Is anyone else having this problem

[hpman247]

Our team hooked up the green lights, however, the camera picks the light up as white light, except for the green tint around the edges. We are using the software that allows us to grab a frame and tell it to track. I don't know if that is the same as Kevins. THe camera also locks onto the flourescent light in the room. They are both white. I don't really understand this. Any ideas.

Also, after we played with the cam for a while today it would move sideways wonderfully but not up and down very well at all. what can cause this??

thanks for your help.

[Alan Anderson]

The camera tracks colors much better in YCrCb color mode. That's what the provided LabView calibration uses, and that's what Kevin's camera code uses. Fluorescent lights are sufficiently more blue than the goal target, and gymnasium-style vapor lights are more red, so the camera can discriminate well between them.

(Sunlight still tricks ours on occasion, with the camera locking on to a window off to the left side of our test bench. I found a way to fix that, but only at the expense of not being able to find the green target more than about 30 feet away.)

[hpman247]

I dont really understand why the camera "sees" the green light as white. The floruescent lights appear as white also. This seems to be causing the problem.

[Alan Anderson]

Change the color mode from RGB to YCrCb and try again. You'll have much better results.

[JJMax7]

also lower the brightness that seemed to work for us very well.

[devicenull]

If you actually read the workbook, it tells you how to fix this problem.

http://kevin.org/frc/CMUcam2_workbook.pdf

[Eldarion]

Originally Posted by JJMax7

also lower the brightness that seemed to work for us very well.

Technically it's the saturation. Setting it to 2 works very well.

[Keith Watson]

In the locking code we changed it from using the location of the "target centroid" to using the confidence value. In testing we found low hanging flourescent lights are picked up in the mx,my values but the confidence for those lights is near zero. But the green target has very high confidence.

[Mike223]

Originally Posted by JJMax7

also lower the brightness that seemed to work for us very well.

Change the brightness of the light or the brightness the camera sees? If the camera how do you change it?

[CaseyKasem_1251]

Originally Posted by Mike223

Change the brightness of the light or the brightness the camera sees? If the camera how do you change it?

Im pretty sure they are referring to the camera and you can change the brightness of the camera through labview. (correct me if im wrong)

[sullivan_owns]

what i would do is try the YCrCb mode, try it again, you should end up with better results, hope it helps