Systems and methods for merging word fragments in optical character recognition-extracted data

Patent metadata

Field	Value
Publication number	US-10679087-B2
Application number	US-201815956547-A
Country	US
Kind code	B2
Filing date	Apr 18, 2018
Priority date	Apr 18, 2018
Publication date	Jun 9, 2020
Grant date	Jun 9, 2020

Abstract

Systems and methods for merging adjacent word fragments in outputs of optical character recognition (OCR) systems can include a processor obtaining word fragments associated with OCR data generated from an image. Each word fragment can be associated with a respective text line of a plurality of text lines. The at least one processor can determine, for each pair of adjacent word fragments in a text line, a respective normalized horizontal distance between the pair of adjacent word fragments. The processor can identify one or more pairs of adjacent word fragments that are candidates for merging based on the determined normalized horizontal distances. The processor can determine that a pair of adjacent word fragments, among the one or more pairs of adjacent word fragments that are candidates for merging, matches a predefined expression of a plurality of predefined expressions, and merge that pair of adjacent word fragments into a single word.

First claim

What is claimed is:

1. A computer system for merging adjacent word fragments in outputs of optical character recognition (OCR) systems, the computer system comprising:
at least one processor; and
a memory storing computer code instructions, the computer code instructions when executed by the at least one processor, cause the at least one processor to:
obtain a plurality of word fragments associated with OCR data generated from an image, each word fragment of the plurality of word fragments associated with a respective text line of a plurality of text lines;
determine, for each pair of adjacent word fragments in a text line of the plurality of text lines, a respective normalized horizontal distance between the pair of adjacent word fragments;
identify, in the text line of the plurality of text lines, one or more pairs of adjacent word fragments that are candidates for merging based on the normalized horizontal distances;
determine that a pair of adjacent word fragments, among the one or more pairs of adjacent word fragments that are candidates for merging, matches a predefined expression of a plurality of predefined expressions; and
merge the pair of adjacent word fragments that matches the predefined expression into a single word, responsive to determining that the pair of adjacent word fragments matches the predefined expression.

2. The computer system of claim 1 , wherein the image includes an image of a receipt.

3. The computer system of claim 2 , wherein the plurality of predefined expressions includes an expression of prices associated with the receipt.

4. The computer system of claim 2 , wherein the plurality of predefined expressions includes an expression of item codes or identifiers (IDs) associated with the receipt.

5. The computer system of claim 1 , wherein the computer code instructions, when executed by the at least one processor, cause the at least one processor to:
determine a length of a gap between the pair of adjacent word fragments; and
divide the length of the gap between the pair of adjacent word fragments by a dimension of the image.

6. The computer system of claim 5 , wherein the dimension of the receipt includes a width of the receipt.

7. The computer system of claim 5 , wherein the dimension of the receipt includes a width of a text segment of the receipt.

8. The computer system of claim 1 , wherein the computer code instructions, when executed by the at least one processor, cause the at least one processor to:
compare, for each pair of adjacent word fragments in the text line of the plurality of text lines, the respective normalized horizontal distance between the pair of adjacent word fragments to a threshold value; and
assign the pair of adjacent word fragments as a candidate for merging upon determining that the respective normalized horizontal distance between the pair of adjacent word fragments is smaller than or equal to the threshold value.

9. The computer system of claim 1 , wherein the computer code instructions, when executed by the at least one processor, cause the at least one processor to:
match three or more consecutive word fragments, among the one or more pairs of adjacent word fragments that are candidates for merging, to one other predefined expression among the plurality of predefined expressions; and
merge the three or more consecutive word fragments into a single word, responsive to matching the three or more consecutive word fragments to the one other predefined expression.

10. The computer system of claim 1 , wherein the plurality of word fragments are arranged into the plurality of text lines.

11. A method of merging adjacent word fragments in outputs of optical character recognition (OCR) systems, the method comprising:
obtaining a plurality of word fragments associated with OCR data generated from an image, each word fragment of the plurality of word fragments associated with a respective text line of a plurality of text lines;
determining, for each pair of adjacent word fragments in a text line of the plurality of text lines, a respective normalized horizontal distance between the pair of adjacent word fragments;
identifying, in the text line of the plurality of text lines, one or more pairs of adjacent word fragments that are candidates for merging based on the normalized horizontal distances;
determining that a pair of adjacent word fragments, among the one or more pairs of adjacent word fragments that are candidates for merging, matches a predefined expression of a plurality of predefined expressions; and
merging the pair of adjacent word fragments that matches the predefined expression into a single word, responsive to determining that the pair of adjacent word fragments matches the predefined expression.

12. The method of claim 11 , wherein the image includes an image of a receipt.

13. The method of claim 12 , wherein the plurality of predefined expressions includes at least one of:
an expression of prices associated with the receipt; or
an expression of item codes or identifiers (IDs) associated with the receipt.

14. The method of claim 11 , further comprising:
determining a length of a gap between the pair of adjacent word fragments; and
dividing the length of the gap between the pair of adjacent word fragments by a dimension of the image.

15. The method of claim 14 , wherein the dimension of the receipt includes a width of the receipt.

16. The method of claim 14 , wherein the dimension of the receipt includes a width of a text segment of the receipt.

17. The method of claim 11 , further comprising:
comparing, for each pair of adjacent word fragments in the text line of the plurality of text lines, the respective normalized horizontal distance between the pair of adjacent word fragments to a threshold value; and
assigning the pair of adjacent word fragments as a candidate for merging upon determining that the respective normalized horizontal distance between the pair of adjacent word fragments is smaller than or equal to the threshold value.

18. The method of claim 11 , further comprising:
matching three or more consecutive word fragments, among the one or more pairs of adjacent word fragments that are candidates for merging, to one other predefined expression among the plurality of predefined expressions; and
merging the three or more consecutive word fragments into a single word, responsive to matching the three or more consecutive word fragments to the one other predefined expression.

19. The method of claim 11 , wherein the plurality of word fragments are arranged into the plurality of text lines.

20. A computer-readable storage device storing instructions that, when executed by one or more processors, cause the one or more processors to perform several operations for assigning word fragments to lines of text in optical character recognition (OCR) generated data, the operations comprise:
obtaining a plurality of word fragments associated with OCR data generated from an image, each word fragment of the plurality of word fragments associated with a respective text line of a plurality of text lines;
determining, for each pair of adjacent word fragments in a text line of the plurality of text lines, a respective normalized horizontal distance between the pair of adjacent word fragments;
identifying, in the text line of the plurality of text lines, one or more pairs of adjacent word fragments that are candidates for merging based on the normalized horizontal distances;
determining that a pair of adjacent word fragments

Assignees

• Google Llc

Inventors

• Ordonez Ivan
• Krishnamurthy Swaminathan
• Paul David
• Udeshi Tushar

Classifications

• G06K9/348

  Physics · mapped topic

• G06K9/344

  Physics · mapped topic

• G06K9/342 Primary

  Physics · mapped topic

• G06K2209/01

  Physics · mapped topic

• G06V30/412

  Layout analysis of documents structured with printed lines or input boxes, e.g. business forms or tables · CPC title