PatternMatch_Identify

Identifies a page by using image pattern matching.

Member of namespace

PatternMatch

Syntax

bool PatternMatch_Identify ()

Parameters

None.

Returns

False, if the rule that contains this action was not applied to a Page object of the Document Hierarchy; if a pattern match is not found; or if fingerprints do not exist. Otherwise, True.

Level

Page level only.

Details

PatternMatch_Identify identifies a page's type and fingerprint by using geometric pattern matching. The locations of unique patterns are configured as Anchor fields for each fingerprint in the Document Hierarchy. One or more Anchor fields can be used to match geometric shapes on a fingerprint to the current image. If one or more Anchor fields on the current page match a fingerprint with equal to or greater than the configured confidence level, the page is identified with that fingerprint. The action does not require all defined anchors to match - the first match is used. The action loads all Anchor field patterns from the fingerprint library, then searches on the current image for each of the patterns in the associated zones. The search area for each zone is increased by the dimensions that are specified in the page METRIC variable. If METRIC is not specified, the default is 500 pixels horizontal and vertical. When this action finds a match, it sets the matching Fingerprint ID and Page Type. It also creates page-level fields and update the Anchor fields with Anchor-specific pattern offset values in a field-level Image_Offset variable. The offset can be used subsequent to matching a fingerprint. The pat_RegisterZones action can be used to align the zones in the fingerprint to the current image, providing more accurately positioned text in each field.

This action requires the current page image to be bi-tonal (black and white). Grayscale or color images cause the action to fail. The fingerprint image must have the same resolution (DPI) as the current page image. The geometric shape that is contained in each Anchor field must be bold and well-defined with clear edges, with crisp, black and white markings that produce a distinct shape. The shape must be thick and compact, not composed of long thin lines. To avoid false positive matches, the shape must not match other shapes or black areas that might exist nearby within the same image.

Example

PatternMatch_Identify()