Jump to content
nfair

Cameras in fog, especially at night with IR

Recommended Posts

I went through some overnight video and I saw what looked like a heavy snowstorm - lots of white dots moving through the image at times getting so heavy I couldn't make heads or tails of the image. The only problem was it didn't snow that night. It wasn't until I got to the early morning video that I figured it out. The IR shut off and the picture cleared up quite a bit. I could see the low lying fog but I still had a much better picture.

 

Is it the IR light reflecting off the fog/water vapor? Is there anything to be done about it? I imagine I could do away with using IR and just keep my white light on overnight but I'd love to find a way to automate it.

Share this post


Link to post
Share on other sites

Is it worth knowing the difference between IR, visible light and low light image capture in conditions of fog, rain, snow or smoke?

Do thermal sensors always have the advantage? (Notwithstanding high price.)

 

Example of image in fog with the Hitachi KP-D5000 vs KP-DE500.

Hitachi claim it's a Hitachi Fog Demo. The description below it says 'smoke'.

 

Share this post


Link to post
Share on other sites
I imagine I could do away with using IR and just keep my white light on overnight but I'd love to find a way to automate it.

Automate what exactly?

 

I just figure it's the price of doing business with IR. All my cameras have it and they all do the same thing. Not really so much of a bother, as heavy fog that causes it happens more infrequently than frequently. Cameras without IR, such as what I monitor at work- it just looks like fog usually does when visibility gets low. That's good, but they have other problems that my bullets and eyeballs don't, so I still favor IR cameras. And motion lights too of course.

Share this post


Link to post
Share on other sites
Cameras without IR, such as what I monitor at work- it just looks like fog usually does when visibility gets low. That's good, but they have other problems that my bullets and eyeballs don't, so I still favor IR cameras. And motion lights too of course.

Strictly speaking, the thicker the fog gets, the worse the visibilty. Light just bounces off the water-laden air. The expert claims that the characteristics of IR light in fog closely match the behaviour of light from the visible spectrum.

Share this post


Link to post
Share on other sites

When monitoring the cameras without IR, fog looks like it does to the naked eye- visibility lowers depending on how thick it is- an overall haze. Although that of course could make it harder to track an intruder at that very moment, it's nowhere near as difficult to do as when you're doing it with an IR camera, where the visibility isn't just lowered- it's all but wiped out due to water droplet noise. It indeed looks like a massive rain storm. But I still prefer cameras with IR, as I said. Most times that IR, coupled with some external lighting, makes all the difference between seeing anything well, and almost nothing at all in low light.

Share this post


Link to post
Share on other sites

I suppose it is not IR problem. Visible light produces the same effect, like headlight in fog.

The problem is in the direction of light and known inverse square law. Fog/snow particles are near to the camera and the build in illuminator therefore the particles are much brighter than other far objects on the scene. Then camera automatic (AESC, AGC) adjust image contrast at these bright particles, at the same time other far objects on the scene become darker.

To improve visibility in fog/snow you should place illuminators not near the cameras, but aside or from above. May be place additional illuminators and turn off build in illuminators during fog/show.

Share this post


Link to post
Share on other sites
To improve visibility in fog/snow you should place illuminators not near the cameras, but aside or from above. May be place additional illuminators and turn off build in illuminators during fog/show.

This is why fog lights on cars are mounted as low as possible - more direct light on the road, less reflecting straight back to the driver.

Share this post


Link to post
Share on other sites

The three still shots come from three different Hitachi cameras aimed at the same subject. The top left shot is the KP-D5000. It retails for about $1800 US.

 

Any explanation as to why it has better fog capabilities? Hitachi refers to “Adaptive Fog Reduction.” It appears to be a highly sensitive CCD, but the details of the sensor have not been established. Also, without a reference shot, it is difficult to estimate the extent of the ‘fog’ released by Hitachi.

Any explanations about the top left photo?

2010256523_HitachiFogDemo.thumb.jpg.fbaa66acd9947e4d436d5bba1221eba5.jpg

Share this post


Link to post
Share on other sites

Any explanations about the top left photo?

 

There is a sufficient explanation in the KP-D5001(P)/D5000(P) manual. This is a contrast correction.

 

Adaptive Fog Reduction

AFR feature provides real-time image correction for improved visibility in dark,

harshly backlit, or unclear/foggy environments, while optimized computational

algorithms enable frame rates up to 30 fps. Through use of this technology the

camera is capable of adjusting the contrast ratio of images that become pale

due to foggy conditions and restoring the fading colors resulting in dramatic

improvement in visibility.

Share this post


Link to post
Share on other sites

I obviously read that. But you miss the point.

I wondered:

"Any explanation as to why it has better fog capabilities?"

The KP-DE500 appears to be marketed as Hitachi's top 1/2" camera and sells for $6000+, while the KP-D5000 sells for $1800. It seems remarkable that a camera that sells for 1/3 less than the KP-DE500 outblasts it in a fog test.

Anyone might wonder why the KP-DE500 doesn't have AFR. And, besides, cameras are replete with terms like this. It seems rather incredible that the processing of an image can remove fog. The question still stands: how can fog, which acts like a vapour laden screen, be penetrated digitally after the fact. Unless this is simply another marketing ploy - Hitachi uses a very, very light fog and then claims to have magically beaten it.

Like I have already told you Stanislav, I read everything. I just don't believe everything I read.

Share this post


Link to post
Share on other sites

I suppose it is just manipulations with the Contrast and the Black Level. Plus Saturation. Brightness of the fogged image is shifted down, thus the fog level becomes the Black Level, then the Contrast of the obtained dark image is increased. If these manipulations is performed by 12-bit video processor, in some cases they can increase visually quality of the output 8-bit image. But these manipulations can't increase camera sensitivity with fixed Signal/Noise ratio.

 

I programmed similar image processing in VideoCAD but for the opposed purpose- modeling image distortions of real cameras in dependence of their parameters and scene conditions

 

The video on youtube was compressed therefore it doesn't show real difference. Don't believe your eyes

Share this post


Link to post
Share on other sites
I imagine I could do away with using IR and just keep my white light on overnight but I'd love to find a way to automate it.

Automate what exactly?

 

I just figure it's the price of doing business with IR. All my cameras have it and they all do the same thing. Not really so much of a bother, as heavy fog that causes it happens more infrequently than frequently. Cameras without IR, such as what I monitor at work- it just looks like fog usually does when visibility gets low. That's good, but they have other problems that my bullets and eyeballs don't, so I still favor IR cameras. And motion lights too of course.

 

I'd love to have firmware on my camera that recognized "whiteout" conditions and shut off IR. With the IR on even when my motion floodlight came on the camera stayed washed out due to the IR reflections mentioned in this thread. Once it got bright enough out to switch off the IR the picture improved considerably.

 

I need to test the idea but I think if in those conditions I shut off the IR and turned on the floodlight I'd get a better picture (for the reasons mentioned in the thread, light source further from the sensor). Camera and floodlight have the right inputs and outputs to do this, but firmware would need to recognize "potential whiteout, let's try alternative lighting scheme".

Share this post


Link to post
Share on other sites

May be a scheme like an "active infra-red sensor" will be suitable. It consist in two parts: a transmitter of modulated IR beam and a receiver. The worse transparence between the transmitter and the receiver is, the less signal level on the receiver. When the signal is less than the specified level, a relay will turn off IR build in cameras and turn on the floodlight.

Share this post


Link to post
Share on other sites
The worse transparence between the transmitter and the receiver is, the less signal level on the receiver. When the signal is less than the specified level, a relay will turn off IR build in cameras and turn on the floodlight.

 

 

 

wow its getting very technical.

 

you could always remove the subatomic hadron particle let the water dissolve the neutron frame work and block it with a nuclear filter.

 

 

but i have found it much easier if you just use a motion light with pir sensor. if its bright enough it will switch your ir off.

 

remember you are talking about fog. i dont see the point in modifying a camera just to deal with a problem that only happends with fog

Share this post


Link to post
Share on other sites
The worse transparence between the transmitter and the receiver is, the less signal level on the receiver. When the signal is less than the specified level, a relay will turn off IR build in cameras and turn on the floodlight.

 

 

 

wow its getting very technical.

 

you could always remove the subatomic hadron particle let the water dissolve the neutron frame work and block it with a nuclear filter.

 

 

but i have found it much easier if you just use a motion light with pir sensor. if its bright enough it will switch your ir off.

 

remember you are talking about fog. i dont see the point in modifying a camera just to deal with a problem that only happends with fog

 

It happens with snow as well, but to a lesser extent. People also leave tracks in the snow so it's a little easier to track down the typical neighborhood thief trying car doors in the snow than in the fog. Plus this winter I've had as many foggy nights as snowy nights, usually there's a lot more snowy nights than foggy nights.

Share this post


Link to post
Share on other sites
May be a scheme like an "active infra-red sensor" will be suitable. It consist in two parts: a transmitter of modulated IR beam and a receiver. The worse transparence between the transmitter and the receiver is, the less signal level on the receiver. When the signal is less than the specified level, a relay will turn off IR build in cameras and turn on the floodlight.

 

Google for "smart IR" or "intelligence IR"

lots of them around

They automatically adjust the infrared light required in a scene as the subject moves closer to or further away from the camera, resulting in a image that is not washed out or too dark.

Share this post


Link to post
Share on other sites
The worse transparence between the transmitter and the receiver is, the less signal level on the receiver. When the signal is less than the specified level, a relay will turn off IR build in cameras and turn on the floodlight.

 

 

 

wow its getting very technical.

 

you could always remove the subatomic hadron particle let the water dissolve the neutron frame work and block it with a nuclear filter.

 

 

but i have found it much easier if you just use a motion light with pir sensor. if its bright enough it will switch your ir off.

 

remember you are talking about fog. i dont see the point in modifying a camera just to deal with a problem that only happends with fog

 

pir sensor will be not confidently switched by fog and snow. Also the illuminator will not confidently switch of the IR built in cameras. The illuminator will produce additional reflection.

We also need to turn on the floodlight. Active IR seems the best. There are a lot of such schemes.

Share this post


Link to post
Share on other sites
pir sensor will not confidently switch by fog and snow.

 

 

i did state motion detect.

 

but with a normal camera with built in IR there is not much you can do anyway (another reason why you dont use built-in ir)

Share this post


Link to post
Share on other sites
pir sensor will not confidently switch by fog and snow.

 

 

i did state motion detect.

 

but with a normal camera with built in IR there is not much you can do anyway (another reason why you dont use built-in ir)

 

We can open the camera and build a relay for IR, but you'll write it is very technical . Generally it is disadvantage of built-in ir, I agree.

Share this post


Link to post
Share on other sites
May be a scheme like an "active infra-red sensor" will be suitable. It consist in two parts: a transmitter of modulated IR beam and a receiver. The worse transparence between the transmitter and the receiver is, the less signal level on the receiver. When the signal is less than the specified level, a relay will turn off IR build in cameras and turn on the floodlight.

 

I've been thinking on using an Arduino for flexible external IR control, and the idea of a transmitter/receiver to determine how it's controlled is an interesting one. The main problem would be where to put the receiver that would allow it to see the external conditions and still be easy to connect to the system. Along the wall would be the easiest, but wouldn't necessarily show the conditions along the camera line of sight.

 

Another option would be side by side transmitter/receiver pair that receives reflected IR from the ambient (similar to a proximity sensor), and once it gets too high, would shut off the IR and trigger the floodlight. You'd have to figure out when to turn it back on again; possibly a photocell to detect when the floodlight turns off, assuming it's on a timer.

 

A simple, non-intelligent way would be to ramp the IR up and down constantly, using a duty cycle that ensured multiple frames captured during the ramp at a given frame rate, taking into account the speed that particular camera changes exposure settings. Your effective frame rate would drop, depending on how much of the IR ramp gave you a useful exposure; outside of this range you'd either be too dark or too bright.

 

You'd probably need a slow duty cycle at night - a few seconds - due to the lower frame rates and the camera's exposure adjustments. This should be easy enough to breadboard and test. I've been considering this for license plate reading using external IR as a way of dealing with the IR washout, but you could tie it in to the onboard IR with a little hardware hacking.

Share this post


Link to post
Share on other sites

I've been thinking on using an Arduino for flexible external IR control, and the idea of a transmitter/receiver to determine how it's controlled is an interesting one. The main problem would be where to put the receiver that would allow it to see the external conditions and still be easy to connect to the system. Along the wall would be the easiest, but wouldn't necessarily show the conditions along the camera line of sight.

I suppose you need not microcontroller to make this device. It can be made by several CMOS chips, transistors and relay. The simplest way is using ready photosell pair for a gate or a bar. You should adjust (decrease) receiver's sensitivity.

Another option would be side by side transmitter/receiver pair that receives reflected IR from the ambient (similar to a proximity sensor), and once it gets too high, would shut off the IR and trigger the floodlight. You'd have to figure out when to turn it back on again; possibly a photocell to detect when the floodlight turns off, assuming it's on a timer.

I suppose in this case you will need the separated modulated IR transmitter/receiver too. Using a mirror allows to place transmitter and receiver on the same wall.

A simple, non-intelligent way would be to ramp the IR up and down constantly, using a duty cycle that ensured multiple frames captured during the ramp at a given frame rate, taking into account the speed that particular camera changes exposure settings. Your effective frame rate would drop, depending on how much of the IR ramp gave you a useful exposure; outside of this range you'd either be too dark or too bright.

 

You'd probably need a slow duty cycle at night - a few seconds - due to the lower frame rates and the camera's exposure adjustments. This should be easy enough to breadboard and test. I've been considering this for license plate reading using external IR as a way of dealing with the IR washout, but you could tie it in to the onboard IR with a little hardware hacking.

Interesting idea, but the problem is the exposure adjusting time as you mentioned. It will make terrible image from the camera.

Share this post


Link to post
Share on other sites

I think there probably two main reason effect it one is the case is not tightly to seal case so the cold air into the case and you know IR to heat the cold air caused the fog and another worst is the temperature is too high to melt sin (becasue some crappied supplier for saving cost to use bad rawmaterial.

 

If is the first you could try open the case make it dry and then seal it tightly ( maby use silicon or something you could) then make pilot run in the house if it happens I htink you beter send it back to fix.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×