ABBYY FineReader Engine 12 for Windows - Release 4

Release date: 28 October 2019 (public release)
Part #: 1342.44
Build #: 12.4.7.0

New: 'Compare Documents' Module

The new 'Compare Documents' Module in ABBYY FineReader Engine provides your software with the ability to quickly and accurately compare two versions of the same document and detect possible differences in their content – for example to compare the originally created contract and its printed and signed version and automatically verify the document's integrity.

The technology can analyze documents in different formats, i.e. Microsoft Word, PDF, JPEG, TIF and other document or image formats and can provide significant value to any document processing solution, especially to applications for the legal, government and business sector.

For easy implementation and demonstration, a new ready-to-use code sample with sample documents is available and can be used in own applications to speed up development work. The sample compares the selected files and, if necessary, saves the detected differences to a file of the specified format.

Higher accuracy for Arabic OCR: New AI-based algorithms

To increase recognition accuracy on low-quality images, new smart AI-based algorithms were implemented for recognition of Arabic. In addition to the ‘standard’ OCR processing approach, a newly trained Recurrent Neural Network for Arabic enables End-to-End recognition processes delivering highly accurate recognition results – even on images with very low quality. To optimize recognition quality and processing speed, a new intelligent built-in classifier selects the appropriate processing methodology.

Higher accuracy for Korean OCR: New AI-based algorithms

To further increase the recognition accuracy of the Korean language, a new Deep Learning Language model was trained on a large document amount. This intelligent model selects the best recognition variant from different recognition hypotheses based on his ‘knowledge’ acquired during an extensive training on documents in Korean. To optimize the balance between the recognition accuracy and OCR processing speed, a new smart built-in classifier selects the usage of this new model only under specific conditions.

Enhanced: Text-based classifier with advanced security of training data

To further increase the security and protection of information contained in training documents for the text-based classification , hashing algorithms were implemented. Using information from checksums eliminates the possibility to recover information from the sample documents.

Enhanced: Classification Demo Sample - now with Office format documents

To reflect the FineReader Engine’s capability of processing Office documents in the classification process, the provided pre-compiled Demo Sample for classification was enhanced and allows now to display Office documents in the classification results (in addition to PDFs and image formats). Furthermore, the newly added sample documents can be used to test and demonstrate the classification capabilities.

Improved: Document layout preservation

To improve the detection and recreation of document layout, a new 'single-column' document model was introduced that provides more exact detection and analysis of tables and charts and significantly improves the detection and recreation of document layout.

Enhanced: Java wrapper documentation

To simplify the usage of the API, the documentation of the ABBYY FineReader Engine 12 has been extended and the documentation for the Java wrapper is now provided in JavaDoc format in addition to the HTML and PDF formats.

New code sample for document comparison

The extensive code samples library was extended by a new sample that allows testing and demonstrating the ability to compare two versions of the same document and detect differences in their content.

Support of the latest .NET Framework versions

The distributive now includes .NET COM Interop wrappers for the .NET Framework versions 3.5, 4.0, 4.5, 4.6, 4.7, 4.8

New operating system support: Microsoft Windows Server 2019

FineReader Engine 12 Release 4 was tested on Windows Server 2019. This is the first release officially tested on this platform.

Improved: Developer's Help documentation

The Developer’s Help of the FineReader Engine 12 has been extended by additional information about different possibilities of licensing the SDK, describing the individual types of licensing options in an easy-to-understand comparison table.

IMPORTANT INFORMATION

GetEngine function was deprecated in R2 ⇒ To load the Engine object, please use the InitializeEngine function. It provides the unified Engine loading procedure for all license types (including the Online License).
Customers updating from previous versions of FineReader Engine such as version 9, 10, or 11 as well as upgrading from FineReader Engine 12 Release 1 to Release 2 and higher who use the GetEngine function would receive an error message if they keep using it in later releases. Please update your code and replace the GetEngine with the InitializeEngine function.
License name change: To better reflect its functionality, the license type ‘Cloud License’ was renamed into ‘Online License’ in the R3 for Windows. This license supports deployment in virtual & cloud environments, usage with Docker containers as well as on premise installations.