For a recent project, I accessed imaging requirements specific to usability (image quality) vs the footprint of the image (size). Prior to the project starting, the client had conducting image quality testing and stated that TIFFs with maximum compression and low image quality resulting in a usable image 30k in size. We conducted some testing and found that the clients tests were in accurate and their sampling of fiche was to small to represent the full spectrum.
 
For those without imaging experience, image format and quality settings affect the image size. For example, if you scan a letter in black and white, you could probably squeeze the file down to 30k. If you had to support grey scales the image might require 200k. Why the concern? Imagine having 20 million items to scan – 30k * 20 mil vs 200k * 20 mil. Consider this…do the math and then consider the impact on a Storage Area Network (SAN). 2 TB vs 10 TB of space – including a safety margin.
 
With any microfiche conversion project, the task of Fiche cleanup and scanning is quite the task. Depending on your approach it could take 2-3 years + to scan all the fiche into the Records Management System. Some client try to avoid this work by conducting a risk benefit analysis to determine the risk of not scanning all the images and just active records or the scanning of new records only – from an agreed upon date forward.
 
Given the range in quality due to image deterioration (some from the 70s) and photography consistency (camera operator sloppy), images are mostly yellow and shadowed. As you probably have guessed, grey scale support is required and that means a larger footprint for the image. Why? We tried TIFF with maximum compression and image pixelated and for the most part was black and not usable.
 
Another aspect of assessing image format is to consider the viewer required to support some of the more advanced image compression types. For example, the format JPEG within TIFF is supported by the native Windows image viewer since Microsoft doesnt license the Kodak viewer since Windows 2000.
 
In the end, our testing found the fiche required TIFF in JPEG compression that resulted in 200k image on average. The SAN sizing had to increase 4-5 times in size which affected the number of drives required and storage cabinets. The moral of the story? Do your testing! Take a wide sample of images from the oldest to the newest. Conduct usability testing by having users read the scanned images on a computer and paper to verify they can be read.
Advertisements