Preprocessing

Preprocessing improves the quality of scanned documents automatically before they are processed further. That includes removing speckles, straightening pages scanned askew, removing blank pages and converting to black and white, for example. It improves the results of text recognition (OCR) and saves storage space.

Permission required

To access this function you need the following permission:

• Edit workflow (AdminUI_WorkflowManagement_EditWorkflow)

You grant permissions through groups — see permissions – overview.

Preprocessing offers various image enhancement methods for processing image documents. It runs as the first working step on incoming documents — that is, before any recognition of text or barcodes and before any output files are saved. Where no preprocessing is wanted, choose Input/output as the workflow type. Preprocessing is page-oriented: each page is processed independently and has no effect on any other page.

Removing speckles

Removes speckles and show-through background on the pages of a scanned document.

The options available are:

• No removal

• Remove

• Remove more

Rotating and mirroring

Rotates and mirrors the pages of a scanned document. The list offers various possibilities. This function is particularly useful with some older multifunction printers that store no information about page orientation and file every page rotated by 90° or mirrored.

The Rotate automatically function attempts to work out the orientation of the page by text recognition. Note that recognising text can be relatively time-consuming and switching this on may reduce the processing speed.

Align automatically attempts to straighten documents scanned slightly askew. The maximum angle corrected is 7°.

Scaling

Offers the option of enlarging or reducing the pages of a scanned document. The value 100 means the size is left unchanged; every other figure stands for the scaling as a percentage.

Changing the resolution

Changes the resolution of pages of a document newly arrived. This function is useful where documents were delivered at too high a resolution, one that is not needed for the processing that follows. If documents were scanned at 1200 dpi, for instance, and only 300 dpi is needed for a PDF with text recognition, changing the resolution brings the following advantages:

• The documents are processed faster
• The file sizes are smaller when the output is saved

Detecting blank pages

The Detect blank pages function analyses every page of a document individually in order to establish whether it contains information or is entirely empty. The detection works essentially by counting the pixels that are not white. Where a given proportion of white, as a percentage, is exceeded, the page counts as blank and is removed. Configuring this option can be demanding, because various parameters that influence the white pixels have to be taken into account.

For example:

• Pale writing, from a pencil for instance
• A speckled background
• A coloured background
• A background showing through
• Scan margins
• Punch holes

Pale writing can be strengthened with the contrast and/or the black-tone dilation. Background disturbances can usually be removed with the other preprocessing options. Where scan margins and punch holes are to be taken into account explicitly in the blank page detection (that is, excluded from the search for white pixels), set the corresponding Take scan margins into account and Take punch holes into account options.

Caution: Configuring this option wrongly or faultily can mean that pages are detected as blank and excluded from the processing that follows.

Removing lines

Where horizontal and/or vertical streaks appear while documents are being scanned, this option removes them automatically.

Removing punch holes

With this function switched on, Docuflair looks for punch holes at the edge of every page and removes them automatically where it finds them.

Black-tone dilation

Documents are often scanned at a high brightness or turned into binary images by a poor algorithm. That results in weakly sampled, unclear characters. The black-tone dilation can improve how the characters appear.

The following modes are supported:

• None (switched off)
• Bitonal dilation 4
• Bitonal dilation 8
• Vertical bitonal dilation
• Horizontal bitonal dilation
• Standard bitonal dilation

Eroding type

Sometimes the characters in documents appear thick and their features indistinct — where too much ink was used in printing, for example. That is usually the result of the document being scanned and printed repeatedly, particularly at low settings such as a low DPI. This function can help remedy problems of that kind.

The following modes are supported:

• None (switched off)
• Erode 8
• Erode 4
• Vertical erode
• Standard erode

Turning the image into a negative

Older scanners in particular sometimes generate TIFF image files that show black pixels as white and the other way round. This function produces a negative by inverting every pixel.

Colour conversion

Changes or reduces the colour model and turns coloured input documents into black-and-white images, for example. Where no colour is needed in the documents output and a PDF with text recognition is to be produced, say, the colour conversion brings the following advantages:

• The documents are processed faster
• The file sizes are smaller when the output is saved

Edge enhancement

Switching this function on improves the transitions at edges.

Removing black margins

Where a scanned original is bounded by dark margins on one or more sides, the document was either not placed exactly on the scanning surface, placed askew, or the area scanned was larger than the page to be scanned. This function removes dark margins around the outside of a page automatically.

Stretching the contrast

This setting optimises the contrast automatically for every page of the scanned document.

Contrast, brightness, saturation and gamma

Brightness, saturation and contrast are typical image processing functions. For more on them, see the Wikipedia article on image editing. Information about gamma correction can be found on Wikipedia, among other places.