Sorting coins by year - possible with photo/optical or laser scanning?

Thread Starter

RogueRose

Joined Oct 10, 2014
375
I have a pretty good understanding of how I would sort different coin denominations and even to their "production run" (meaning their years of production like 1948-1984 or 1999-current) as they can vary somewhat significantly by weight.

What I need to know is if there is a way to sort coins to their year and possibly even minting location. I would think that something like a laser or camera may work (take 1-4 images a second - 2 cameras - both sides), mixed with some OCR software which could orient the picture with a specific distinct spot on the coin. I would even bet it is possible to grade the coins to some degree in this process.

Can anyone tell me which method would be better - laser or photo, and any suggestions on how to handle imageing at speeds like this - say 4 per second.
 

Thread Starter

RogueRose

Joined Oct 10, 2014
375
I checked out one of those sensors and I have to say that seems like it would do the job nicely but may be a little more than I need.

I don't see why a something like the camera from an iphone (old one at that, or Samsung Galaxy, or most $50+ digital camera from the last 5 years) would be able to do this task at 4 pics/sec. Most cameras have an action setting that takes 3-7 shots per second at a less than maximum resolution (my old Casio 10.1 MP does this at a 5+MP resolution @ 5 / second). I know the Raspberry Pi camera module can do HD at 1080p @ 30fps and static imagry at 5MP. If video can be done at 1920x1080 @ 30fps and 640x480 @ 90 fps, then I should be able to get static images of those resolutions at those frame speeds - or at least 4 pics / second. Tell me if I am missing some glaring point in the technology to which I am ignorant which makes these video frames not equal to static stills in some way - or something else I'm missing.
 

Sensacell

Joined Jun 19, 2012
3,432
The amount of work involved in interfacing to a consumer phone or digital camera imaging device would be large and entirely not worth it.
Buy an off-the-shelf machine vision camera designed for the purpose.
 

GopherT

Joined Nov 23, 2012
8,009
The amount of work involved in interfacing to a consumer phone or digital camera imaging device would be large and entirely not worth it.
Buy an off-the-shelf machine vision camera designed for the purpose.
Do you know how much a machine vision system would cost (that can do this level of task at 4 coins per second). Just wondering.
 
National Instruments LabView probably has a vision add-on kit. The LabView language is a data-flow programming language, Your program looks a lot like a schematic. When all of the inputs to a vi (Virtual Instrument) has data, it executes. It's easy to do stuff in parallel.
 

mcgyvr

Joined Oct 15, 2009
5,394
I checked out one of those sensors and I have to say that seems like it would do the job nicely but may be a little more than I need.

I don't see why a something like the camera from an iphone (old one at that, or Samsung Galaxy, or most $50+ digital camera from the last 5 years) would be able to do this task at 4 pics/sec. Most cameras have an action setting that takes 3-7 shots per second at a less than maximum resolution (my old Casio 10.1 MP does this at a 5+MP resolution @ 5 / second). I know the Raspberry Pi camera module can do HD at 1080p @ 30fps and static imagry at 5MP. If video can be done at 1920x1080 @ 30fps and 640x480 @ 90 fps, then I should be able to get static images of those resolutions at those frame speeds - or at least 4 pics / second. Tell me if I am missing some glaring point in the technology to which I am ignorant which makes these video frames not equal to static stills in some way - or something else I'm missing.
Sure.. you could start with a PI/ PI camera module..
IMO getting the image is simple... Processing it and fast enough is where the difficulty comes in.
Are you very familiar with OpenCV/SimpleCV/Python? You are going to need to be.. :)

There are many ways (all with varying costs) to achieve your goal..
Like I said.. I would use an off the shelf system that I can pick up for around $1000..
One could easily spend that much in time developing something from scratch..
But with something off the shelf as much as 90% of the work is done and programming it to recognize what you want is mostly point and click with their easy software..
 

sirch2

Joined Jan 21, 2013
1,037
I assume this is a hobby project? if so do you really need 4 coins per second?

You can get data off a phone using bluetooth so if you did all the processing on the phone (they are quite capable processors) then tripping a solenoid based on a bluetooth signal would not be hard. But as said above it is going to take some time to program

A laser scanner for sub-mm measurements would be quite expensive but would give you easier to process data. From experience I would say that if you use a camera lighting will be an issue, to give contrast to the detail on worn coins is probably going to need the coin to be lit from the side but then the captured detail will depend on the coin orientation.
 

wayneh

Joined Sep 9, 2010
17,496
You're going to need to define an acceptable error rate, and save a sorting bin for the coins that cannot be sorted with your required level of certainty. I manually sorted a large penny collection not long ago. I used lighting from multiple angles and of course a magnifying glass. (A dissecting microscope with binocular vision would've been great.) Some coins still defied sorting.

An automated sorter would be interesting to coin collectors, since one of their methods is to get a bag of coins and search for the needle in the haystack.
 

tcmtech

Joined Nov 4, 2013
2,867
Can anyone tell me which method would be better - laser or photo, and any suggestions on how to handle imageing at speeds like this - say 4 per second.
How many coins do you have to sort? To me that's the #1 criteria for the feasibility of this concept.

Unless you work at a bank and plan to scan every single coin, if you were even allowed, that passed through their doors a day every day in order to find the rare ones I have serious doubts the whole process would be worth the time effort and costs involved.

Personally I suspect a person with a good eye that knows what they are looking for could easily sort through 1 - 2 thousand coins and hour and pick out the interesting ones easily enough.
 

wayneh

Joined Sep 9, 2010
17,496
Personally I suspect a person with a good eye that knows what they are looking for could easily sort through 1 - 2 thousand coins and hour and pick out the interesting ones easily enough.
I'll guess you've never tried this. o_O It's quite a bit more difficult than that, once you have narrowed the pile down the "interesting" pennies.

It's true that the bulk of random pennies you will find are post-1981 and not made of copper, just zinc with a copper coating. Only the one's with mint errors would be worth more than a penny. They have no melt value. I wouldn't bother reading them at all - just get them out of the way.

Copper pennies, (pre-1981 except the war-time ones) are worth at least about 2.5¢ for their copper content. It's theoretically illegal to destroy pennies but their value tracks the value of copper. 'Nuff said. It's worth discriminating copper pennies from modern ones.

The "Lincoln Memorial" pennies (after 1958) are rarely worth anything more than their copper value. This is, surprisingly to me, also true of most wheat pennies (pre-1958) as well. The chances of an old wheat penny being worth more than its copper is very low.

As you might suppose, value goes up with age and quality. But the biggest factor by far is original rarity. It requires reading both year and mint to determine that. As the quality drops, it gets harder and harder to read the year and mint.
 

wayneh

Joined Sep 9, 2010
17,496
You need a better lawyer. ;) There are exceptions (such as those machines), but in general you can't melt the small value coins. The larger value coins are protected by their face value.

From 31CFR part 82

§ 82.1 Prohibitions.
Except as specifically authorized by the Secretary of the Treasury (or designee) or as otherwise provided in this part, no person shall export, melt, or treat:

(a) Any 5-cent coin of the United States; or

(b) Any one-cent coin of the United States.

§ 82.4 Penalties.
(a) Any person who exports, melts, or treats 5-cent coins or one-cent coins of the United States in violation of § 82.1 shall be subject to the penalties specified in 31 U.S.C. 5111(d), including a fine of not more than $10,000 and/or imprisonment of not more than 5 years.

(b) In addition to the penalties prescribed by 31 U.S.C. 5111(d), a person violating the prohibitions of this part may be subject to other penalties provided by law, including 18 U.S.C. 1001(a).
 

tcmtech

Joined Nov 4, 2013
2,867
I had doubts on the picking solid copper pennies out or the masses for scrap metal value as well.

Given the low number of solid copper pennies still in circulation now doing scrap work for profit would be far more lucrative just walking roadsides and trash dumpsters picking up aluminum cans or getting permission to scrounge the landfills or construction site dumpsters for discarded copper wire and aluminum and what not.
 

mcgyvr

Joined Oct 15, 2009
5,394
You need a better lawyer. ;) There are exceptions (such as those machines), but in general you can't melt the small value coins. The larger value coins are protected by their face value.
My lawyer says that you left out 82.2 in your argument :p

82.2
The prohibition contained in Sec. 82.1 against the treatment of 5-cent coins and one-cent coins shall not apply to the treatment of these coins for educational, amusement, novelty, jewelry, and similar purposes as long as the volumes treated and the nature of the treatment makes it clear that such treatment is not intended as a means by which to profit solely from the value of the metal content of the coins.
 

wayneh

Joined Sep 9, 2010
17,496
My lawyer says that you left out 82.2 in your argument :p

82.2
The prohibition contained in Sec. 82.1 against the treatment of 5-cent coins and one-cent coins shall not apply to the treatment of these coins for educational, amusement, novelty, jewelry, and similar purposes as long as the volumes treated and the nature of the treatment makes it clear that such treatment is not intended as a means by which to profit solely from the value of the metal content of the coins.
Ummm...

"....as long as the ... treatment is not intended as a means by which to profit solely from the value of the metal content of the coins."

If you're buying pennies at face and selling them for their melt value, I think that qualifies. Not that the Feds would know or much care about you.
 
Top