Module name:	HNR
Module identifier:	RM_HNR
Filling methods supported:	FM_HANDPRINT
Filters supported:	FILTER_ALL, FILTER_DIGIT, FILTER_PUNCTUATION and FILTER_MISCELLANEOUS
Trade-off supported:	TO_FAST, TO_ACCURATE (includes TO_BALANCED)
Knowledge base file:	`KEZIRSTR.TBS`
Training file supported:	no

This recognition module is supported on: Windows.

This module is included only in the Professional Recognition Kit (not the OCR kit). To make this technology available in your application, it must be covered by your distribution licensing. See the topic on Licensing in the General Information help system.

Application areas

This recognition module can be used for recognition of handprinted numerals and four additional signs. If further handprinted characters are to be recognized, the use of the RM_RER recognition module is recommended, which is supported on: Windows, Linux, Mac OS X.

Range of characters

This recognition module can recognize the following handprinted characters:

the digits (0-9),
the Plus Sign (+)
the Minus Sign (–)
the Period or Full-stop (.)
the Comma (,).

The filter FILTER_DIGIT can be used to exclude the last four characters. The filters FILTER_PUNCTUATION and FILTER_MISCELLANEOUS function, other filters have no effect.

Accuracy issues

The functioning of the module can be influenced with the following parameters:

Recognition trade-off settings: The page-level trade-off value TO_FAST is respected, while TO_ACCURATE and TO_BALANCED are merged to a single value.
Writing style:
- STYLE_EURO: lets you specify European styled numbers,
- STYLE_US: lets you specify North-American styled numbers.
Character box flag: This lets you pass information to the recognition module on the positions of characters to be recognized, which is likely to improve accuracy. If a zero value is given, the recognition module will determine the positions automatically.
Space inclusion flag: This lets you set whether spaces detected between the recognized characters should be transmitted or not.

Conditions

For successful recognition, the characters should not touch each other. Each character can be zoned individually or a zone may contain several lines of characters. Each character must have a height of 30-180 pixels. Well-formed numbers written in pen are best recognized. Pencil and felt-tip pens give poorer results. When reading from pre-printed forms, dropout colored boxes can be useful to encourage respondents to write numbers of even size and spacing. But then, they mustn’t use a pen with the dropout color!

Optimum box size for such cases is 5 x 6 mm separated by 3 mm.

Maximum number of characters in a zone: 3000. Maximum number of lines in a zone: 40. Maximum number of characters in a line: 600.

Output

This can consist of the digits, the four additional characters, rejection symbols and, if requested, spaces. The RM_HNR module can benefit from the checking module to improve recognition accuracy, by encouraging output to conform to the systems’ pre-defined requirements. But note that spell checking from a language dictionary should be disabled. For zones whose contents are known to be limited in some way, a set of permitted numerical strings pre-defined in a User dictionary could be very useful.

Note:: See HNR Recognition Engine Module.