Chinese printing style formula identification method

Abstract

The invention provides an identification method to identify Chinese printed formula, including analysis of typeface, character recognition and mathematical formula recognition three modules; typeface analysis module two-value pretreats BMP images, split out literal block, image block and list block using projection method combined with bottom-up typeface analysis algorithm, preserve image block and list block; Chinese character recognition module false merge rows against literal block, select segmentation parameters, extract characteristics and recognize Chinese characters, record the rejected results, combine adjacent rejected results in the same row in order to locate formula region; mathematical formula recognition is to extract,divide formula characters in the rejected character region, merger some characters and recognize; finally obtain the relationship between characters through structure analysis of formula characters, and output the final one-dimensional character strings. the identification effect of this invention is famous after test.

Description

(1) Technical field [0001] The invention relates to a recognition method, in particular to a method for recognizing the content of printed Chinese documents, especially for printed mathematical formulas. (2) Background technology [0002] In 1929, Tausheck obtained a patent for Optical Character Recognition (OCR). Because it is easy to be accepted and mastered by people, it has increasingly become the focus of people's research together with speech recognition and behavior recognition. After nearly a century of development, OCR has become one of the most active research contents in the field of pattern recognition today. So far, the pure one-dimensional character recognition technology has been quite mature, and there are relatively well-established recognition systems (such as Ziguang, Hanwang, etc.), which have a high recognition rate, but these systems cannot recognize mathematical formulas in documents. Therefore, this two-dimensional structural mathematical formula has...

AI Technical Summary

Problems solved by technology

So far, the pure one-dimensional character recognition technology has been quite mature, and there are relatively well-formed recognition systems (such as Ziguang, Hanwang, etc.), which have a high recognition rate, but these systems cannot recognize the mathematical formulas in the document

Therefore, this two-dimensional structural mathematical formula has become a bottleneck restricting the development of OCR technology.

Images