TEXT RECOGNITION METHOD, TEXT RECOGNITION APPARATUS, ELECTRONIC DEVICE, AND STORAGE MEDIUM

To provide a text recognition method, a text recognition apparatus, an electronic device, and a storage medium.SOLUTION: A method includes the steps of: extracting character prediction regions of a text line image using an existing text recognition model, to obtain a character region set; calculatin...

Full description

Saved in:
Bibliographic Details
Main Authors XU HUI, ZHANG FAEN, LU QIANG, SUN TIANQI, YUAN ZHICHAO
Format Patent
LanguageEnglish
Japanese
Published 13.09.2021
Subjects
Online AccessGet full text

Cover

Loading…
More Information
Summary:To provide a text recognition method, a text recognition apparatus, an electronic device, and a storage medium.SOLUTION: A method includes the steps of: extracting character prediction regions of a text line image using an existing text recognition model, to obtain a character region set; calculating horizon-distance IOU between the character prediction regions in the character region set on the basis of a position of each character prediction region in the character region set; removing overlapped character prediction regions from the character region set, on the basis of the horizon-distance IOU between the character prediction regions in the character region set, to obtain a character position set; and ranking and outputting character classes corresponding to the character prediction regions in the character position set on the basis of positions of the character prediction regions in the character position set. According to the technical idea of the embodiment of the present application, accuracy of text recognition is improved.SELECTED DRAWING: Figure 2 【課題】本出願は、テキスト認識方法及びテキスト認識装置、電子設備、記憶媒体を提供する。【解決手段】該方法は、既設のテキスト認識モデルにより、テキスト行画像の文字予測領域を抽出し、文字領域集合を取得するステップと、文字領域集合における各文字予測領域の位置に基づいて、文字領域集合における文字予測領域間の水平距離IoUを算出するステップと、文字領域集合における文字予測領域間の水平距離IoUに基づいて、文字領域集合における重なりがあった文字予測領域を除去して、文字位置集合を得るステップと、文字位置集合における文字予測領域に対応する文字クラスを、文字位置集合における文字予測領域の位置に基づいて順位をつけて出力するステップとを含む。本出願の実施例の技術案によって、テキスト認識の精度を向上させた。【選択図】図2
Bibliography:Application Number: JP20200085898