Document Scanning and Image Quality

FileHold was recently speaking with one of our partners, Micro Com, who's business includes document scanning and imaging. They scan millions of pages a month on behalf of our customers, at their facilities, or at client sites. FileHold asked Micro Com to provide results of scanning 10 different invoices using ABBYY FlexiCapture software.

FileHold asked:

"It would be interesting to know if scanning and imaging results would improve with higher DPI resolution. At 300 DPI we got 7 out of 10 invoices fully captured using ABBYY FlexiCapture software. What would happen if the images were scanned at 500 or higher? It seems to me prospects might be willing to buy higher resolution scanners if performance improved."

Micro Com replied:

"We too wondered if we could do anything to increase the accuracy of the results we received when preparing the test we ran on your material. We briefly considered scanning your forms on our Opex DS1225, as we believe the images produced on this platform are of a higher quality than the usual Fujitsu scanners we use every day. We ultimately rejected this notion, as we reasoned most FileHold clients would be averse to purchasing $50K scanners.

We decided we would use more mainstream scanners, and focus on using VRS technology to ensure the highest quality images. We performed the scans on a Fujitsu fi-5900, using VRS. We experimented with a significant range of settings, and having found a good balance of image quality, clarity and definition, we spent many hours testing the images in ABBYY FlexiCapture. Once we reached a plateau level and could no longer improve the results of the tests, we asked one of the solutions engineers at ABBYY to have a look at the setups to see if they had any suggestions that might improve the data extraction. Apart from some minor tweaks, he felt Micro Com had done a very thorough job, and left things pretty much as they were.

The same engineer at ABBYY explained that 300 DPI scans created using image enhancing technology like VRS have proven to be the most reliable source for OCR/ICR processing. Micro Com feels that ABBYY software has the best product available in this field, but even they won't claim to be 100% accurate. Their philosophy is to significantly reduce manual data entry, not eliminate it entirely. Unfortunately, even the most sophisticated software in the world still sometimes requires a human to make the final judgment on whether a questionable character is truly an h or a 4."

To learn more about FileHold and ABBYY FlexiCapture scanning results and integration, please contact [email protected].