couple of areas to inflate their low light performance specs, and get the desired marketing spin they are after.

Area 1 - "Usable Video"

The term "usable video" is often used to determine at exactly what point a video image is deemed "usable" by a surveillance footage reviewer. As one could imagine, this is very subjective. Furthermore, the industry is completely unregulated, and does not force manufacturers to test in identical test setups. Many manufacturers simply buy a CCTV chart and pick a spot to look at. They point their camera at that spot, dim the lights to the desired lux rating, and determine if the video they are viewing is deemed "usable". You can only imagine the conversation in the darkroom....

Tom: "Can you see that outline of the boys face on the chart Bob?"
Bob: "Um...ya..a little."
Tom: "Well, then ok, that's usable, great lux rating!"

You may think I'm kidding, but half the camera manufacturers out there don't even check the lux rating, they simply import ccd imagers and use the specs that came with the imager. Others take those imager specs, then count up the number of IR led's they're going to add to it, multiply by the lumens rating of each LED, and develop an algorithm that results in some bogus lux rating of "0.0032 Lux!". Better labs will run their video through a waveform analyzer, point their camera at a particular resolution chart, then ensure the waveform meets certain voltage criteria. Again, though, this criteria is established by the manufacturer. This leads me to cheat area #2.

Area 2 - Lux Meters:

The funny thing about these claims of "0.00002 Lux!" is that few light meters in the price range of most labs even display this many digits. I once sat across the table of the largest camera manufacturer in North America and listened to them try to impress me with the specs of their new camera. Don't get me wrong, they make great cameras, but when asked what I thought about the camera's impressive lux ratings of "0.00002", I responded, "I'm more impressed that you have a meter that measures that many digits." They didn't get it, but that bogus lux rating still made it to their marketing material.

So the question you might have now is: How do I find a camera with great night vision? Here are my recommended steps:

1) Know your CCDs - Pick a Sony Ex-View. These CCDs are excellent in my experience. Look it up yourself, these chips are so sensitive that they have been known to detect cosmic rays. (Seriously - the rays cause dead pixels over time). A VERY good chip.

2)LEDs are usually a bad sign - Shy away from cameras with lots of IR LEDs. LEDs are like a short guy with a monster truck...they're making up for something. Nowadays it is harder to find an outdoor camera that doesn't offer some kind of IR option....and don't get me wrong, IR has its place (like shooting 200ft down a runway), but If the camera is littered with LEDs and claims 100ft range, check the CCD. If it is a HAD or SUPER HAD (not ex-view), then you're going to see little to nothing past the range of those LED's and overall be unhappy with the selection. The Ex-View will show you , what I like to call, "Desert Storm Night Vision". Remember those green videos of night vision from Desert Storm...ya, that's pretty much what ex-view will show you at night. You can see your driveway, your car, your neighbors house across the road, the grass in the distance....very good cam.

3)DSPs Matter - Once you've picked the right chipset, be sure the camera has a solid DSP supporting it. The better DSPs have low light enhance modes. (The Bosch Dinion 0495 for example has multiple light enhancing modes) Check the specs for additional features that add to the low light experience. Anything without an on-screen menu of some type should not be considered.

4) Light touches the lens - Pick an aspherical lens that is made for IR light. Otherwise, you'll be putting cheap tires on your Ferarri of a camera.

If you follow these guidelines, there is no question you will be amazed by the night time performance you'll experience. And if you're not happy...well there's always thermal.

Speaking of...if you plan on using analytics with your night system, you may want to carefully evaluate the effects of distant lights that the camera may see. A headlight from a car a mile away can cause your super sensitive camera to start triggering phantom movement well over the threshold of your analytic boundary rules. I've experienced this first hand, and the project would have been much better had the original planners chosen thermal instead of excellent night IR vision. The night vision was so sensitive, it was picking up distant headlight IR reflections that were not visible to the naked eye and causing the analytic boundaries to trigger falsely.


Last word: Lux ratings are bogus...the only place I would use them to compare anything is when comparing camera models from the same manufacturer. Lux ratings across manufacturers are much too subjective to be used otherwise.

- Security Insider

Friday, January 2, 2009

Beating traffic cams...not so easy.

Well, let's turn this into a new post instead of a comment.

Intercooler said...

Security Insider. You posted an entry on boing boing, indicating that IR LED’s would not interfere with CCTV cameras, like Pelco, Panasonic, etc…

I’m curious if you could state the effectiveness IR LED’s aimed at photo radar cameras would have. Both day and night conditions.

Thank you.
November 11, 2008 3:50 PM


Security.Insider said...

It depends on how the light hits the camera, the camera type, and the timing. The camera should automatically adjust to the light, if it has wide dynamic range, and most radar cameras do. The timing of most WDR's is slow however, so the IR light could disrupt the image at night if it was a quick pulse. During the day, you will see little effect, if any by shining any light into a camera. Most are setup to withstand direct hits from sunshine, and it doesn't get much brighter than that. I know where you're heading with this though. To beat a traffic cam, a speeding cam, at night, you'd have to trick the cameras own iris. If the camera is the kind that flash when you're speeding then you're not going to overpower them with another light unless it was a laser or spotlight, and good luck hitting a camera lens with a laser (the angle has to also be right to washout the entire image, we've done this in the lab)...all while speeding past. Also, most video surveillance systems have alarms setup to detect this. It ,technically, is called a video occlusion filter. The same type of filter that detects a camera knocked out of focus or other dramatic scene changes. Also, some cameras will have light blanking, where they will make a negative out of bright lights. So bright becomes black spots and dark becomes white spots. Most license capture systems utilize blanking, so you're not going to wash it out. Many video capture systems use thermal as well. If it is thermal, then IR will not effect the camera at all. Most, if not all, incorporate a ccd along with the thermal, so they'll catch the image one way or another, even if you use simple methods to beat the thermal. Check out the links below for more info on one of the more popular systems out there. (http://www.extremecctv.com/products_video.php?vid=9) In the meantime, realize the speeding cameras are ultimately there for everyones safety and we all share the roads. Keep the racing on the 1/4 mile, and less kids will die. Also remember, you never know when this might happen when road racing.

http://www.youtube.com/watch?v=pdxmAq3TOjI

-Security Insider