Settle a bet for me

[CameraMan]

I have an argument going with a customer, and I need you guys to prove I'm right as usual.

Here's the thing. Customer will have a camera, mounted at about 10 feet high, aiming down a very long driveway. Customer wants to capture license plates, I spec'ed a wide dynamic range camera (the Sanyo VCC-W8774, I saw it at ASIS this year, it's awesome) and a very long Tamron lens, to get a tight shot of the car. The WDR is to eliminate glare from the ehadlights and reflections from the license plate.

Now, my customer was not happy with this. He informed me of the existance of a mythical creature called the "license plate camera". I looked at a few of these, and I noticed, they all have IR illuminators! Wouldn't the light of the IR give you a reflection off the reflective surface of the license plate?

I insisted that my setup would work, and the customer insisted that it would not, and now we have a $1 bet going. Anyone care to weigh in? I appreciate it.

[Erron S.]

Yes sir, this creature does exist....
http://www.extremecctv.com/products.php?producttype_id=6

Having said that, and I've also sold a number of them, this would be a nice addition to the camera you have in place. As you can see from their demo video, it will give you the license plate, but not the car or driver. Which would be the perfect application for a WDR camera.

[integrator97]

Nessie exists :eek: ha-ha, jk. I've seen the REG cameras our new junior friend Erron mentions, but I can't attest to their performance. I don't think IR reflection would be an issue in any case.

I've also seen a camera built into a speed bump. Unless you're taking high speed (many ips), you will always have a problem if you don't have a way to get them to slow down. And dirty dusty plates are also a problem.

[CameraMan]

I'm sorry, I ws unclear in my original post. My contention is that a WDR camera will work just as well, if not better, than the gazillion dollar "license plate reader" cameras out there, and for cheaper.

The camera will focuse at the end of a driveway, just before you turn into the parking lot or crash, so I think that people will be going at 10-25 mph in any case.

[integrator97]

That may be so, I don't know that the camera's are any better or different than a WDR, other than coming in a nice package with illuminators. But I still think you may need an illuminator if you're trying to see them after dark. Unless everyone has a well lit plate.

[Erron S.]

As you can see, Extreme seems to have done their homework on this one. They have adjusted the gain on the camera to compensate for the IR. I believe they also tout it for being able to get plates at fairly high speeds. (don't remember the exact speeds) It certainly works well from the responses I've recieved from the dealers I sell to. It's a little on the pricey side but sure does the trick! :D

[Silva Consultants]

We had a very demanding application for outdoor license plate recognition a few years back and did some extensive experimenting with different types of cameras and IR illuminators. We found that IR illuminators can create a glare, depending on the angle that they hit the license plate. This application can be much more complex than it first appears, as there are so many different types of vehicles, each which seem to have their license plates and lights at a different height and location.

The results of our tests were not very encouraging; we could only guarantee capture of the plate number about 85% of the time. What seemed to work best was to get the cameras down to the license plate level (2' to 3') and as close to the drive lane as possible. You definitely need a camera dedicated to license plate capture; trying to get a shot of the license plate with a camera that views the overall scene is wishful thinking. Having cameras on both the front and back side of the vehicle increases the success rate dramatically.

[integrator97]

Silva,
You may be right, but are you sure the glare wasn't from chrome around it, which I hadn't thought of. But who knows. Or it could be a hot spot the camera can't adjust for with a dark background, which would be the same as glare I guess.

Did you happen to try the cameras in the speed bump? I'd like to know how well they work. If you have any data or more info to post, I'm sure we'd find it helpful.

You're spot on that you need a very small field of view, to get a useable image. Going from the front wouldn't help you here, we don't have front plates, unless you want to know what they are a fan of.

[CameraMan]

I will have a 100mm (VF) lens and I will mount the camera 40 feet away from the spot I will focus on, at a height of less than 10 feet. So I am certainly going to get a tight shot of the vehicle coming and going (16 feet across at most).

I spoke to this customer who told me "it's not possible that a 'license plate camera' is nothing more than a WDR camera and an IR illuminator, why would ADI sell license plate cameras for four grand, then?".

"Because I know how lenses and DSPs work" is not going to be an acceptable answer, here.

[integrator97]

Sell him the $4k camera then. :D Get the rep to get you a demo. The REG uses a 1/2" CCD, and is designed to get plates a 100mph. If he needs that, spend the money. But for your app, you can probably do it for alot less your way.

16 feet across, I don't think your going read plates, unless you're recording hi-def, and/or have some sweet software. But your lens can go down to about 1.5' high by 2' wide according to my handy dandy lens wheel, assuming a 1/3 inch CCD camera.

[CameraMan]

Trust me when I say that my first instinct was to just sell the alleged so called "license plate camera" and bank the difference. The only thing is this customer is bidding against a lot of bigger, better-known installation companies, and I'm trying to get the bid as low as possible.

But, screw it- I can only argue so much before I agree and sell the customer something he doesn't need and costs 5x what I originally offered.

[Silva Consultants]

Silva,
You may be right, but are you sure the glare wasn't from chrome around it, which I hadn't thought of. But who knows. Or it could be a hot spot the camera can't adjust for with a dark background, which would be the same as glare I guess.

Did you happen to try the cameras in the speed bump? I'd like to know how well they work. If you have any data or more info to post, I'm sure we'd find it helpful.

You're spot on that you need a very small field of view, to get a useable image. Going from the front wouldn't help you here, we don't have front plates, unless you want to know what they are a fan of.

The actual application that I had was on a bridge that passed over the tracks of a light-rail commuter train system. People were stopping their cars on the bridge, and throwing objects down on the trains, including things such as television sets and bowling balls (I kid you not...). Our client wanted to use CCTV to record this activity, including capturing the license plate numbers of the cars involved.

We couldn't test on the bridge itself, so set up a simulation in the alley and parking lot behind our office. We had cameras and illuminators set-up on tripods, and recorded our results using VCRs. We drove vehicles of various types up and down the alley and back and forth into the parking lot for several hours, constantly varying the placement of the cameras and illuminators (our neighbors thought we were nuts..)

We tried various cameras and illuminators (including the Extreme product) and nothing gave us anything other than mediocre results. We found that the IR illuminators worked just like spotlights - they cast a narrowly focused beam of light that would cause glare if it hit the plate at the wrong angle.

In the end, what worked best was a camera and separate illuminator mounted down low no more than 10' away from the car. Like I said earlier, even this combo could be blinded at least some of the time. Brake lights were a particular problem, so I'm thinking if you are looking at a car at speed bump, this could be an issue. Not having both a front and rear license plate makes things even more difficult.

It has been several years since we did our tests. There may be new technology out there now that I'm not aware of, but I would be very skeptical until I saw the product tested in a real-world environment.

[Silva Consultants]

I should clarify my last message to say that I tested a couple of different types of the Extreme IR illuminators, but did not test the Extreme licence plate camera system that was referred to earlier. This product was not available at the time I did my tests or we surely would have tried it.

[Vegas Tech]

Hmmm I need to post some pics of car traps I've done for valet here in vegas... They need to see a scratch!!! errrr!!! But I have found that the camera placement is more important than the camera type and controling the cars approach speed and position is the hardest. We had about 60 % success at this. Valet drivers would fly throuh the recorded area not giving the cam time to do it's job or there was no specific lane to drive through. So we added a gate with a 2-3 second delay to force the car to sit still for a bit. That got us up to 80%. As far as the rest the customer was using AD Intellex recorders and you can turn a photo/video clip into a negative and get the plate #'s usually.

Oh and I was using Elmo 1/2" cameras in the typical Pelco housings...

Camera Man you are correct imo

:D

[Curtis Baillie]

Here's an article on license plate recognition: http://www.securityinfowatch.com/article/article.jsp?siteSection=430&id=18163