Recognition quality greatly depends on the quality of the source image. When you want to recognize paper documents the first step is scanning (see Using a scanner from Getting an image page). Before scanning you can check the print quality of your paper document. If the document contains: noise (i.e. random black dots or speckles), blurred and uneven letters or skewed lines then this document can be determined as poor-quality and may require specific scanning settings.

For better recognition results on poor-quality paper documents scanning in grayscale at 300 DPI is recommended.

Working with image files

When you work with image files for best results you can manually edit selected regions, specify new regions, delete regions and/or change region types.

The recognition process is applied only on all of the regions for given image. You have a full control on regions. So manipulating regions you can control the recognition process too. For example if a picture region is marked as text during automatic analysis, you can convert the region type to picture region or if some text is not detected as a part of text region you can use the Draw Text Region tool from Image View panel to draw a text region around the text. You are free to resize or delete existing regions and/or define new ones.

If the page orientation is not standard (for example the text is upside-down) the rotation buttons on Image View panel can be used.

Working with digital photographic images

In some cases you have a paper document but don't have a scanner. In such situations you can take a photo of your document and use this photographic image for recognition.

For best results in these situations:

Make sure that the paper fits entirely within the frame that there are no dark areas or shadows on it.
Position the camera parallel to the plane of the document so that the lens looks to the center of the page being photographed.

The minimum required digital camera specifications are:

2-megapixel sensor (5-megapixel or above is recommended).
Variable focus lens (Usually cell phones and hand-held devices use fixed-focus cameras. The images produced with these cameras are not suitable for recognition).