A method for recognizing similar characters based on ocr fonts

Abstract

The present invention relates to the field of computer technology, in particular to the field of pattern recognition and deep learning, and more specifically to a method for recognizing similar characters based on OCR fonts. Changing the traditional glyph recognition method, it can recognize both character text and fonts. Through multi-sample comparison and adding threshold screening, it not only greatly improves the accuracy of text recognition, but also effectively recognizes character fonts. It is especially suitable for character recognition of similar glyphs and similar fonts, and realizes double accurate recognition of glyphs and fonts. Each character is cut into a size of 96*96 pixels by horizontal and vertical segmentation, which facilitates the extraction of pixel feature information, avoids mutual interference between adjacent characters, and effectively improves recognition efficiency. The designers of the present invention combine books, newspapers, clothes and Each character in various pictures such as screenshots is cut into 96*96 pixels to extract character pixel feature information, and the extraction rate is close to 100%.

Description

Technical field [0001] The present invention relates to the field of computer technology, in particular to the field of pattern recognition and deep learning, and more particularly to a method based on OCR glyph-like text recognition. Background [0002] Optical Character Recognition (OCR) is a combination of optical technology and computer technology to convert image files printed on paper into text files, OCR recognition can be used for automatic scanning and long-term storage of bank bills, a large number of documents, archival files, tax bills and other bills. [0003] OCR recognition is usually measured by recognition rate, recognition speed, layout understanding and layout refactoring. This technology has a relatively good recognition rate for general characters, but there are still certain technical problems in the field of Chinese characters with rich structure and glyphs, especially for characters with similar glyphs, such as: (noon, dry, dry), (run, bubble, cannon) and...

AI Technical Summary

Problems solved by technology

This technology has a relatively good recognition rate for general characters, but there are still certain technical difficulties in the field of Chinese characters with rich structures and fonts, especially for similar fonts, such as: (午, gan, gan), (run, bubble, cannon) ) and other characters have the problems of low recognition efficiency and low accuracy

In addition, the existing technology cannot judge characters with the same glyph and different fonts. It is very easy to make mistakes when recognizing the same glyph and different fonts. The results of repeated recognition are different. Sometimes manual intervention is required to correct errors, which greatly reduces the accuracy of recognition.

Images