Table correction and recognition method based on combination of image processing and deep learning

A technology of deep learning and image processing, applied in character and pattern recognition, instruments, biological neural network models, etc., can solve the problem of low detection accuracy and achieve the effect of improving accuracy

Pending Publication Date: 2020-12-04
晶璞(上海)人工智能科技有限公司
View PDF0 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0002] Forms are a means of highly refined and concentrated expression of information, with the characteristics of visualization and convenient communication. Forms are widely used in the collection of various data. With the advancement of paperless office, paper forms are converted into electronic files. Tables are an inevitable trend of informatization. The current method of identifying tables in images is to use equipment to obtain table images, and then analyze the optical characteristics of the entire page digital image to detect the position of the row and column frame lines of the table to detect the layout structure of the table. Generally, it is only applicable to The quality of the input image is relatively good, the position and layout of the table are relatively fixed, and the frame line of the table is relatively obvious. If the image has problems such as reversed text direction, tilt, perspective distortion, etc., the detection accuracy is low

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
  • Table correction and recognition method based on combination of image processing and deep learning
  • Table correction and recognition method based on combination of image processing and deep learning
  • Table correction and recognition method based on combination of image processing and deep learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0058] see figure 1 , the present invention discloses a method for table correction and recognition based on the combination of image processing and deep learning. The method includes the following steps:

[0059] [Step 110] Obtain the original image data of the form.

[0060] Get the original form image, obtained with a scanner, such as Image 6 shown.

[0061] [Step 120] Preprocessing of table images.

[0062] see figure 2 , step 120 specifically includes the following steps:

[0063] Step 210, determine the character direction. Using the lightweight convolutional neural network MobileNet to train a model for detecting text in 4 directions (0 degrees, 90 degrees, 180 degrees, 270 degrees), the original image is passed through the text direction detection model to get the correct direction of the text, and the original image is processed rotate;

[0064] Step 220, rotate the tilted image. Use the canny edge detection operator to detect the boundary, look for straight...

Embodiment 2

[0083] On the smart phone, the mobile phone has a 720,000-pixel camera, and the method of the invention can be used to correct and identify the form image captured. Figure 9 is the form image captured by the mobile phone, Figure 10 is a schematic diagram of the association between text blocks and tables, Figure 11 It is a screenshot that is finally saved to the excel file.

Embodiment 3

[0085] On the digital camera, using the method of the present invention, the form image correction and recognition can also be performed on the form image obtained by shooting.

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 relates to the technical field of image processing and image recognition, in particular to a table correction and recognition method based on combination of image processing and deep learning, which comprises the following steps: step 110, acquiring original image data of a table; step 120, image preprocessing; step 130, positioning a character area; 140, reconstructing table information; by designing and improving the existing form recognition method, the accuracy of form recognition is improved by performing text direction judgment, inclination correction and perspective distortion processing when the form image is recognized, and the problem that the form recognition accuracy is low after the form image is obtained by using equipment in the existing method for recognizingthe form in the image is solved. The table row and column frame line positions are detected by analyzing the optical characteristics of the whole-page digital image so as to detect the format structure of the table, and the method is generally only suitable for the conditions that the input image quality is good, the table positions and formats are fixed, and the table frame lines are obvious, andthe problems of character direction overturning, inclination, perspective distortion and the like exist in the image.

Description

technical field [0001] The invention relates to the technical field of image processing and image recognition, in particular to a form correction and recognition method based on the combination of image processing and deep learning. Background technique [0002] Forms are a means of highly refined and concentrated expression of information, with the characteristics of visualization and convenient communication. Forms are widely used in the collection of various data. With the advancement of paperless office, paper forms are converted into electronic files. Tables are an inevitable trend of informatization. The current method of identifying tables in images is to use equipment to obtain table images, and then analyze the optical characteristics of the entire page digital image to detect the position of the row and column frame lines of the table to detect the layout structure of the table. Generally, it is only applicable to If the quality of the input image is relatively goo...

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): G06K9/00G06K9/32G06K9/62G06N3/04
CPCG06V30/413G06V30/1478G06N3/045G06F18/214
Inventor 罗宝娟李进文严京旗卞志强张成栋
Owner 晶璞(上海)人工智能科技有限公司
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