Comparative Scanning Of Coins For Variations?

Gincoin43


How is that done?

I have no clue.

Can it be done with coins saved as JPEG files or PNG files? All of my coin pictures are in one of those two formats.

Do I need new software?

Kcm I was actually focused on coin authentication and die typing more than grading. Grading is a swampy mire and I agree with you that a computer lacks the necessary ability to evaluate "eye appeal" because it is totally subjective.

Computer comparison would have the ability to display differences in the design of two coins proving they were struck from different dies. It is like scanning for a fingerprint authentication. It seems as if a computer would be tailor made to detect both Counterfeit coins and
die varieties of counterfeit or genuine coins.

Gincoin43 - You ask




My photographs are all either jpeg or png. I do not have an image editor on my laptop as far as I know. What would you suggest? If such a program may have come already installed on this laptop what would it be called?

@swamperbob,

Yeah, I guess I should try to be more explanatory in my posts. I got your drift, but failed to put my own within reach. When I used the terms "grade" and "grading," I had in mind all the numismatic properties TPG's concern themselves with except for eye appeal. As to that property, I wish they'd keep their opinions out of the mix. Many coins who've spent their whole lives on a battleground appeal to this man's eye more so than the lustrous cousins whose lives were spent in holders albums and safety deposit boxes.

Kevin

bjherbison Gincoin43 Thank you for the suggestion. I will try that program and see if I can get it to work on a simple overlay.

Kcm Your statement regarding coins from battlegrounds versus pristine ones from vaults actually resonates with my own beliefs. I have always had a dislike for proof or even MS coins because I see them as never having been "real" circulating money. On the other hand, I personally have always gravitated towards the counterfeit coins that circulated but are often found in very ugly conditions. Coin collecting is a broad field that can accommodate both views and certainly it is less expensive for me to indulge in my specialty than it is for people searching for perfect coins from the early 1800's.

The technology does exist. The question is time and money.

An instance that I am aware is used in Japan mint where optical scan is used to scan for coins that is not within acceptable strike. Therefore it is quite rare to find any mis strike or error coins from Japan mint. This will become more difficult as the scan is only optimized for UNC. With different wear grade, this will more difficult.

The only way I can see around this is that one creates an AI database and consistently train it. In theory, this would also make it possible for such machines to grade coins in the near future.

I installed GIMP last night and began experimenting with it. It is complex for me - I really do not see any organization yet in the controls. I try to find crop and I end up looking all over. It only gets worse as I go deeper into the program.

My pet project is right now Cap and Ray 8R counterfeits. So the comments below pertain to just those coins.

GIMP has a very interesting Measurement tool which allows me to level each die photo based on a plane that I can select for each die. The best plane for the Obverse die is the base of the cap corner to corner. From that plane I can determine the angle and distance to any feature on the obverse die. Comparing the angle and distance to ray tips is logical to compare die types.

The reverse die can be leveled with the water line on the hill upon which the cactus grows. Sharp features like wing tips - leaf tips - alpha features can map that side.

The key for general use would be selection of a standard plane that is absolute or two people might select slightly different planes.

I would opt for the base plane of the cap to be a line that touches (tangentially) the two lowest parts of the cap near the corners so that all of the cap sits above the line. This would exclude breaks in the bottom of the cap due to multiple strikes (or re-hubing) and also exclude die chips and breaks that effect the base of the cap.

The water line is a bit easier to define as most dies use a continuous level line for the water surface. If that line is NOT straight I would propose a line end to end as the plane.

In a former career (College and 4 years after) I was a land surveyor and this approach is old hat for me. So I actually could establish a survey of each die for comparison. Time is a consideration since it could take nearly one hour per side (unless as I suspect there are surveying programs that could speed the process.)

One large critical issue is creating pictures that are exactly the same size (scale). This in turn requires all pictures or scans to be done exactly perpendicular to the coin faces. (Perfect for a scanner only slightly more difficult for a tri-pod mounted camera).

Using an accurately measured diameter of the coin to standardize the pictures would be preferential but not 100% necessary.

An approximation could be done based on the supposition that the die face inside the arc of the denticals is a circle. This would allow pictures taken of coins by others to be mapped approximately. The first step would to adjust the vertical to horizontal ratio to form a circle then adjust the diameter to a standard size. The results might be different, however the pattern of results could be compared and the size adjusted to force a match of extremities.

Then each example scanned could be mapped and results compared.

At my age 73 it is a daunting task since my collection alone consists of over 4,000 examples.

So I guess it could be done - a ton of work and time.

So I have a new question.