Is it possible to scan an identity zone once and then compare with multiple strings to identify the specific form template?

[Jonathan_Willco]

I have configured approximately 40 templates, but they all look at the same header area on the page to identify the form template. It appears that Advanced Capture is scanning the same header area on the page each time, potentially 40 times. This seems like a big performance hit, particularly for larger documents (50-100 pages).

Is it possible to scan the same header area once, OCR the text from that area once, and then attempt to match that text with multiple compare strings to identify the form template?

[Steve_Reed · Accepted solution]

I would recommend using the 'form definition lookup text' methodology to match the different header values to their associated templates - this way you only need to have the engine OCR the page once, get the text, then do a direct lookup again the database to find the template.

[Jonathan_Willco]

That sounds promising, but then how do we add the registration points?

[Steve_Reed]

The lookup text is used to filter the candidate template list down (hopefully to only one candidate, but it can be several) - then the form identity zones are run for those template(s) - if they have registration points assigned to them, they will still be used.

[Jonathan_Willco]

Thanks, this is working pretty well. The one-page forms process in a second or two, but the larger forms take over 20 seconds each. By tuning the minimum page count, I could reduce the time from ~25 seconds to ~6 seconds. By using the lookup text, I could reduce it from ~6 to ~5 seconds.