Method and system for processing candidate strings generated by an optical character recognition process
US10438083B1 · kind B1 · utility
Assignee
Inventor
Key dates
| Filing date | Dec 20, 2016 |
| Grant date | Oct 8, 2019 |
| Priority date | — |
| Expiry date | Dec 20, 2036 |
Classification
- Technology area (CPC G)Physics
- CPC primaryG06V20/625
- WIPO fieldComputer technology
- WIPO sectorElectrical engineering
Abstract
A method and system of recognizing a string of characters in a target image. An acquired target image is analyzed using an optical character recognition process to identify a candidate string, the candidate string having an associated plurality of character positions, each character position being associated with a set of one or more candidate characters located at related positions in the target image. A minimum edit cost is determined between the candidate string and a template for an output string. Determining the minimum edit cost includes, for a given character position in the candidate string and a given output character position in the template, identifying, among the respective set of candidate characters of the candidate string, a subset of candidate characters that satisfy the respective character set of the template; and determining an edit cost based on the recognition score of one of the candidate characters belonging to the identified subset of candidate characters. An output string corresponding to the determined minimum edit cost is returned.
Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.