Peter -
I have the programming guide for the
Gretag/Macbeth Spectrolino & Spectroscan. It describes the serial I/O
communication for the Spectrolino, including commands and responses. The
Spectrolino itself is compatible with the communications protocol used for the
SPM spectrophotometers. It's a PDF file. I'd be happy to email it. I may also
have somewhere a programming guide for the ColorSavvy ColorMouse (CM2). If
you're interested, I will look for it. Finally, you can usually get this
information by simply contacting the vendor's technical support folks
directly. Gretag and Xrite are probably the two most common device vendors in
use.
Just out of curiosity, can you tell me
what sensor you found? There are a number of low-cost color sensors out there;
filtered photodiodes and photodiode arrays. TAOS (Texas Advanced
Optoelectronic Solutions) has a selection of light-to-voltage and
light-to-frequency devices spun out of Texas Instruments' mid-90s
optoelectronics research. A number of their sensors are already in use for
robotic color vision, monitor calibration, and spot-color matching. Parallax
sells a simple "colorimeter" that is a TAOS TCS230 development board that
interfaces to a Microchip PIC microcontroller (BasicStamp). Can be used to
match sample colors to a lookup table (there is an example on their Web site
of an M&M sorter). If you wanted to get more sophisticsated, you could
correlate the sensor output to actual colorimetric values (ex. XYZ) but I'm
not sure if it would be sufficiently accurate or repeatable to be
useful.
Have you seen any of the papers
describing the ColorSavvy ColorMouse? It's a very simple (in terms of
hardware) device that employs a VIS photodiode, NIR filter (to improve the
SNR), and a ring of multi-colored LEDs (red, green, blue, green/yellow,
orange, yellow, etc.). The photodiode responds to the amount of light
reflected back off the target for each LED color, then approximates the
spectral reflectance of the target. From this, colorimetric data is derived
(XYZ or Lab). I have a couple of papers that describe the math, if you'd be
interested. Viptronics in the UK has a similar device that uses a photodiode
and multi-colored LEDs.
Xerox (as I recall) recently received a
patent on the use of a photodiode with three LEDs - red, green and blue -
housed in a single module and used to auto-calibrate a color printer. This is
the simpliest approach; measure RGB reflectance with the photodiode, and apply
a 3x3 matrix and gamma correction to derive XYZ. ColorSavvy and Viptronics
added additional LED colors to attempt to improve device accuracy; ColorSavvy
attempts to approximate the spectral reflectance of the patch using (as I
recall) nine data points (which is not a lot). The TAOS demo board uses two
white LEDs and a TCS230 RGBI 64-element photodiode array, sort of the opposite
of the Xerox approach. TAOS I think has the right idea, because the white
LEDs will generate more signal for the photodiode to "see" than
individual red, green or blue LEDs. ColorSavvy found that signal strength was
critical to ensuring accuracy and repeatability; because of the inherent
signal-to-noise ratio of the electronics.
Anyway, with so many sensors and
implementations of sensors in devices, there always seems to be something new
out there. So, that's why I'm curious as to what you found.
Kind Regards,
- Chris Brown
Is there anyone on this list
who can tell me where I can find
specifications of the communication protocol
between
a computer and any reflection
colorimeter.?
I believe that I
have sourced a sensor which would make
construction of a
colorimeter relatively simple for someone
with a little experience in construction of microcontroller-based
electronic devices (that's what I have - a little
experience).
I would like to build a spot-reading reflection
colorimeter
and it occurred to me that the best way to go
would be to
make the device emulate a widely-used colorimeter
for
which a fair amount of software is available
already.
So how about it? Can anyone point me in the
right
direction?
