EasyOCR is
a font-dependent printed character reader based
on a template matching algorithm. It allows training of the font
to be recognized by showing sample images of all possible characters.
For this reason, it is able to read any kind of short
text (serial numbers, labels...) such as those found
in industrial environments. 
The training phase uses an interactive utility to show samples of the characters
and store them in a font file. Such a font editor is available, ready-made, in
EasyAccess. Anyway, EasyOCR provides
the means for you to write your own, custom, font editor.
EasyOCR provides
three standard fonts: OCR-A, OCR-B and Semi.
This avoids a training phase to recognize a text using one of
these fonts. |
|
Beside
the provided fonts EasyOCR can
be tea ched
for user defined fonts. In EasyAccess a
font editor is integrated where simple training can be done by
presenting an image containing the characters. Features will
be extracted and stored in a configurable data base. Parameters
for character separation, maximum and minimum width / height
and in certain restriction for character reconstruction can be
set. |
Information request eVision: |