Text detection method, system and equipment based on multi-receptive field depth characteristics and medium

A multi-receptive field, text detection technology, applied in image data processing, instruments, biological neural network models, etc., can solve the problem of low text box detection accuracy and recall rate, text box can not completely cover the text area, text box regression exists Position deviation and other problems, to achieve good text detection performance, minimize information loss, and expand the effect of the receptive field

Inactive Publication Date: 2019-07-16
SOUTH CHINA UNIV OF TECH
View PDF3 Cites 36 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] At present, the disadvantage of implementing text boxes based on image segmentation methods is that the use of single-dimensional convolutional features will cause position deviation in text box regression, or the text box cannot completely cover the text area. The problem of low accuracy and recall of text box detection with multi-scale and complex background

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 detection method, system and equipment based on multi-receptive field depth characteristics and medium
  • Text detection method, system and equipment based on multi-receptive field depth characteristics and medium
  • Text detection method, system and equipment based on multi-receptive field depth characteristics and medium

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0077] Such as figure 1 As shown, the present embodiment provides a text detection method based on multi-receptive field depth features, the method includes the following steps:

[0078] S101. Data acquisition

[0079] Specifically, one of the text detection databases widely used by academic circles in the field of text detection in natural scenes is obtained, and the text detection database is used as a network training database.

[0080] In this embodiment, the text detection database adopts the RCTW17 competition database. The name of the data set is CTW-12k, which consists of 12263 natural scene pictures containing Chinese, most of which are directly taken by cameras or mobile phones, and a small part is for generating images. And each image contains at least one line of Chinese.

[0081] The labeling of the data is done manually with the labeling tool. A text line is marked by drawing a quadrilateral instead of a word. The content of each text line is marked with a UTF-...

Embodiment 2

[0135] Such as Figure 4 As shown, this embodiment provides a text detection system based on multi-receptive field depth features, the system includes an acquisition unit 401, a construction unit 402, a training unit 403, a calculation unit 404 and a processing unit 405, and the specific functions of each unit are as follows :

[0136] The obtaining unit 401 is configured to obtain a text detection database, and use the text detection database as a network training database.

[0137] The building unit 402 is used to build a deep network model with multiple receptive fields.

[0138] The training unit 403 is configured to input the natural scene text pictures in the network training database and the corresponding true value data of the text box coordinates into the multi-receptive field depth network model for training.

[0139] The calculation unit 404 is used to calculate the image mask for segmentation through the trained multi-receptive field depth network model, obtain t...

Embodiment 3

[0143] This embodiment provides a computer device, which may be a computer, such as Figure 5 As shown, it includes a processor 502 connected through a system bus 501, a memory, an input device 503, a display 504, and a network interface 505. The processor is used to provide computing and control capabilities. The memory includes a non-volatile storage medium 506 and internal Memory 507, the non-volatile storage medium 506 stores an operating system, computer programs and databases, the internal memory 507 provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium, and the processor 502 executes memory storage During the computer program, realize the text detection method of above-mentioned embodiment 1, as follows:

[0144] Obtain a text detection database, and use the text detection database as a network training database;

[0145] Build a multi-receptive field deep network model;

[0146] Input the natural sc...

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

The invention discloses a text detection method, system and device based on multi-receptive field depth characteristics and a medium, and the method comprises the steps: obtaining a text detection database, and taking the text detection database as a network training database; building a multi-receptive field depth network model; inputting a natural scene text picture and corresponding textbox coordinate true value data in the network training database into a multi-receptive field depth network model for training; calculating an image mask for segmentation through the trained multi-receptive field depth network model to obtain a segmentation result, and converting the segmentation region into a regression textbox coordinate; and counting the textbox size of the network training database, designing a textbox filtering condition, and screening out a target textbox according to the textbox filtering condition. The method fully utilizes the feature learning capability and classification performance of the deep network model, combines the characteristics of image segmentation, has the characteristics of high detection accuracy, high recall rate, strong robustness and the like, and has agood text detection effect in a natural scene.

Description

technical field [0001] The invention relates to a text detection method, system, equipment and medium based on multi-receptive field depth features, belonging to the field of image text analysis. Background technique [0002] With the development of computer vision technology, the application of image understanding technology is becoming more and more extensive. As an information-dense point of an image, text information is the key to image understanding. Text localization is an important step in extracting key information from images. However, due to the variety of backgrounds and the uncertainty of size and direction, the positioning of Chinese text in natural scene images faces the following challenges: 1) changes in text format and aspect ratio of text lines; 2) text boxes with multiple directions 3) The size of the text box varies greatly; 4) The confusion caused by the similarity between the text and the background. [0003] There are currently two main methods for ...

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 Applications(China)
IPC IPC(8): G06K9/62G06N3/04G06T7/11
CPCG06T7/11G06N3/045G06F18/253G06F18/214
Inventor 高学杨雯
Owner SOUTH CHINA UNIV OF TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap