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

Deep handwriting identification method and device based on double-tower network

An identification method and handwriting technology, applied in the field of deep neural network identification, can solve the problem of low accuracy of handwriting identification

Active Publication Date: 2021-01-12
安徽深信科创信息技术有限公司 +1
View PDF4 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] In order to solve the above-mentioned problems existing in the prior art, the present invention provides a deep handwriting identification method and device based on the twin-tower network, which solves the problem of low accuracy of handwriting identification

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
  • Deep handwriting identification method and device based on double-tower network
  • Deep handwriting identification method and device based on double-tower network
  • Deep handwriting identification method and device based on double-tower network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0077] like figure 1 As shown, a kind of deep handwriting identification method based on double tower network that the embodiment of the present invention provides, comprises:

[0078] S11, acquiring the first handwriting image to be identified and the second handwriting image of the person to be identified;

[0079] It can be understood that the person to be identified refers to the person to be identified who writes the handwriting to be identified, and the person to be identified refers to a person who may write the handwriting to be identified. The handwriting to be identified refers to the handwriting formed by words, characters, and graphics that need to be identified. The first handwriting image can be obtained by converting the handwriting to be identified into a picture, and the second handwriting image can be obtained by converting the handwriting written by the person to be identified into a picture. Of course, in the transformation process, methods such as taking...

Embodiment 2

[0088] As an optional implementation mode provided by the embodiment of the present invention, before the above step S12, the deep handwriting identification method based on the twin-tower network provided by the embodiment of the present invention further includes:

[0089] Step 1: Binarize the first handwriting image and the second handwriting image respectively;

[0090]Step 2: Perform image channel conversion on the binarized first handwriting image and the binarized second handwriting image, so that the image channel of the first handwriting image matches the image channel required by the first input channel or the second input channel The image channel of the second handwriting image matches the required image channel of the second input channel;

[0091] Wherein, the image channel required by the first input channel is the same as the image channel required by the second input channel.

[0092] It can be understood that when obtaining handwriting pictures, due to the d...

Embodiment 3

[0096] As an optional implementation of the embodiment of the present invention, the preset twin-tower network model includes: an encoding and decoding network, a backbone network, a feature intersection network, and a classification output network, and the input channels of the encoding and decoding network include: the first input channel And the second input channel, the encoding and decoding network is used for data reconstruction of the input sample image, the backbone network extracts feature data after the image information is reconstructed, and the feature cross network is used to combine the features of the two channel networks of the backbone network Perform feature difference calculations on the data, and output recognition results based on the calculation results of feature differences;

[0097] Wherein, the feature data includes: first feature data and second feature data. The encoding and decoding network includes: the first channel network and the second channel...

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 embodiment of the invention provides a deep handwriting identification method and device based on a double-tower network, and the method comprises the steps: obtaining a to-be-identified first handwriting image and a second handwriting image of a to-be-identified person, and inputting the first handwriting image and the second handwriting image into a trained double-tower network model; and obtaining an identification result output by the double-tower network model. In the recognition process, noise interference in the first handwriting image and the second handwriting image is removed ina data reconstruction mode, the robustness is improved, and due to weight sharing in the trained double-tower network model, more attention is paid to the characteristics of global handwriting data, so that similar data is closer, the difference between dissimilar data is increased, and the recognition accuracy is improved. According to the method, the generalization ability of the method is improved, and after data reconstruction, the first feature data and the second feature data of the extracted reconstructed data are fused, so that the feature data value of the similar image is increased,and the recognition result is more accurate.

Description

technical field [0001] The invention belongs to the field of deep neural network identification, and in particular relates to a deep handwriting identification method and device based on a twin-tower network. Background technique [0002] The writing process of adults has been trained for a long time, and a stable writing style has been formed. The biological behavior characteristics of written characters are playing an increasingly important role in judicial identification, medical disputes and other fields. In these fields it is often necessary to authenticate written words, a process known as handwriting identification. [0003] The handwriting identification process of prior art is as follows: [0004] Collect the text written by multiple people, convert the text into pictures, obtain multiple sample sets containing picture data, input the sample sets into the deep learning network model, and iteratively train the deep learning network model until the training cut-off ...

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): G06K9/00G06K9/62G06N3/04G06N3/08
CPCG06N3/084G06V40/33G06V30/10G06N3/045G06F18/214G06F18/2415G06F18/253
Inventor 不公告发明人
Owner 安徽深信科创信息技术有限公司