How to Improve Recognition Results

A variety of factors may affect the quality of recognition results. If you are working with complex multi-page documents, it is important to take into account the characteristics of original images and specify the correct settings. This will significantly improve the quality of recognition results.

This section describes how you can take into account the characteristics of your original document, select the right settings, resolve non-standard situations, and get high-quality recognition results.

Recognition errors may occur when recognizing facing book pages or when the program fails to detect the orientation of a page. Working with such documents is described in "Splitting Facing Pages and Correcting Text Orientation".
The quality of recognition results is influenced by the document layout. Better results can be achieved by checking and adjusting the detected areas manually for documents with complex structures and very many elements, such as tables or barcodes. For more information, see "Working with Areas".
If a document is written in more than one language, including a complex script language, such as Chinese or Japanese, be sure to specify the actual languages used in the document. For more information, see "Tips for Working with CJK Texts".
The quality of original images greatly affects recognition times and quality. You can process document images using the tools available in the Image Editor. For more information, see "How to Edit Images Manually (Image Editor)".
FineReader Sprint monitors image loading, structure analysis, recognition, and export. If a problem occurs that may affect the quality of recognition, the application will display a warning or an error message. For more information, see "Errors and Warnings".