Text information processing method, system and device and computer readable storage medium

Abstract

The invention discloses a text information processing method, system and device and a computer readable storage medium. According to the method, the error correction model obtained by training the error correction training set corresponding to the type of the to-be-processed text is adopted in advance, error correction processing is conducted on the to-be-processed text, at least one correction text of the to-be-processed text is obtained, and correction of glyph errors and the like in the to-be-processed text is achieved; a named entity recognition model is obtained by training a structured feature training set corresponding to the type of a to-be-processed text in advance; extracting structural features of the corrected text; matching the structural features of the corrected text with the structural features of each part of standard text information in a trusted data set; and determining standard text information corresponding to the corrected text, thereby further correcting named entity errors existing in the corrected text through the structured features, and improving the accuracy of text information recognition.

Description

Technical field [0001] The present invention relates to the field of image processing technology, in particular to a text information processing method, system, equipment and computer-readable storage medium. Background technique [0002] In daily work or life, paper documents such as various bills and certificates are used, such as invoices, business licenses, etc. In order to realize the identification of paper documents, computer technology is used to automatically identify text information printed on paper Become a trend. Especially for key text information such as company names, it has specific structural features and requires high recognition accuracy. In many financial situations, company names and similar text information are not allowed to have any errors. [0003] At present, the recognition of text information printed on paper mainly uses optical character recognition (Optical Character Recognition, hereinafter referred to as OCR) technology, which uses optical technolo...

AI Technical Summary

Problems solved by technology

[0004] However, in the process of text recognition through OCR technology, the following recognition errors often occur: for some long text information on paper documents, the head and / or tail of the text are cut due to text positioning deviation; Structural fonts, it is easy to mistakenly identify Chinese characters with left-right structure or left-right structure in the text information as two or more characters, such as recognizing "ka" as "mouth card", etc.; because paper documents are not clear, printing Difficulties in recognition caused by tilting, partially covered text, overlapping text coverage, low brightness, etc., resulting in the recognition of some complex characters as other characters, etc., and errors in the recognition of mixed Chinese and English characters, such as recognizing "IBM" as "18M "Wait

The existing recognition of text information on paper documents has a high error rate. How to correct the recognition results and improve the accuracy of text information recognition has become an urgent technical problem to be solved.

Images