Monday, July 26, 2010

60 % at 10 o'clock, or how can you name your scanned detail images ?

This detail map (click to enlarge) shows the railway line of the Kristianstad-Hessleholms railway company. It is a vignette on a 1906 share of the Kristianstad-Hessleholms Järnvägsaktiebolag. The railway company was founded in 1863 and nationalized in 1944.

I often scan details of certificates, but how do I name these images ? The scanned image of a whole certificate gets the same filename as the ID - identification number - of the certificate in my database. In this case, the certificate has an ID of 1405.  So, the image has 1405.jpg as its filename being the name of the certificates ID and the file type "jpg" in this case.  

Scanned image files of details on certificates also need a filename. This is how I do it. Just take a look at the example of the Swedish railway certificate.

  1. First, we need to position the vignette from a reference point.  I use the center of the certificate, which is at the crossing of the bisector lines. Now, imagine a clock in the center of the certificate.

  2. We position the vignette as a combination of a direction (an hour on the clock) and a distance of the center of the vignette from the center of the clock (0%) to the border (100%).
So, our detail image can be "positioned" at 10 o'clock - 60% . The filename of the resulting scan will be 1405_10_60.jpg, which is the outcome of the certificate ID, the hour position and the distance % and the file extension jpg.

Purists, can question this example and argue that 1405_1030_55.jpg is more accurate. However, it is an approach that has worked well for many years now, and, when ordering the files by filename in your file folder list, detail images get sorted along with the images of the whole certificate.

So, that's how I do it. I'd happy to learn from other approaches. In case you'd wonder, here is the whole certificate's image.



2 comments:

  1. Dear Franky (sorry I mispelled you name in the last post).

    I name my scans like this:

    US- The Alton and Chicago Railway 1889.jpg

    The country, the name of the company and the issue date. If I have a detail like a signature I do:

    US- The Alton and Chicago Railway 1889 Signature of Joe Blow.jpg

    I tend to spell things out as completely as I can so that my search engine on the computer can find things.

    Interesting post by the way.

    Tim W.

    ReplyDelete
  2. Hi Tim,
    Sure, you can add the name of the organization and the person that signed the document to the file name of the scan. In case of signatures it makes sense to do so, because it works relatively unambiguous.
    In case of vignette scans, I uncoupled the description of the vignette from the name of the scan because the naming process can be quite challenging and the search for images can fail to return all relevant matches due to the usage of synonyms .
    Several vignettes with the same subject can appear on a certificate. Imagine a certificate with a small vignette of a train on the left part of the certificate, a small vignette of another train on the right part, and a larger vignette of a train in the middle. Composing the file name of each scan can become a challenge to distinguish each scan sufficiently.
    Further, unintended use of synonyms will not return all matches when searching for an image. Suppose, I scan a vignette of a cow near a railway and save the image as ‘cow_near_railway.jpg’. Six months later, I query my scans for animals near railroads. The query action will not come up with that image because its name contains ‘cow’ instead of ‘animal’ and ‘railway’ instead of ‘railroad. That is why I uncouple the description of the vignette from the name of its scan. I put the description of the vignettes in the certificate’s database record.

    What about vignettes on the rear ?
    I use the same approach to name their scanned images but I add the word ‘rear’ to the filename.
    An example : ‘1405_10_60_rear.jpg’.
    Signatures can also appear on the rear, so you could end up with multiple scans of signatures on different locations. To solve the image naming question, you could follow the same approach.
    Example :
    myRailwayCompany_nameOfSignature_10_50.jpg ( a signature on the front) and myRailwayCompany_nameOfSignature_5_80_rear.jpg (same signature on the rear).

    Regards,
    Franky

    ReplyDelete