Optical character recognition (OCR) engines having confidence values for text types

• US 20120134589A1
• Filed: 11/27/2010
• Published: 05/31/2012
• Est. Priority Date: 11/27/2010
• Status: Active Grant

• Current Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Original Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Inventors: Reddy, Prakash
• Granted Patent: US 8,452,099 B2
• Time in Patent Office: Days
• US Class Current: 382/182
• CPC Class Codes: G06F 18/254 of classification results, e.g. of results related to same input data
• CPC Class Codes: G06F 18/28 Determining representative reference patterns, e.g. by averaging or distorting; Generating dictionaries
• CPC Class Codes: G06V 30/10 Character recognition
• CPC Class Codes: G06V 30/1914 Determining representative reference patterns, e.g. averaging or distorting patterns; Generating dictionaries, e.g. user dictionaries

Abstract

An image of a known text sample having a text type is generated. The image of the known text sample is input into each OCR engine of a number of OCR engines. Output text corresponding to the image of the known text sample is received from each OCR engine. For each OCR engine, the output text received from the OCR engine is compared with the known text sample, to determine a confidence value of the OCR engine for the text type of the known text sample.

Claims

1. A method comprising:

for each known text sample of a plurality of known text samples, each known text sample having a text type,

generating, by a processor, an image of the known text sample;

for each optical character recognition (OCR) engine of a plurality of OCR engines,

inputting the image of the known text sample, by the processor, into the OCR engine;

receiving output text corresponding to the image of the known text sample, by the processor, from the OCR engine; and

,

comparing the output text received from the OCR engine with the known text sample, by the processor, to determine a confidence value of the OCR engine for the text type of the known text sample.

2. The method of claim 1, further comprising, for each OCR engine of one or more of the OCR engines, periodically redetermining the confidence values of the OCR engine.

3. The method of claim 2, wherein the confidence values of a particular OCR engine of the OCR engines is redetermined responsive to the particular OCR engine being upgraded.

4. The method of claim 2, wherein the confidence values of each OCR engine of the one or more of the OCR engines is redetermined responsive to an additional known text sample being added to the plurality of known text samples.

5. The method of claim 1, wherein the text type of each known text sample comprises one or more of:

a font type;

a font size;

presence of italics;

presence of bold;

presence of underlining; and

, presence of strikethrough.

6. The method of claim 1, wherein comparing the output text received from the OCR engine with the known text sample to determine the confidence value of the OCR engine for the text type of the known text sample comprises:

for each character of a plurality of characters of the output text, comparing the character with a corresponding character of the known text sample to determine whether the character of the output text is identical to the corresponding character of the known text sample;

dividing a number of the characters of the output text that are identical to the corresponding characters of the known text sample, by a total number of the characters of the output text, to yield the confidence value of the OCR engine for the text type of the known text sample.

7. The method of claim 1, wherein comparing the output text received from the OCR engine with the known text sample to determine the confidence value of the OCR engine for the text type of the known text sample comprises:

for each word of a plurality of words of the output text, comparing the word with a corresponding word of the known text sample to determine whether the word of the output text is identical to the corresponding word of the known text sample;

dividing a number of the words of the output text that are identical to the corresponding words of the known text sample, by a total number of the word of the output text, to yield the confidence value of the OCR engine for the text type of the known text sample.

8. The method of claim 1, wherein comparing the output text received from the OCR engine with the known text sample to determine the confidence value of the OCR engine for the text type of the known text sample comprises:

for each word of a plurality of words of the output text,

comparing the word with a corresponding word of the known text sample to determine whether the word of the output text is identical to the corresponding word of the known text sample;

where the word of the output text is identical to the corresponding word of the known text sample, incrementing a score by a number of characters within the word of the output text;

dividing the score by a total number of characters within the words of the output text, to yield the confidence value of the OCR engine for the text type of the known text sample.

9. The method of claim 1, wherein comparing the output text received from the OCR engine with the known text sample to determine the confidence value of the OCR engine for the text type of the known text sample comprises, where the OCR engine consistently makes a same error for a given word of the known text sample, setting the confidence value of the OCR engine for the given word as equal to zero.

10. The method of claim 1, further comprising:

receiving, by the processor, an image of unknown text having a given text type;

inputting the image of the unknown text, by the processor, into each OCR engine;

receiving output text corresponding to the image of the unknown text, by the processor, from each OCR engine;

where the output text received from each OCR engine is not identical, selecting the output text to use as at least provisionally correct for the unknown text, by the processor, based on the confidence values of the OCR engines for the given text type of the unknown text.

11. The method of claim 10, wherein selecting the output text to use comprises selecting the output text received from the OCR engine having a highest confidence value for the given text type of the unknown text.

12. The method of claim 10, wherein selecting the output text to use comprises:

for each OCR engine, setting a weight for the output text received from the OCR engine as equal to the confidence value of the OCR engine for the given text type of the unknown text;

where the output text received from two or more OCR engines of the plurality of OCR engines is identical, summing the weights for the two or more OCR engines as the weight for the output text received from the two or more OCR engines; and

,

selecting the output text having a highest weight.

13. A non-transitory computer-readable data storage medium having a computer program stored thereon for execution by a processor to perform a method comprising:

receiving an image of unknown text having a text type;

inputting the image of the unknown text into each optical character recognition (OCR) engine of a plurality of OCR engines, each OCR having a confidence value for the text type;

receiving output text corresponding to the image of the unknown text from each OCR engine; and

,

where the output text received from each OCR engine is not identical, selecting the output text to use as at least provisionally correct for the unknown text, based on the confidence values of the OCR engines for the text type of the unknown text.

14. The non-transitory computer-readable data storage medium of claim 13, wherein the text type comprises one or more of:

a font type;

a font size;

presence of italics;

presence of bold;

presence of underlining; and

, presence of strikethrough.

15. The non-transitory computer-readable data storage medium of claim 13, wherein selecting the output text to use comprises selecting the output text received from the OCR engine having a highest confidence value for the text type of the unknown text.

16. The non-transitory computer-readable data storage medium of claim 13, wherein selecting the output text to use comprises:

for each OCR engine, setting a weight for the output text received from the OCR engine as equal to the confidence value of the OCR engine for the text type of the unknown text;

where the output text received from two or more OCR engines of the plurality of OCR engines is identical, summing the weights for the two or more OCR engines as the weight for the output text received from the two or more OCR engines; and

,

selecting the output text having a highest weight.

17. A computing system comprising:

a processor;

a computer-readable data storage medium to store an image of unknown text having a text type;

a plurality of optical character recognition (OCR) engines executable by the processor, each OCR engine having a confidence value for the text type, each OCR engine to generate output text corresponding to the image of the unknown text; and

,

logic executable by the processor to, where the output text received from each OCR engine is not identical, select the output text to use as at least provisionally correct for the unknown text, based on the confidence values of the OCR engines for the text type of the unknown text.

18. The computing system of claim 17, wherein the text type comprises one or more of:

a font type;

a font size;

presence of italics;

presence of bold;

presence of underlining; and

, presence of strikethrough.

19. The computing system of claim 17, wherein the logic is first logic,

wherein the computer-readable data storage medium is to further store a plurality of known text samples and an image of each known text sample, each known text sample having a given text type,

wherein each OCR engine is to generate output text corresponding to the image of each known text sample,

wherein the computing system further comprises second logic to, for each OCR engine, compare the output text generated by each OCR engine corresponding to the image of each known text sample with the known text sample, to determine a confidence value of the OCR engine for the given text type of the known text sample,

and wherein the text type of the unknown text is one or more of the given text types of the known text samples.