SetMinimumMatchingTolerance

Syntax

bool SetMinimumMatchingTolerance()

Parameters

Minimum: A decimal value from 0.00 (Lowest Tolerance) to 0.99 (Highest Tolerance). The decimal separator must be appropriate for the locale. Smart parameters are supported.

Returns

Level

Details

This action sets the minimum confidence matching tolerance rating to determine if a fingerprint is a match. If none of the existing fingerprints match with the specified tolerance or greater, then it is considered that none of the existing fingerprints correctly match the fingerprint for the current page.

A non-matching fingerprint is not necessarily an error. For some applications, all of the expected types of pages are known. Maybe there are 10 known page types or perhaps there are 10,000 known page types, but they are all known when the application goes into production so it is expected that the FindFingerprint action should match one of the predetermined fingerprints. If it does not match, then the page may have been ingested as an error, or perhaps needs other handling to be correctly processed.

For other applications, such as APT, they may be deployed with a set of known fingerprints, but can also learn new fingerprints as the application is used. For example, a new vendor starts sending a new kind of document. Fingerprint matching would also fail in that case, however, for APT, it would cause alternate processing which will define the fingerprint at runtime during the course of processing the new document. Once the document has been fully processed, a new fingerprint will be generated, allowing a successful fingerprint match the next time that new page format is ingested.

A lower tolerance rating will allow greater chance for a match of a fingerprint, but also a greater chance for a False match. It is recommended to adjust the tolerance against the set of fingerprints for an application to allow for poor quality documents to successfully match the fingerprint, but not so low that fingerprints match incorrectly.

Once set, this setting will be in effect for the length of the current task profile. If fingerprint actions are used again in a later task profile, then this action must be called again to configure the setting before using subsequent fingerprint actions.

Example

AnalyzeImage()
RotateImage()
RecognizePageOCR_S()
SetSearchArea("0.0","0.5")
SetMinimumMatchingTolerance("0.70")
FindFingerprint("True","")

In this sequence, the fingerprint is created for the current page to create the CCO and the recognition results are also placed into the CCO. The search area is configured to the top 50% of the page and the matching tolerance is set to ".7". The FindFingerprint action then attempts to match the fingerprint for the current page to the existing set of fingerprints. If a fingerprint does not match the minimum tolerance, then the fingerprint for the new page is added to the set of existing fingerprints.