Post by thread:How to tell the color of test results

Hi everyone I was wondering how I could tell the color (electronically) of different types of color coded test strips like humidity measurement (you know the ones that come with some IC's). I would need to test for a variety of colors: brown, purple, blue, pink, red to name a few. Does anyone know how this could be accomplished? I suppose I would need to measure reflected wavelength? Most sensors I have seen like TSL 235 light to frequency converter seem to measure intensity. I wonder if you could use some kind of a calibrated color wheel and test each color on the wheel against the test paper. It seems if you were to use the blue part of the color wheel and the test paper was blue a simple sensor like a TSL 235 might detect a comparison. What do you think? I may be headed in the wrong direction, any help appreciated. Another thought- Does the refractive index change with color? Thanks Jon

Jon, One of the "classic" methods for determining color is to reflect white light off of the object, send it through a color filter, and measure the intensity of the result. You do this for THREE different color filters, such as cyan, magenta, and yellow, or red, blue, and yellow. The color is then determined by the ratio of the three results and the absolute intensity readings. For example, green will have a fair amount of transmission through both the blue and yellow filters, but near zero transmission through the red filter. Thus the ratio tells us we have lotsa_blue+lotsa_yellow+no_red=green. Ah, but what *shade* of green? That is where the absolute values are important. Bright green will have larger absolute values than dark green. To identify a rather fixed set of colors is not too hard. You measure out each one and then you have a table of values. You compare a color set against this table, and go with the closest match. Instead of using white light and filters, some designs use a red, a blue, and a yellow or green LED. (Sometimes TWO blue LEDs. Each color must be adjusted to give essentially equal amounts of light, but of differing wavelength). The LEDs are turned on one at a time and the relative reflected light for each is recorded and compared against a color table of known colors. If your colors to be tested are saturated you will get excellent results. If you use washed-out colors, then there is not much difference between colrs, and identifying them becomes more difficult. Fr. Tom McGahee ---------- {Quote hidden}> From: Jon Petty <spam_OUTPhxsys3TakeThisOuTAOL.COM> > To: .....PICLISTKILLspam@spam@MITVMA.MIT.EDU > Subject: How to tell the color of test results > Date: Wednesday, March 31, 1999 1:21 AM > > Hi everyone > > I was wondering how I could tell the color (electronically) of different types > of color coded test strips like humidity measurement (you know the ones that > come with some IC's). > > I would need to test for a variety of colors: brown, purple, blue, pink, red > to name a few. > > Does anyone know how this could be accomplished? > I suppose I would need to measure reflected wavelength? > > Most sensors I have seen like TSL 235 light to frequency converter seem to > measure intensity. I wonder if you could use some kind of a calibrated color > wheel and test each color on the wheel against the test paper. It seems if > you were to use the blue part of the color wheel and the test paper was blue a > simple sensor like a TSL 235 might detect a comparison. > > What do you think? > I may be headed in the wrong direction, any help appreciated. > > Another thought- > Does the refractive index change with color? > > > Thanks > > Jon

Off the top of my head and without understanding any of the implications, could you not use a CCD camera? I suspect you may be able to get some kind of RGB signal from it that may be useful. Or may be some LDR's with different coloured filters in front of them perhaps. Just a thought, though I could of course be mad. kind regards, SW. ____________________Reply Separator____________________ Subject: How to tell the color of test results Author: pic microcontroller discussion list <PICLISTKILLspamMITVMA.MIT.EDU> Date: 31/03/99 01:21 Hi everyone I was wondering how I could tell the color (electronically) of different types of color coded test strips like humidity measurement (you know the ones that come with some IC's). I would need to test for a variety of colors: brown, purple, blue, pink, red to name a few. Does anyone know how this could be accomplished? I suppose I would need to measure reflected wavelength? Most sensors I have seen like TSL 235 light to frequency converter seem to measure intensity. I wonder if you could use some kind of a calibrated color wheel and test each color on the wheel against the test paper. It seems if you were to use the blue part of the color wheel and the test paper was blue a simple sensor like a TSL 235 might detect a comparison. What do you think? I may be headed in the wrong direction, any help appreciated. Another thought- Does the refractive index change with color? Thanks Jon

Use the Texas TSL401, a 128 bits serial output photosensor, DIP8 package, and a color filter with all the colors you want in a continuous strip in front of it. 10 colors would use aprox 12 bits for each color, plenty to check which set of 12 bits has higher output level. The problem would be to produce this tiny filter set... the DIP8 has less than 10mm of visible window, probably the use of a lens system to expand the area of the filter. Electronic wave length is also possible, more expensive. It is used to recognize different gases, based on the fact that they reflect diferent wave length. A wide spectrum photo sensor with several electronic filters can do the same job, but... lots of work.

-----Original Message----- From: Wagner Lipnharski <.....wagnerlKILLspam.....EARTHLINK.NET> To: EraseMEPICLISTspam_OUTTakeThisOuTMITVMA.MIT.EDU <PICLISTspam_OUTMITVMA.MIT.EDU> Date: Wednesday, March 31, 1999 10:29 AM Subject: Re: How to tell the color of test results >Use the Texas TSL401, a 128 bits serial output photosensor, DIP8 >package, and a color filter with all the colors you want in a continuous >strip in front of it. I understand that another manufacturer has taken over production of the Texas Instruments photosensors. TI lists this device as obsolete. Does anyone know who is now manufacturing these? Michael