abbyy-software formreader ユーザーガイド
text image quality can result in the appearance of field borders or the background on the form image, and
consequently, cause a deterioration in the recognition quality.
consequently, cause a deterioration in the recognition quality.
2. If the printer makes unauthorized changes to the technical print parameters (i.e. different paper, other color
components) then the background may become too dark and could prove difficult to remove regardless of
the scanning parameters chosen.
the scanning parameters chosen.
Black&white forms with raster background
Fields on such forms are simply white spaces (usually rectangles) on a raster background. The background is made
up of individual dots, no more than 0.1 mm in size, with the distance between each dot about 1 mm. This is much
greater than is the case with gray forms, where dot density is such that the eye perceives the background as smooth
gray.
Fields on such forms are simply white spaces (usually rectangles) on a raster background. The background is made
up of individual dots, no more than 0.1 mm in size, with the distance between each dot about 1 mm. This is much
greater than is the case with gray forms, where dot density is such that the eye perceives the background as smooth
gray.
Background filtering
The raster background does not disappear during scanning itself; instead, the raster dots are classified as garbage and
removed from the image during despeckling.
The raster background does not disappear during scanning itself; instead, the raster dots are classified as garbage and
removed from the image during despeckling.
Advantages and disadvantages
Advantages
1. If both the scanning parameters and the dot size are chosen correctly, the form image will be despeckled
and the recognition module will acquire the field image free of garbage and superfluous characters.
2. Letters/digits overlapping field borders is less of a problem; field borders are part of the background, and
therefore disappear during image cleaning, leaving only the field contents left to be recognized..
Disadvantages
1. Raster forms require periods, commas and other small characters to be written thickly. This is because their
size must be greater than that of the raster dots; otherwise they will be removed as part of the background.
2. Scanning parameters (brightness and contrast) can only be altered to a limited extent. This can prove
problematic when scanning forms completed using a very light ink, as decreasing the brightness to increase
the text image quality can result in the field borders or the background appearing on the form image, and
consequently, worsen the recognition quality.
the text image quality can result in the field borders or the background appearing on the form image, and
consequently, worsen the recognition quality.
3. Not all graphic editors and word processors (e.g. MS Word) have the shading style described above (i.e.
raster) in their standard styles palette. , In addition, word processors normally only have a limited number
of raster set up tools, leading to difficulties, for example, when trying to change the distance between raster
dots, or their size.
of raster set up tools, leading to difficulties, for example, when trying to change the distance between raster
dots, or their size.
4. A raster background can prove tiring to the eye, and consequently discourage form completion.
5. If printing density is increased, dots may become larger and, as a result, left on the image as garbage. This,
5. If printing density is increased, dots may become larger and, as a result, left on the image as garbage. This,
in turn can make character recognition impossible.
Black&white forms with raster borders
Field borders here are made up of raster lines i.e. sequences of small black dots. Raster dot size should be 0.39 –
0.5 pt.
Field borders here are made up of raster lines i.e. sequences of small black dots. Raster dot size should be 0.39 –
0.5 pt.
The recommended raster dot size is 0.39 pt, with the distance between the raster dots being at least five times larger
than the dot size:
than the dot size:
If the distance is less, the dots may become glued during scanning, leading to them remaining on the image after
despeckling. This, in turn, leads to lower recognition quality.
Acceptable ways of completing fields with raster borders are shown in the figures below:
despeckling. This, in turn, leads to lower recognition quality.
Acceptable ways of completing fields with raster borders are shown in the figures below: