Color Vision

Look up (TAOS) TCS230 for a colour sensor, or CMUcam and AVRcam for camera based solutions.

Dave

Posted Via Nuthinbutnews.Com Premium Usenet Newsgroup Services

---------------------------------------------------------- ** SPEED ** RETENTION ** COMPLETION ** ANONYMITY **

----------------------------------------------------------

Hi

Go to

Cheers

Dale

And I forgot to mention - be aware that just about all colour sensors are also sensitive to Infra Red, and additionally the filter responses for the individual colours also overlap to some extent.

If your colours are very distinct, then you might get away with a simple phototransistor with a colour filter in front - make three, one for each colour you want to detect.

Posted Via Nuthinbutnews.Com Premium Usenet Newsgroup Services

---------------------------------------------------------- ** SPEED ** RETENTION ** COMPLETION ** ANONYMITY **

----------------------------------------------------------

Thank you a lot for your replays, they are very helpful.

I want to ask about another thing, we build a white color detector using infrared trans/teceiver ,we emite infrared to the surface being tested and receive the reflected rays on the receiver and threthold it using a schmittriger. the reliablity of the sensor is not good as it not always work, and also the distance from which it could detect the color is variable and don't know how to calculate it. Any ideas.

Thank you all again

As you are discovering, making a detector like this is not completely straightforward. Think of the problem as maximising the signal/noise ratio. 'Noise' in this case comes from other sources of light and infrared. The 'signal' part is related to the amount of energy that you receive from the intended target - which is related to the amount that you send out, as well as the reflective properties of the target itself.

Actions you can take:-

Reduce the noise by filtering out as much unwanted light/infrared as possible.

Increase the signal by using a 'brighter' source, and also by making the target reflect more if possible (retro reflectors are especially useful)

Because of the variables mentioned above, using a simple level detector (Schmitt trigger) is rarely a good idea as the threshold will vary.

The 'best' solution is to pulse the transmitter on and off. If you measure the received signal with transmitter off, and then measure it again with transmitter on, the difference will tell you what is reflected and detected, and will ignore variations in room lighting etc.

Have Fun ! D

Posted Via Nuthinbutnews.Com Premium Usenet Newsgroup Services

---------------------------------------------------------- ** SPEED ** RETENTION ** COMPLETION ** ANONYMITY **

----------------------------------------------------------

And don't forget to calibrate your sensor. You'll need to subtract off the 'blackpoint' measurement, or the measurement when nothing is in front of the sensor.

Also recognize that IR us a limitation - not all 'white' object reflect IR the same as the rest of the visible light. it may be OK as an approximation.

Fo your white detector, why don't you use a white LED?

Jim

Thank you all for your help

Actually I want to detect the white color from a green background. Counting the white areas will give me the position of the robot. Any way I was thinking of using a votting system by using 3 sensor instead of one and vote them for the correct deceition and then but every thing in a closed box that remove any lite from outside. Second I can adjust the sencitivity of the sensor by controlling the output level using a variable resistor.

I want to ask about the filter, how can I filter the Input to the sensor. You mean using lenses or such things.

do you mean using a white LED instead of the transmitter, or I should use another receiver.

Thank you again