Harvesting the Fitzwilliam

This project is now complete. These pages are provided for reference and are no longer updated.

Contents

Partners

Project Cluster Group

 

 

Coin Imaging Procedures

The major steps in scanning coin images are:

  • Scan obverse of a group of coins as a one page image
  • Scan reverse of same coins as a one page image
  • 'Pick' single coin images off each of the one page images
  • Rotate each coin for correct orientation
  • Crop each coin image to remove excess whitespace
  • Rename image files to reflect the coin's object number

At this point the images are the 'unadjusted master set' of images. Using a copy of 'unadjusted masters' manual processing can be carried out (e.g. on colour images or greyscale images).

Another automated process is used to create a set of images (from final adjusted colour images) used for the Museum's collection management system, ADLIB (these images are also the images which are displayed via the Museum's OPAC website)

The following sections give more detail on each of the steps outlined above.

Page Scans

Temporary Numbering

In order to be able to physically keep track of coins as they are being handled for scanning it is practical to assign each coin a temporary number. The temporary numbers are also used to name the coin image files throughout the early stages of the imaging process. The temporary numbering scheme is influenced by the EyeBatch program which does provide automatic appended numbers on each of the 'picked images' (more on that in Coin Image 'Picking').

So, temporary numbering is designed to support the physical and processing needs of imaging coins.

The general format of a temporary coin number is XYY_ZZZ.tif where:

  • X is a letter representing the series being scanned (e.g. R for Roman)
  • YY is the scan number (e.g. 84 is the 84th page scan of 20 coins in this series)
  • ZZZ is a row/column number based on position on scanned page (e.g the table below demonstrates the numbering for a 5 x 4 (20 coin) scan
000 001 002 003
100 101 102 103
200 201 202 203
300 301 302 303
400 401 402 403
  • the reverse scan is identified by adding '_r' to the temporary coin number (e.g XYY_r_ZZZ.tif)

There is a numbering spreadsheet in S:\Coins & Medals\scantools\numbersheet (s:\ = \\heffa\shared) designed to assist with the use of temporary numbers. In order to keep track of coins it is often helpful to put a copy of the temporary number firstly in the coin tray and secondly on the scanner bed itself (it is necessary to put the number behind the coin, as any numbers showing up on the scans will stop the crop batch process from working properly.)

Page Scanning a Group of Coins

The coins are laid out on a lightbox , in order to ensure a uniform white background. Foam disks are then used to support an upturned flatbed scanner and prevent it from touching the coins. It is important to keep the scanner and lightbox free of dust as this will show up on the scans.
Placing a grid on the light box is helpful for the laying out of coins. This grid should be constructed relative to the size of the coins to be scanned (ie 4x5). This grid must not be visible on the scans-( a grid drawn in biro on an over head acetate works well. ) For EyeBatch to work, each coin must be placed in directly in the centre of its square and the squares must be equal.
When selecting the final area to be scanned, attention must be paid to selecting the outside of the grid. I.e. EyeBatch cuts the grid into squares starting from the outside edge of the scan area, so extra space around the scan will distort the results. It is helpful to place a visible marker (such as the discs used to support the scanner) at the outside edge of the area to be selected, These markers must be erased from the scan in PhotoShop before further processing.

Coin Image 'Picking'

Coin image 'picking' is the process of creating an individual file for each coin from a page scan of a group of coins. EyeBatch processing software is used to do this. EyeBatch has an operation called 'slicing' which is able to 'pick' multiple images off a single image based on a set grid. For example, if 20 coins are laid in a 4 x 5 grid on the scanner and scanned into a single image file then EyeBatch can be told to 'pick' images off the single image based on a 4 x 5 grid division of the image; creating 20 separate image files.

EyeBatch automatically numbers the separate files based on row and column numbers (e.g the table below demonstrates the numbering for a 5 row x 4 column 'pick') appended to the source file name. For example if the input filename is R70.tif 20 image files will be created names R70_xxx where xxx equals the number corresponding to the table below.

000 001 002 003
100 101 102 103
200 201 202 203
300 301 302 303
400 401 402 403

Step-by-Step Guide

1. With EyeBatch open, chose the ‘Select Images’ tab. Click on the folder containing the scans to be processed. They will appear in the ‘Image file’ column.
2. Click on the double blue arrow to move the scans into the ‘Selected Images’ column.
3. Click on the ‘Batch settings’ tab. Chose the ‘Load’ button and then select the script relevant to the grid you have used on the page scan ( eg. ‘4by5pick.ebs’) , and then click open to load the selected script
4. Click the ‘Process’ button. A dialogue box will prompt you to select a destination file. Browse to a suitable subdirectory (ie create a folder called ‘Sliced’ within the folder containing the scans.) Press ok and wait a few moments for the process.
5. Check the results using XnView

Resources

Scripts are 4by5pick.ebs, 7by5pick.ebs and 2by2pick.ebs.
EyeBatch scripts stored locally on network at S:\Coins & Medals\scantools\ebscripts
EyeBatch website http://www.atalasoft.com/EyeBatch/index.asp

Coin Image Rotation

The individual coin image files are taken into photoshop and manually rotated until the coin is oriented properly in the image.

Coin Image Cropping

Coin image cropping removes excess whitespace from around the coin in the image. This process is automated by EyeBatch which does an 'autocrop' at 3% sensitivity and then increases the canvas size by 2%. Script is crop.ebs.

Step-by-Step Guide

1. With EyeBatch open, chose the ‘Select Images’ tab. Click on the folder containing the images to be processed. They will appear in the ‘Image files ‘ column.
2. Click on the double blue arrow to move the scans into the ‘Selected Images’ column.
3. Click on the ‘Batch settings’ tab. Chose the ‘Load’ button and then select the script crop.ebs, and then click ‘open’ to load the selected script
4. Click the ‘Process’ button. A dialogue box will prompt you to select a destination file. Browse to a suitable subdirectory (ie create a folder called ‘Cropped’ within the folder) .Press ok and wait a few moments for the process.
5. Check the results using XnView. If any dust etc. is on the scans individual coins may not have cropped properly. These can be easily opened in Photoshop and cropped manually.

Coin Image File Renaming

Coin image file renaming renames image files from the temporary form (e.g R70_002.tif and R70_r_002.tif) to the object number form (e.g. CM.28.113-1920(1).tif and CM.28.113-1920(2).tif). The process is semi automated.

Firstly a cross reference sheet ('XRefSheet') in a spreadsheet is filled out with the object numbers against each temporary number. A copy of this information is then cut and pasted into an input data file (crbdata.txt). The processor (createcrb.php) creates an output file (rename.bat) which is a command line batch file with a separate rename command for each image file. rename.bat should be copied into the directory where the images to be renamed reside and then run (double-clicked).

An example...

The data from the spreadhseet which is cut and paste from the spreadsheet to the input data file (crbdata.txt) will look like:

R78_000 CM.RI.459-R

Note that no reference to obverse or reverse is necessary as the processor (createcrb.php) assumes R78_000 (the obverse image) will have a reverse partner image named R78_r_000. The .tif image extensions are also assumed by the processor. When the processor is run it will create two lines for the above input data:

ren R78_000.tif CM.RI.459-R(1).tif
ren R78_r_000.tif CM.RI.459-R(2).tif

When this batch file (rename.bat) is copied into the appropriate directory and run the two 'ren' commands actually perform the renaming of the files.

Technical Notes

The PHP scripts require php.exe and php4ts.dll (v4.3) to run. To enable the user to double-click the scripts to run them, windows should have a file association set so that *.php files are run by php.exe.

Colour Image Adjustments

A copy should be made of the final 'unadjusted masters'. This set should be adjusted in PhotoShop, typically using ‘Adjust Levels’ and ‘Unsharp Mask’. See \\Heffa\Shared\Coins & Medals\Documentation\SEW24 Documentation Manual.doc for further details.

Producing ADLIB Images

Each original image is used to create two images for use in Adlib. The first of these is a lower resolution JPEG of the original TIFF - not bigger that 360 x 380 pixels, the second is a 100x100 JPEG thumbnail. These can be automatically created from the adjusted colour TIFF using the EyeBatch script cmadlibimages.ebs.
(See Step by Step guides to using EyeBatch above- this time in Step 5 chose cmadlibimages.ebs.)

 

 

Contacts, Page modified date: 22Dec04