Unlock instant, AI-driven research and patent intelligence for your innovation.

Text recognition method and device

A text recognition and text technology, applied in the computer field, can solve the problem of low text detection accuracy and achieve the effect of improving text detection rate and accuracy

Active Publication Date: 2021-07-30
BEIJING CENTURY TAL EDUCATION TECH CO LTD
View PDF13 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] In view of this, the embodiment of the present invention provides a text recognition scheme to solve the problem in the prior art that the accuracy of text detection is not high for images containing text topics, especially for images with poor quality

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Text recognition method and device
  • Text recognition method and device
  • Text recognition method and device

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0022] refer to figure 1 , shows a flow chart of steps of a text recognition method according to Embodiment 1 of the present invention.

[0023] The text recognition method of the present embodiment comprises the following steps:

[0024] Step S102: Obtain a text image to be detected.

[0025] Wherein, the text image to be detected contains information of a plurality of characters, and the characters include but not limited to one or more of characters, symbols, numbers, and letters.

[0026] In the embodiment of the present invention, the text image to be detected is an image of a question title. But not limited thereto, other text images are also applicable to the solution of this embodiment of the present invention.

[0027] Step S104: Perform multi-scale transformation on the text image to be detected to obtain multiple sub-text images of different sizes.

[0028] Multi-scale image technology, also known as multi-resolution technology (MRA), refers to the use of multi-...

Embodiment 2

[0040] refer to figure 2 , shows a flow chart of steps of a text recognition method according to Embodiment 2 of the present invention.

[0041] The text recognition method of the present embodiment comprises the following steps:

[0042] Step S202: training a convolutional neural network model for text detection on text images.

[0043] This step is optional. As mentioned above, in actual use, the convolutional neural network model trained by a third party can also be directly used for text detection.

[0044] During training, a large number of text images can be automatically generated using methods such as freetype and pygame based on the existing topic text data; image blurring, image rotation, scaling, distortion and other geometric deformations, image contrast transformation, image noise pollution, etc. One or more of the techniques enhances the diversity of text images. Thus, a large batch of sample images used for training the convolutional neural network model is ...

Embodiment 3

[0104] refer to Figure 8 , shows a structural block diagram of a text recognition device according to Embodiment 3 of the present invention.

[0105] The text recognition device of this embodiment includes: a first acquisition module 302, configured to acquire a text image to be detected, wherein the text image contains information of a plurality of characters; a second acquisition module 304, configured to Perform multi-scale transformation of the text image to obtain multiple sub-text images of different sizes; the third acquisition module 306 is used to perform text detection on each sub-text image using a convolutional neural network model, and obtain each character in each sub-text image Corresponding candidate text detection frame; determination module 308, for carrying out non-maximum suppression NMS processing to multiple candidate text detection frames of all sub-text images of the same character, and filtering the processed candidate text detection frame, Determine...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

Embodiments of the present invention provide a text recognition method and device, wherein the text recognition method includes: acquiring a text image to be detected, wherein the text image contains information of a plurality of characters; Scale transformation, to obtain multiple sub-text images of different sizes; use the convolutional neural network model to perform text detection on each sub-text image, and obtain the candidate text detection frame corresponding to each character in each sub-text image; for the same character Multiple candidate text detection frames of all subtext images are subjected to non-maximum suppression NMS processing, and the processed candidate text detection frames are filtered to determine an effective text detection frame; based on the effective text detection frame, the text is The image is subjected to text recognition, and the text recognition result is obtained. Through the embodiments of the present invention, the accuracy of text detection and recognition for text images is greatly improved.

Description

technical field [0001] The embodiments of the present invention relate to the field of computer technology, and in particular to a text recognition method and device. Background technique [0002] With the development of computer and Internet technology, it has become a trend to assist learning and teaching with the help of computers and networks. Among them, taking pictures and searching questions has gradually become an important means for students to answer questions in this kind of learning and teaching process. [0003] Photo search refers to the process in which users take pictures of questions through mobile phones and other devices, submit them to the learning platform, and the learning platform searches the database according to the content of the photos and returns the stems and analysis of the corresponding questions. One of the key technologies in the process of questioning. [0004] The existing text detection technology mainly realizes text detection based on...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Patents(China)
IPC IPC(8): G06K9/00G06K9/62
CPCG06V30/414G06F18/29
Inventor 高立宁
Owner BEIJING CENTURY TAL EDUCATION TECH CO LTD