Page 1 of 1

Monitoring a Stop Light.

Posted: Tue Jun 30, 2009 3:24 pm
by kylejohnson
Say that I have two cameras. Camera 1 is zoomed into a stop light, with (mostly) the 3 lights (green, yellow and red) in view. Camera 2 is pointed at an intersection.

I want to record camera 2 when camera 1 detects that the light has turned red, and stop recording when the light turns green. I have camera 2 set to nodect and linked to camera 1, which is set to modect.

What I need help with are the zones on camera 1... Dealing with false positives is tricky because recording needs to happen continuously while the light is red, and stop when the light turns green and I assume the only way to make this happen will be with some zone-type (inclusive, etc) magic.

How could I make this happen?

I do not think that I need 'color detection' as, by defining 3 small zones - 1 for each light - zm should be able to detect 'motion' when the colors change.

What are your recommendations? Any ideas? Thanks!

Posted: Tue Jun 30, 2009 8:08 pm
by kylejohnson
I've made some progress...

I have set an active zone on the red light and a preclusive zone on the green light.
Then using a high (60) reference blend image, recording is pretty accurately triggering when the light turns red, and not when the light turns green (which I assume is thanks to the preclusive zone) - which is correct.

Now my problem is that, while recording is taking place when the light turns red, I need it to take place WHILE the light is red. So how would I define my zones to record when the light turns red, and stop when the light turns green? Somehow I want to start recording when motion is detected in the Red zone, and continue recording until motion is detected in the Green zone.

Image

Posted: Tue Jun 30, 2009 10:33 pm
by coke
There's probably a clever way to do that, on the other hand, how often over the course of a day does the light change the amount of time it's red/green? Is it run from a traffic sensor in the pavement?

Where I am, the lights are pretty much consistently a certain amount of time, though they adjust it somewhere inbetween coming into work and leaving it.

Posted: Wed Jul 01, 2009 1:14 am
by kylejohnson
That is a good point. I was monitoring the camera all day and did notice that the red and green times were not very consistent - we do not install the lights, so I am not sure what determines the length, but I am willing to bet there is a sensor or video camera changing it dynamically.

I will work on finding the range of the duration of the lights, but have a feeling that it will not be a very accurate solution.

Status of your project?

Posted: Fri Jan 22, 2010 10:36 pm
by caribeanstefan
Hi kylejohnson,
What is the status of your project, did you make it work?

I am very interested because I want to deploy a similar setup, if possible with only 1 camera.