The automatic page-layout decomposition process (auto-zoning) can be activated directly by calling the ImGearRecZoneCollection.Locate Method for finding text blocks on the image. It creates an entire zone list for the given image.
The Locate Method has an overload that takes an ImGearRectangle as a parameter. This rectangle serves to limit the area of the image inside which zones will be searched and defined.
The recognition engine offers three different algorithms to be applied during auto-zoning: use the ImGearRecRecognitionSettings.DecompMethod Property to specify which Page parser algorithm setting should be applied during auto-zoning from among the following:
- STANDARD : The page parser algorithm introduced in the previous version of the toolkit
- LEGACY : The one incorporated in the earlier versions of the toolkit
- FAST : The quickest, but not able to handle complex cases (including tables)
In addition, the integrating application can set the zoning method to AUTO. With this setting, the recognition engine determines the method to use. See ImGearRecDecompositionMethod Enumeration for details.
Applying the new STANDARD algorithm results in the best throughput in overall accuracy, though in some cases it might require significantly more time to complete.
When auto-zoning is used, each resulting zone is initialized with the following:
- A unique auto-assigned zone identifier
- The zone's detected coordinates
- The zone type determined (FLOW, TABLE, or GRAPHIC)
All zones created by this function will have the following:
- The default filling method, DEFAULT
- The automatic recognition module selection AUTO setting
- The DEFAULT Character Set filter to specify the use of global filtering, if any
- None of the checking prohibit flags (e.g., LANGDICT_PROHIBIT) set
Any zone can be locally overridden with the Zones Property of ImGearRecPage, which allows you to access the zone list and change the attributes of a zone.
