Longest common substring automatic error correction method and system based on OCR recognition result

Abstract

The invention provides a longest common substring automatic error correction method and system based on an OCR recognition result, and relates to the field of image optical character recognition, andthe method comprises the following steps: step 1, obtaining a character string in a to-be-detected image file through OCR software recognition; step 2, preprocessing the character string obtained by the OCR software identification; step 3, performing character error correction replacement processing on the preprocessed character string; and step 4, performing longest common substring matching calculation processing based on the character string subjected to character error correction replacement processing, and outputting a correct result. According to the method, automatic error correction and replacement are carried out on the character string recognized and input by the OCR software, then longest common substring matching calculation is carried out on the character string subjected to error correction and replacement and the target character string needing to be output, and the correct target character string is output. The problem that when the image file with low definition is recognized, the image file recognition output success rate is low is solved.

Description

technical field [0001] The invention relates to the field of image optical character recognition, in particular to an automatic error correction method and system for the longest common substring based on OCR recognition results. Background technique [0002] OCR software refers to software that uses OCR (Optical Character Recognition) technology to recognize, extract and convert text content on pictures, photos, electronic faxes and other images into editable text, through scanners, cameras, electronic fax machines, etc. The device acquires and saves image files, and then reads and analyzes image files through OCR software and extracts character strings through character recognition. [0003] When digitally managing various documents such as tax invoices, contracts, fund transaction notes, and transfer instructions, OCR software is required to automatically identify the image content of the fixed area of ​​​​the document and extract information such as the account and amoun...

AI Technical Summary

Problems solved by technology

This method uses multiple OCR software to identify the wrongly identified data by voting. It does not pay attention to the own rules of each field in the identified file content and the business rules between fields, and does not use these rules and rules to correct the wrong data. Implement inspection and correction, so when identifying electronic fax image files with very low resolution, the success rate of image file recognition output is very unsatisfactory, and it is difficult to be practically applied in a business production environment

Images