Hello,
I have a gas counter that has nothing to help me mesure its usage... no relexion on the 0, no pulse, no rotating wheel... just the numbers wheel running.
IS there a simple way to have an image recognition of a number, let's say for example the 0 that could be used to replace the pulse light missing ?
Thanks,
As far as I know, everyone uses the rotation wheel or a pulsing led or something like that.
Can you make a photo of it ?
It is possible to recogize the number, but you would need a (mini)computer for that.
Gas meters generally do not have an electrical supply , they are purely mechanical.
They are available but cost a lot, even if you can persuade th co to fit them.
If the dial is acessible you could put a dab o silver paint on the wheel and use an opto sensor, if its behind a window though i would not reccomend messing with a gas meter.
Its possible that just using an opto sensor may give something usable off the digits although some experimentation is required.
Boardburner2:
If the dial is acessible you could put a dab o silver paint on the wheel and use an opto sensor,
The dial is NEVER accessible as, if it were, you could turn it back and get you gas for free.
if its behind a window though i would not reccomend messing with a gas meter.
The gas supply company wouldn't be too pleased with you, as you could have stolen their gas.
Some meters have a small red pointer that goes around.
If you have something like that, you could try using one of these optical distance thingies (see also the picture at bottom of posting_
I've not tried this myself, but another thought I had was to use the optical detection system in a mouse. The new style mice seem to have a small camera which photographs the area under the mouse and performs calculations based on the differences in the images.
If you hold a mouse up to the meter and the mouse pointer moves as the dial rotates, it would be theoretically possible to use this via a USB Host sheild to read the movement data
Aside from that, full blown optical character recognition, you'd need a Raspberry PI + camera
Hello,
I have already explored all the above... sadly... here is a picture, it is mecanical without any interresting feature... and yes I shall not open it or pray and pay... ]
Yesterday I saw someone speaking of LineScan nig http://parallax.com/product/28317 but it looks like very complicated to do, or watching a precise pixel but here I haven't identified something that could do it.
You should be able to stick a very small optical detector to it (phototransistor), to detect the white stripes on the right.
If you use a blue or green led, the red gets darker, and I think you have to use a lense with the phototransistor.
Peter_n:
You should be able to stick a very small optical detector to it (phototransistor), to detect the white stripes on the right.
If you use a blue or green led, the red gets darker, and I think you have to use a lense with the phototransistor.
Hello,
Thanks for the idea, could this do it : Optical Detector / Phototransistor - QRD1114 - SEN-00246 - SparkFun Electronics or a classic TCRT5000L would be applicable ?
I understand the idea but I don't know what hardware I would need to achieve this...
Looking for ytour answer,
Emmanuel
This looks like it functions very similar to the module I posted.
BTW. The module I posted actually has a sensitivity setting. ie although its sold as a distance detection device, thats not how they appear to work
I have one of the modules I linked to, and they appear to work merely by detecting the intensity of the reflected IR light.
And have a comparitor chip and pot to set the threshold.
So they detect shiny surfaces at a larger distance than mat ones.
Its hard to know whether you could adjust the module to trigger with the difference between the white stipes on the meter, but those modules are dirt cheap, so it may be worth a go.
BTW. If you do use optical detection of the white stripes. You'd need to make sure you exclude all other light from the meter other wise you are likely to get false readings.
BTW. AFIK, you can also buy colour detection devices, but I'm not sure whether they would be of any use in this instance e.g. detect red vs white lines
epierre, Those optical detectors you mentioned have an angle that is probably too wide. You need something smaller, perhaps something with a lense. But I can't tell which hardware you need, I have seen some very small optical detectors in optical mice, but I don't know if that can be used.
Further to what I wrote above, there's nothing to stop you (other than cost or space) having a second, private meter fitted AFTER the supply company's meter, that you can get to the dial on. The readings on this may not be exactly the same as those on the supply company's meter, but should be fairly close. You will, of course, have to observe all the safety rules that apply in your jurisdiction.
Hi all,
Well I would prefer to try and learn... and I don't like playing with gas and some smart grid project may give me a. new one within 5 years...
I've also looked at the freescale/parallax linescan but too few things about its use, and the board added to it make it pricey... unless I find something on its use.
Isn't there a pixel reading sensor or more that could be used through a pinhole ? or addi g a pinhole to those discussed above to focus it on a smaller surface ?
A pinhole instead of a lense, that might be the best idea so far.
I can think of a package made of sealent with a green led, a pinhole and a phototransistor. With no influence of outside light. If I had to make it, I would make a few of those, to see which one is best in detecting the white stripes.
An optical mouse sensor reads 16x16 pixels (typical). But again it's a problem of getting the right focus on the dial. And finding a compatible mouse sensor 
Those numbers seem pretty clear to me, and you know exactly where they are and don't need to worry too much about handling variations. That ought to be relatively easy to recognise using any OCR system, just with a web cam and PC.
for the mouse approach, I'll investigate it.
for the ocr, that would be the rich man verion, but it is likely to be a good fall back.
for tbe linescan/pinhole I've stubled uppon this :446
-taos tsl201r 64x1 linear sensor or pinhole vishay TEPT4400 camera – Arduining
the latter is quite cheap and would be mosr fitted in my case. what do you think of that ?
epierre:
for the ocr, that would be the rich man verion
I don't see why there would be any significant expense. All it would need would be the use of a PC with a web cam, and wireless web cams are readily available and probably cheaper than an Arduino even before you consider the other hardware you'd need to make the Arduino version work.
for the ocr, that would be the rich man verion
No raraspberry Pi and a web cam would set you back about $35, you don't have to be rich.
