Confidence values are calculated on several factors depending upon recognition characteristics. They should only be considered as relative and only to other results of the same type.

The Confidence property of the result object specifies the confidence value the recognition engine uses for the analysis of the OCR character, text line, or text block. The value range is 0 - 100, where 100 is the greatest confidence.

How Confidence Value is Determined

When an image is submitted to SmartZone OCR, the SmartZone OCR engine matches the character to character sets and field type, and regular expressions or data validation lists you've defined to determine the best recognition results.

		Y	80%
		U	71%
Y	=	V	65%
		v	40%
		u	27%

Increase Confidence Values through Methods and Properties

Set and modify the properties below as needed prior to calling the Reader for zonal recognition of characters. The properties to set for accuracy and confidence are:

Property	Description
Area	The rectangular coordinates of the image to perform recognition on. The area uses the Rectangle object which contains the height, width, x and y coordinates to define the area on an image.
CharacterSet	The set of valid characters to match for recognition. Modify the character set as needed for the closest possible match.
FieldType	Used to define the expected format of text being recognized.
RegularExpression	Used to define a custom format of the text being recognized.
DataValidationList	Used to define the expected contents of data being recognized.
MinimumCharacterConfidence	The confidence value used to determine if the character is rejected in the text output. Adjust this value as needed based on the results for better accuracy.
RejectionCharacter	The character used as a replacement value when the recognition confidence is below the MinimumCharacterConfidence value. Confidence will always be 0 and only be 0 when the rejection character is used, since the recognition engine did not produce the value.
Segmentation	Gets and sets Segmentation parameters including detecting spaces, blobs, lines, and merged characters.

The SmartZone OCR Result object is the base class for all result objects including TextBlockResult, CharacterResult, and TextLineResult. The properties to set in these classes are:

Property	Description
Area	The rectangular coordinates of the image of the Result object.
Confidence	The resulting confidence value of the Result object.
Text	The text string of one or more characters of the Result object.

 

See Also