Subject:
|
Re: Black
|
Newsgroups:
|
lugnet.robotics.rcx
|
Date:
|
Sat, 29 Jan 2000 20:58:58 GMT
|
Viewed:
|
2569 times
|
| |
|
|
In lugnet.robotics.rcx, Steve Hodge writes:
> Keep in mind that the light sensor generates a red light, so you're really
> measuring the reflectivity of the target in the red part of the spectrum>
True, I was shocked at all the colors that were "white". I think part
of this is due to the poor quality of the construction paper. Close
examination indicates a fair amount of unstained fiber, for instance the
black read a lousy 47, compared to 31 for my India Ink lines on
newsprint. The idea of just cutting out patches of construction paper
and using them for floor markers looks less viable than it did.
It does indicate the great problem of what a sensor sees vs what
a highly adaptable human eye sees.
I did more testing. I have used a carefully selected Radio Shack CDS
cell. I have noticed that not all Radio Shack cells in the five pack
are necessarily equal. I found one really great one about 1/2 inch:
see www.aloha.net/~mburger
to make my own unpowered primative light sensor. I compared it to the
Lego one. Of course it is very sensitive to room light levels, but
if one is careful to use multiple light sources and bounce light off
a white ceiling to make a good soft steady light, this problem can
be minimized.
As expected it had a different response to lots of colors. I then
mounted mine right next to the LED side of a Lego light sensor at
the same height so both could see the LED illumination.
The CDS cell's definition of "white" increased. But it still responded
differently than the Lego sensor. This tempts one to think that a
homemade CDS cell sensor mounted next to a Lego sensor might make for
some reasonable color recognition, if ambient light is controlled and
rather shadow free. Also many colors indicated "more white" on the
CDS cell than the Lego sensor, so the CDS cell had readings both
higher and lower than the Lego sensor in this specific case.
But most important:
Black stayed black. That is truely black blacks, deep dark absorbing
blacks with no surface shine, like my India Ink soaked into newsprint
to make a very velvety black line, did not change nearly as much as
the white did with increased light. In one instance, the black stayed
at 27 on my sensor while my sensor with the same increased room light
(Lego sensor off) went from 70 to 90. Results were similar with the
Lego sensor powered. Really black black remained unchanged. Plastic
black tape, the horrible creepe photo tape, both changed quite a
bit with room light changes, due to reflectiveness of surfaces.
This indicates that a good velvelty matt black will make life
generally a lot simplier.
Also I tried the trick of partially blocking the LED on the Lego
sensor. I used a small piece of opaque vinyl wedged in the little
channel between the LED and the sensor, just a tiny piece to fill
up the channel. Yes a big difference in levels, but the range between
black and white hardly changed on mine, from 19 units to 22 units.
Also the stability of the white was seriously disrupted and the LEGO
sensor became more sensitive to ambient light by a disturbing amount.
A CDS cell next to a Lego sensor may be a viable approach to
limited color recognition.
Really black blacks and shadow free light from multiple light
sources make life easier, good blacks tend to stay black as light
levels change.
What looks obvious to the eye is a very different matter to a
light sensor.
|
|
Message is in Reply To:
 | | Re: Black
|
| (...) Keep in mind that the light sensor generates a red light, so you're really measuring the reflectivity of the target in the red part of the spectrum. Thus red is as reflective as white (some of the other results, e.g lavendar, are still pretty (...) (24 years ago, 29-Jan-00, to lugnet.robotics.rcx)
|
5 Messages in This Thread:
  
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
