Method and device for recognizing graph in image, computer equipment and storage medium

A graphic recognition and image-based technology, applied in the field of image recognition, can solve problems such as unsatisfactory recognition accuracy, achieve the effects of improving recognition accuracy and recognition rate, fast convergence speed, and suppressing useless information

Active Publication Date: 2021-05-14
GREE ELECTRIC APPLIANCES INC +1
View PDF8 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] In order to improve the recognition accuracy and recognition rate of two-dimensional codes, traditional methods are often used in two-dimensional code recognition technology, such as threshold-based segmentation methods, edge detection methods, etc. Although these methods are fast in recognition, they are difficult to identify Accuracy is not very good

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
  • Method and device for recognizing graph in image, computer equipment and storage medium
  • Method and device for recognizing graph in image, computer equipment and storage medium
  • Method and device for recognizing graph in image, computer equipment and storage medium

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0049] The graphic recognition method in the image provided by this application can be applied to such as figure 1 shown in the application environment. Wherein, the computer 102 communicates with the server 104 through the network. Wherein, the terminal 102 can be, but not limited to, various personal computers, servers, notebook computers, smart phones, tablet computers and portable wearable devices, and the server 104 can be realized by an independent server or a server cluster composed of multiple servers. The self-inhibiting residual neural network and the attention mechanism neural network are deployed on the server 104 .

[0050] The terminal 102 first sends a large number of learning images marked with the boundary coordinates of the graphics to the self-inhibiting residual neural network and the attention mechanism neural network on the server 104, the server 104 obtains the learning images, and uses the self-inhibiting residual neural network and the attention mecha...

Embodiment 2

[0053] In this example, if figure 2 As shown, a pattern recognition method in an image is provided, which includes:

[0054] Step 210, acquiring an image to be recognized, wherein the image to be recognized contains a figure to be recognized.

[0055] In this embodiment, the image to be recognized is an image that needs to be separated from the graphics contained therein. The image to be recognized includes graphics to be recognized and background graphics. The graphics to be recognized are graphics that can carry information. The background graphics can also be called backgrounds. The background is graphics other than graphics to be recognized in the image to be recognized. The graphics are graphics that need to be separated from the background in the image to be recognized, and the graphics to be recognized are graphics that carry information, while the background graphics do not have graphics that carry information. In this embodiment, the to-be-recognized image is acqui...

Embodiment 3

[0081]In this embodiment, the image is firstly preprocessed, grayscaled and filtered, and image preprocessing can reduce noise and reduce the amount of calculation in the next step. Then mark a large number of QR code pictures for the supervised learning of the algorithm. It is then input to the self-suppressing residual network layer. The self-suppressing residual network layer can use a deeper network to capture deeper information. Compared with the ordinary residual network, the self-suppressing residual network converges faster, making the two-dimensional code The recognition accuracy has been improved. When it is passed to the attention mechanism layer, the vector representing the weight of the relevant image is obtained. These vectors can obtain more detailed information of the target that needs to be focused on, while suppressing other useless information. The two-dimensional code recognition layer uses the trained model to output the coordinates of the boundary points...

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 provides a method and a device for recognizing a graph in an image, computer equipment and a storage medium. The method comprises the following steps: acquiring a to-be-identified image; analyzing the to-be-recognized image through a self-suppression residual neural network to obtain a graphic feature of a to-be-recognized graph in the to-be-recognized image; analyzing through an attention mechanism neural network to obtain a to-be-recognized image, and obtaining a weight vector of the graphic feature; and obtaining boundary coordinates of the to-be-recognized graph in the to-be-recognized image based on the weight vector of the graphic feature. Through the self-suppression residual neural network, a deeper network can be used to capture deeper information, because the self-suppression residual neural network is faster in convergence speed, the features of a two-dimensional code can be accurately obtained, and through the attention mechanism neural network, a weight vector representing the features of the two-dimensional code is obtained, therefore, the coordinate of the boundary point of the two-dimensional code can be determined based on the weight vector of the features of the two-dimensional code, and the two-dimensional code in the to-be-recognized image is accurately separated from the background.

Description

technical field [0001] The invention relates to the technical field of image recognition, in particular to a pattern recognition method, device, computer equipment and storage medium in an image. Background technique [0002] With the continuous development of the mobile Internet, the application of QR codes has been greatly developed. QR codes can be seen in various fields of mobile Internet applications. For example, you need to recognize a QR code to log in to an account; you need to recognize a QR code to add friends; you need to recognize a QR code to rent a shared bicycle. QR code recognition has penetrated into many aspects of life. [0003] Limited by the display pixels of some display devices or the recognition distance of the QR code, the QR code cannot be accurately recognized. There are many factors that affect the recognition of QR codes. For example, the background of the picture to be recognized is similar to the background of the QR code, and the proportion...

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): G06K7/14G06K9/46G06K9/62G06N3/04G06N3/08
CPCG06K7/1417G06K7/1439G06N3/04G06N3/08G06V10/44G06F18/241
Inventor 唐光远陈海波罗琴张俊杰李润静
Owner GREE ELECTRIC APPLIANCES INC
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