Neighborhood CCTV

[tomcctv 187]

we seam to be talking about differant cameras here i thought the first post was for a REG-X which can not give you a number plate at 18% the REG-X is for close work only 5-7m max giving you 120% as seen on the bosh site. we dont get are cameras from bosh are cameras come from our CNB distributor.

 

if you are looking for a camera for 18% then it would need to be REG-L1 to reach 35m but the price for that will be $2000 plus and for a camera that ONLY gives a number plate and not colour or make of car is a waste of money if not used how it was intended.

 

For $2000 you could use 4 of the CNB cameras http://www.videcon.co.uk/product_details.php?StockID=1203&l1=3&l2=23 and have money left over

that will do just the same but give you also what cctv is all about video evidence with as much detail as possable

[Bean00 0]

we seam to be talking about differant cameras here i thought the first post was for a REG-X which can not give you a number plate at 18% the REG-X is for close work only 5-7m max giving you 120% as seen on the bosh site. we dont get are cameras from bosh are cameras come from our CNB distributor.

 

You know what? I think I'm done with discussion. Birdman Adam gets it, and thats the only reason why I posted this info. I don't need to debate whether 18% of an image at 5m is the same as 18% of an image at 20m.

 

Scruit,

 

I got to say. You have the sweetest mobile setup I've ever seen.

[Scruit 0]

tomcctv,

 

The OP is talking about catching plates from cars that pass by in the closest lane within 2m of the camera as it sits on the side of the road. This is perfectly doable, even at a 35deg angle, within the 7.6m range of the REG-X.

The current thought process is not calling for 35m range, more like 6m.

 

Don't forget they are also talking about putting in a standard WDR cam that will get the vehicle description.

[Scruit 0]

Scruit,

 

I got to say. You have the sweetest mobile setup I've ever seen.

 

Thanks!

[Scruit 0]

we seam to be talking about differant cameras here i thought the first post was for a REG-X which can not give you a number plate at 18% the REG-X is for close work only 5-7m max giving you 120% as seen on the bosh site. we dont get are cameras from bosh are cameras come from our CNB distributor.

 

I don't get what you mean by 120%. When I say 18% I mean the width of the plate is 18% of the width of the screen. Using US license plates (same character height as UK number plates, but the lettering is thinner and the characters are closer together) this means the car takes up the whole image.

[Birdman Adam 0]

Aren't these percentage values always a percentage of the cameras view? The distance at which I'm capturing doesn't really matter, as long as the plate takes up at least 18% of the pixels.

 

For D1 resolution, we have 720x480 = 345,600 total pixels

345,600*18% or 0.18 = 62,208 pixels

 

So the license plate must make up at least 62,208 pixels to be readable by a human.

 

According to the manual, the REG-X maintains a shutter speed of 1/1000 all the time - shouldn't have any problem with blur there.

 

All I need is the REG-X - I'm only capturing cars that are starting out about 10 feet in front of the camera. After that they are pulling directly away from it, there will be plenty of time in which the plate will be above even 30%. Maybe I should draw another overview, as the camera pole will now be only a few feet off the side of the road, looking down a ~200' straightaway, up a slight hill. Angle will be less than 30 degrees.

 

 

Calculation shows the 16mm lens will be perfect for viewing the right lane of the street, and a wider view after that. (Our street is not full width, each is actually about 7' wide).

[Scruit 0]

Aren't these percentage values always a percentage of the cameras view? The distance at which I'm capturing doesn't really matter, as long as the plate takes up at least 18% of the pixels.

 

For D1 resolution, we have 720x480 = 345,600 total pixels

345,600*18% or 0.18 = 62,208 pixels

 

So the license plate must make up at least 62,208 pixels to be readable by a human.

 

According to the manual, the REG-X maintains a shutter speed of 1/1000 all the time - shouldn't have any problem with blur there.

 

All I need is the REG-X - I'm only capturing cars that are starting out about 10 feet in front of the camera. After that they are pulling directly away from it, there will be plenty of time in which the plate will be above even 30%. Maybe I should draw another overview, as the camera pole will now be only a few feet off the side of the road, looking down a ~200' straightaway, up a slight hill. Angle will be less than 30 degrees.

 

 

Calculation shows the 16mm lens will be perfect for viewing the right lane of the street, and a wider view after that. (Our street is not full width, each is actually about 7' wide).

 

Your 18% calculation is more complex than it needs to be. If your image is 720x??? then make sure the license plate is at least 720*.18 = 130 pixels wide. For US license plates, if you can see past *both* sides of the car then the plate is too small.

 

This is the smallest I would have the plates and still expect to read them...

 

 

Too small:

 

 

More thoughts:

[tomcctv 187]

tomcctv,

 

The OP is talking about catching plates from cars that pass by in the closest lane within 2m of the camera as it sits on the side of the road. This is perfectly doable, even at a 35deg angle, within the 7.6m range of the REG-X.

The current thought process is not calling for 35m range, more like 6m.

 

Don't forget they are also talking about putting in a standard WDR cam that will get the vehicle description.

 

 

this is were i am getting lost . so the camera is on the side of the road so in meters the car is 2m from camera to plate and the camera is from kerb to centre of road say another 2m and the camera is mounted 2m high.

 

 

with the bosh american calculator the REG-X can not see as close as 2m so entre 3m then distance from kerb to centre of road 2m and the camera is mounted 2m high. bosh calculator tells you that you are out of the 7m range and install for the REG-X and to use a REG-L1

 

 

http://products.boschsecurity.co.uk/en/GB/products/bxp/SKUF.01U.080.125-CATM4ee759c007a61cbce559f31fd53d9314

[tomcctv 187]

Aren't these percentage values always a percentage of the cameras view? The distance at which I'm capturing doesn't really matter, as long as the plate takes up at least 18% of the pixels.

 

For D1 resolution, we have 720x480 = 345,600 total pixels

345,600*18% or 0.18 = 62,208 pixels

 

So the license plate must make up at least 62,208 pixels to be readable by a human.

 

According to the manual, the REG-X maintains a shutter speed of 1/1000 all the time - shouldn't have any problem with blur there.

 

All I need is the REG-X - I'm only capturing cars that are starting out about 10 feet in front of the camera. After that they are pulling directly away from it, there will be plenty of time in which the plate will be above even 30%. Maybe I should draw another overview, as the camera pole will now be only a few feet off the side of the road, looking down a ~200' straightaway, up a slight hill. Angle will be less than 30 degrees.

 

 

Calculation shows the 16mm lens will be perfect for viewing the right lane of the street, and a wider view after that. (Our street is not full width, each is actually about 7' wide).

 

Your 18% calculation is more complex than it needs to be. If your image is 720x??? then make sure the license plate is at least 720*.18 = 130 pixels wide. For US license plates, if you can see past *both* sides of the car then the plate is too small.

 

This is the smallest I would have the plates and still expect to read them...

 

 

Too small:

 

 

More thoughts:

 

 

 

lens calculation can not be used on the bosh REG cameras its a black image camera it only sees ir. dont mix standard cameras settings with a bosh REG camera this is daytime image

[Birdman Adam 0]

Your images you just posted are very interesting - really shows how different settings will help with capture! I don't think the DVR I've chosen can duplicate views on to other channels.

 

If I need to manipulate a picture, (contrast, brightness, polarize, etc) I can use GIMP to do that easily. Looks like contrast can help out a lot to bring out the difference between the plate background color and the lettering color.

[tomcctv 187]

Your images you just posted are very interesting - really shows how different settings will help with capture! I don't think the DVR I've chosen can duplicate views on to other channels.

 

If I need to manipulate a picture, (contrast, brightness, polarize, etc) I can use GIMP to do that easily. Looks like contrast can help out a lot to bring out the difference between the plate background color and the lettering color.

 

 

 

Birdman Adam. that is what i have been saying all along use standard cameras and not the bosh REG-x also see how far scrits car is away from the camera the REG-x can not get a plate from that distance. also if you use your settings as

For D1 resolution, we have 720x480 = 345,600 total pixels

345,600*18% or 0.18 = 62,208 pixels

 

you will also have a bad image they need to record in cif. remember a cif image with a very high bitrate give good detail images

[Scruit 0]

Why do you keep saying CIF gives better image than D1?? I don't get it...

 

Edited January 6, 2011 by Guest

[Scruit 0]

Your images you just posted are very interesting - really shows how different settings will help with capture! I don't think the DVR I've chosen can duplicate views on to other channels.

 

If I need to manipulate a picture, (contrast, brightness, polarize, etc) I can use GIMP to do that easily. Looks like contrast can help out a lot to bring out the difference between the plate background color and the lettering color.

 

 

I don't think GIMP will help you in this case... The brightness and contrast is set on the DVR at before the image is recorded. I don't think you will get the same effect altering the resulting jpg image after the fact.

[Gator 0]

I'm about to try the BBB-34 in this application as a driveway camera. If it doesn't work for this, I'll stick it somewhere else. I'll try a 50mm lens at first. I may need to go to 80. I'll let you know if it works with bright headlights.

 

You photos shows cars rounding the bend. You may get better results if you use a longer lens and catch them at a straight spot in the road. Then they won't be moving side to side in the frame. There will be less interlace hatch in the images because there is effectively less movement.

[Scruit 0]

You photos shows cars rounding the bend. You may get better results if you use a longer lens and catch them at a straight spot in the road. Then they won't be moving side to side in the frame. There will be less interlace hatch in the images because there is effectively less movement.

 

Agreed. That's why I moved my own LP camera to be mounted dead center on the driveway.

 

However, the longer lens camera is almost triple the cost of the REG-X. And the reg-x IS spec'd for 50mph at 40deg looking down and 40 deg looking across the road. So although the image would undoubtably be *better* with the car driving directly away from the camera rather than past it at an angle... If the "past it as an angle" image is perfectly satisfactory then it wouldn't justify the extra cost of the bigger camera.

[Birdman Adam 0]

Maybe I should delete my previous posts on the first-third pages...

 

-The REG-X will be looking up the straightaway, right next to the road. Viewing angle will be less than 30 degrees, with starting distance from cars ~10'.

 

-I will use a WDR camera from CNB to get car details.

 

-I will be recording at D1 at 30FPS.

 

-Cars will be driving almost perfectly straight away from the camera between 10' and like 50'.

 

-Why would lens calculations be different for the REG-X compared to a normal camera, its still a CCD sensor, still 1/3", still a normal 16mm lens... Sure the sensor is sensitive to only IR light, but why would that change the viewing angles???

[tomcctv 187]

-Why would lens calculations be different for the REG-X compared to a normal camera, its still a CCD sensor, still 1/3", still a normal 16mm lens... Sure the sensor is sensitive to only IR light, but why would that change the viewing angles???

 

 

Hi Birdman Adam. standard lens calculations are for field of view. standard lens. the REG-X and the rest have a Black lens it cant see field of view its there to look for reflections from its IR reflecting back. if you stand infront of the camera it will not see you at all but it will pick-up reflection from your watch or belt buckle.

[rory 0]

-Why would lens calculations be different for the REG-X compared to a normal camera, its still a CCD sensor, still 1/3", still a normal 16mm lens... Sure the sensor is sensitive to only IR light, but why would that change the viewing angles???

 

 

Hi Birdman Adam. standard lens calculations are for field of view. standard lens. the REG-X and the rest have a Black lens it cant see field of view its there to look for reflections from its IR reflecting back. if you stand infront of the camera it will not see you at all but it will pick-up reflection from your watch or belt buckle.

 

Black lens or black cover?

The REG-L i used in the past had a black cover.

 

this REG-L could see things. Its basically just a dark tinted cover and IR for night to bring out the reflective objects, and a standard BW camera.

 

 

[Bean00 0]

-Why would lens calculations be different for the REG-X compared to a normal camera, its still a CCD sensor, still 1/3", still a normal 16mm lens... Sure the sensor is sensitive to only IR light, but why would that change the viewing angles???

 

 

Hi Birdman Adam. standard lens calculations are for field of view. standard lens. the REG-X and the rest have a Black lens it cant see field of view its there to look for reflections from its IR reflecting back. if you stand infront of the camera it will not see you at all but it will pick-up reflection from your watch or belt buckle.

 

Well it has an IR pass filter, so that it blocks light that it's not emitting. It helps with LPR tag rec on reflective plates, eliminating other light sources in the image such as headlights.

 

But a 1/3" CCD with a 1/3" 16MM lens is the same field of view regardless of whether it has a IR pass filter or not.

[Scruit 0]

I'm still trying to figure out why tomcctv keeps saying that a cif image is better than a d1 image in this configuration:

 

you will also have a bad image they need to record in cif. remember a cif image with a very high bitrate give good detail images

 

because it says in the manual for the camera dont use D1 as artifacts (datalace) will result in bad image. cif is recommended. with good bitrate.

 

dvr should record 2cif or 4cif ............ the camera will not work with D1 and will need to run 30fps.

 

You get interlacing with d1, sure. That's why you de-interlace. The 18% calculation works for D1 but does it work for cif? At 320 pixels wide that means the plate is ~60 pixels wide. IS that enough detail to read a plate? I'm be aiming for 35-40% plate/screen ratio if I was forced to use CIF. But I wouldn't use CIF, I'd use D1 on a DVR that has a good de-interlace tha is tuned for motion. (aver's de-interlace #2 works great for my needs).

 

 

And as far as lens calculations go - they work as expected for my camera setup which is a standard camera with a standard lens and IR filter. I can't figure out what part of the LP camera would change the physics of light to a point where the CCD size plus focal length cannot determine the field of view. Is an LP camer not simply a higher shutter speed camera with a standard lens and an IR filter?

 

Again, folks, you guys are the experts on here and I'm just a hobbyist, but I like to think I've learned enough to be competent. If there is some explanation why CIF at 29.97fps is better for LP capture than D1 at 29.97fps (which is then deinterlaced) then just tell me - don't leave me sitting here wondering why I'm being told I'm wrong when I point out that it is counter-intuitive. Same with the lens calculations... Is 16mm on a 1/3" ccd through an 850nm ir pass filter REALLY a different FOV than the same withouth the filter?

[Bean00 0]

I'm still trying to figure out why tomcctv keeps saying that a cif image is better than a d1 image in this configuration:

 

you will also have a bad image they need to record in cif. remember a cif image with a very high bitrate give good detail images

 

because it says in the manual for the camera dont use D1 as artifacts (datalace) will result in bad image. cif is recommended. with good bitrate.

 

dvr should record 2cif or 4cif ............ the camera will not work with D1 and will need to run 30fps.

 

You get interlacing with d1, sure. That's why you de-interlace. The 18% calculation works for D1 but does it work for cif? At 320 pixels wide that means the plate is ~60 pixels wide. IS that enough detail to read a plate? I'm be aiming for 35-40% plate/screen ratio if I was forced to use CIF. But I wouldn't use CIF, I'd use D1 on a DVR that has a good de-interlace tha is tuned for motion. (aver's de-interlace #2 works great for my needs).

 

 

It's not better. Bosch/Extreme recommends trying 2CIF if you have an issue with D1 because of possible deinterlacing issues.

 

They absolutely do not recommend CIF. Matter of fact they state to record higher than CIF in those docs I linked to earlier in the post.

 

And the 18% rule won't work in CIF as it would be too pixelized to read in that small of an image.

Edited January 8, 2011 by Guest

[Bean00 0]

Same with the lens calculations... Is 16mm on a 1/3" ccd through an 850nm ir pass filter REALLY a different FOV than the same withouth the filter?

 

Nope, exactly the same FOV.

[tomcctv 187]

Same with the lens calculations... Is 16mm on a 1/3" ccd through an 850nm ir pass filter REALLY a different FOV than the same withouth the filter?

 

Nope, exactly the same FOV.

 

 

 

Bean00 what i was saying was not to record in D1 (like bosh says) cif/ 2cif/ does not matter which. the software you get with the camera is what you zoom into the plate with not the dvr. (you do the zooming on a pc it also comes with enhacement tools)

 

using a lens calculator .............. yes its fine for a standard camera the REG-X at 8.3m is dead will not see anything at all .......... so no point using the lens calculator.

 

am trying to get permision so as i can give birdman an ip to 3 REG-X cameras i should have it monday

 

as i use the bosh REG-X for campsites and the barrier control (which will be the link) i dont understand why use this camera just in the way you would use a standard camera it was not designed for that use.

 

you could use a camera like this http://www.videcon.co.uk/product_details.php?StockID=1203&l1=3&l2=23 ........ it will get plates day and night at a good range also colour of car and detail. and you will save $700 that is what i cant get my head around this post

 

Scruits set-up is very good even had comments on it. and scruits system is my point he has the licence plate and you can also see it is a subaru forester (good car i had one when i lived in westvirgina) and its done with a stanard camera. the REG-X will only give you a plate in a black image. so you are paying more for less detail and information.

[bpzle 0]

Same with the lens calculations... Is 16mm on a 1/3" ccd through an 850nm ir pass filter REALLY a different FOV than the same withouth the filter?

 

Nope, exactly the same FOV.

 

 

 

Bean00 what i was saying was not to record in D1 (like bosh says) cif/ 2cif/ does not matter which. the software you get with the camera is what you zoom into the plate with not the dvr. (you do the zooming on a pc it also comes with enhacement tools)

 

using a lens calculator .............. yes its fine for a standard camera the REG-X at 8.3m is dead will not see anything at all .......... so no point using the lens calculator.

 

am trying to get permision so as i can give birdman an ip to 3 REG-X cameras i should have it monday

 

as i use the bosh REG-X for campsites and the barrier control (which will be the link) i dont understand why use this camera just in the way you would use a standard camera it was not designed for that use.

 

you could use a camera like this http://www.videcon.co.uk/product_details.php?StockID=1203&l1=3&l2=23 ........ it will get plates day and night at a good range also colour of car and detail. and you will save $700 that is what i cant get my head around this post

 

Scruits set-up is very good even had comments on it. and scruits system is my point he has the licence plate and you can also see it is a subaru forester (good car i had one when i lived in westvirgina) and its done with a stanard camera. the REG-X will only give you a plate in a black image. so you are paying more for less detail and information.

 

I always read your posts Tom... Lots of things to learn from you.

[Birdman Adam 0]

It seems to me the REG-X using the filter to only see IR actually improves plate readability. If I was using a normal camera, or even a 'license plate camera' without an IR only filter, it would get blinded by headlights/taillights/etc.