A component-based handwritten Chinese character evaluation method

Abstract

The application relates to a component-based handwritten Chinese character evaluation method, relating to computer graphics. The method comprises the following steps: 1) performing 'random movement' and'size change' adjustment on partial components of a given handwritten Chinese character; 2) using an unaltered standard layout handwritten Chinese character and data generated in step 1) as a data set, pre-training a Chinese character component reconstruction module; 3) using the pre-trained Chinese character component reconstruction module, extracting layout features of the handwritten Chinese character; 4) taking the layout features of the handwritten Chinese character as input, combining a layout evaluation module to perform handwritten Chinese character layout evaluation, and obtaining an evaluation conclusion and modification suggestions; and 5) using the layout evaluation suggestions to adjust the handwritten Chinese character. The method solves the deficiencies of the prior art in the aspects of Chinese character evaluation mode and evaluation result display, effectively shortens the cycle of traditional manual design evaluation templates and reduces the cost. The method has wide application prospects in the fields of calligraphy education, font design, character recognition and the like.

Description

Technical Field

[0001] This invention relates to the field of computer graphics research, and in particular to a component-based method for evaluating handwritten Chinese characters. Background Technology

[0002] Calligraphy uses Chinese characters as its medium, and Chinese characters are ideographic, possessing unique structural regularities and aesthetic characteristics. Calligraphers use brush and ink to depict the beauty of the structure and layout of Chinese characters, showcasing their rhythm and aesthetic appeal. Therefore, evaluating the layout and structure of Chinese characters is of great significance to the art of calligraphy.

[0003] Computational evaluation of handwritten Chinese characters is a method for evaluating the aesthetic quality of calligraphy, providing substantial convenience for Chinese character writing education. Handwritten character evaluation plays a crucial role in many areas, including the generation of handwritten Chinese characters and the beautification of Chinese characters. However, previous research has rarely delved into using character decomposition and layout reconstruction to achieve automatic evaluation of Chinese characters. Early calligraphy evaluation methods mostly employed template matching, comparing the target character with a template character to measure its visual characteristics, such as stroke length, width, angle, number, and order. However, these methods are heavily influenced by the selection of template characters and require the development of corresponding evaluation criteria for each template character, making the evaluation rules overly complex.

[0004] With the development of deep learning, one possible solution for calligraphy evaluation is to utilize deep learning methods, such as convolutional neural networks. However, compared to other languages, Chinese characters have a unique structure because they can be considered as being composed of multiple components. The shape and relationship of these components collectively influence the aesthetics of Chinese characters. Therefore, the evaluation of Chinese characters must consider the rationality of their internal structure. Summary of the Invention

[0005] The purpose of this invention is to address the problem that existing technologies fail to consider the relationships between components when evaluating handwritten Chinese characters. It provides a component-based evaluation method for handwritten Chinese characters, aiming to simulate the layout problem of handwritten Chinese characters, use deep learning technology to evaluate the components of handwritten Chinese characters as a whole, and provide corresponding evaluation conclusions and modification suggestions.

[0006] This invention includes the following steps:

[0007] 1) Generate handwritten Chinese characters with layout problems: Adjust the components of a given handwritten Chinese character by "randomly moving" and "changing the size" to obtain handwritten Chinese character data with layout problems;

[0008] 2) Pre-trained Chinese character component reconstruction module: Using the standard layout handwritten Chinese characters without any changes and the data generated in step 1) as the dataset, the Chinese character component reconstruction module is pre-trained; the input of the Chinese character component reconstruction module is a multi-channel image containing standard layout handwritten Chinese characters and problem layout handwritten Chinese characters, and the network output is the features of the problem layout handwritten Chinese character components and the reconstructed standard layout handwritten Chinese characters.

[0009] 3) Extract layout features: Using the pre-trained Chinese character component reconstruction module from step 2), extract the layout features of handwritten Chinese characters;

[0010] 4) Layout evaluation: Using the layout features of handwritten Chinese characters as input, the layout evaluation module is used to evaluate the layout of handwritten Chinese characters. Based on the pre-set classification criteria, evaluation conclusions and modification suggestions are obtained.

[0011] 5) Adjust according to evaluation suggestions: Based on the evaluation conclusions and modification suggestions obtained in step 4), adjust the handwritten Chinese characters according to the pre-set classification standards, including adjusting the position and size of components.

[0012] In step 1), the components of the given handwritten Chinese character refer to the 3755 commonly used Chinese characters at level 1 in GB2312_80, and the strokes of each handwritten Chinese character are saved independently.

[0013] In step 1), "random movement" refers to randomly moving one or more components of a handwritten Chinese character to change the layout of the components.

[0014] In step 1), "size change" refers to adjusting the size of one or more components in a handwritten Chinese character.

[0015] In step 2), the unaltered standard layout handwritten Chinese characters refer to the original handwritten Chinese characters that have not been adjusted by "random movement" and "size change".

[0016] In step 3), the extraction of layout features of handwritten Chinese characters refers to sending the input handwritten Chinese characters into the Chinese character component reconstruction module, obtaining the problematic layout features of handwritten Chinese characters, and then subtracting them from the features of standard handwritten Chinese character layouts to obtain the image features.

[0017] In step 4), the input of the layout evaluation module is the layout features of the handwritten Chinese characters. In the layout evaluation module, the positional relationship between each component of the handwritten Chinese characters is classified one by one until the classification of the relationship between all components is completed, and corresponding evaluation conclusions and text suggestions are given.

[0018] In step 4), the pre-set classification criteria include: (1) the position is appropriate compared with other components; (2) the position is higher compared with other components; (3) the position is lower compared with other components; (4) the position is to the left compared with other components; (5) the position is to the right compared with other components; (6) the position is smaller compared with other components; (7) the position is larger compared with other components.

[0019] Compared with the prior art, the present invention has the following outstanding technical effects:

[0020] This invention proposes a component-based Chinese character evaluation method, which evaluates handwritten Chinese characters based on their component relationships and provides evaluation conclusions and modification suggestions. First, it proposes a Chinese character structure layout reconstruction module, using reconstruction features instead of template words to evaluate various handwriting styles without the need for extensive template word feature extraction. Second, it proposes a handwritten Chinese character evaluation method based on layout reconstruction and designs a graph-structure-based Chinese character component layout evaluation module, enabling more accurate evaluation of the layout structure of handwritten Chinese characters. Third, addressing the issues of using scores to evaluate handwritten Chinese characters being too vague and using template matching being too inflexible, it employs a commentary evaluation and modification suggestion method, which has significant practical value for calligraphy teaching. This invention, through the comprehensive application of preprocessing, feature extraction, and component layout evaluation steps, achieves intelligent evaluation of the component relationships of handwritten Chinese characters, providing an efficient solution for handwritten Chinese character evaluation and having significant implications for the application of computer graphics. Attached Figure Description

[0021] Figure 1 This is a flowchart of the method of the present invention.

[0022] Figure 2 This diagram illustrates a method for moving components of handwritten Chinese characters. It includes: a. the original character and its corresponding IDS; b. random movement of the components; and c. resizing the components.

[0023] Figure 3 A diagram of the module for reconstructing Chinese character components.

[0024] Figure 4 This is a diagram of a module for evaluating the layout of handwritten Chinese characters. Specifically: a) the sequence of components obtained after segmenting the handwritten Chinese character; b) the spatial features of the handwritten Chinese character obtained by reconstructing the layout of the components; and c) the input of the spatial features of the handwritten Chinese character into the network for layout relationship prediction.

[0025] Figure 5 Example image of handwritten Chinese characters modified based on evaluation conclusions. Detailed Implementation

[0026] The following description of at least one exemplary embodiment is merely illustrative and in no way restrictive of the present invention or its application or use. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present invention without creative efforts fall within the scope of protection of the present invention.

[0027] The present invention constructs a system for evaluating and adjusting the layout of handwritten Chinese characters, which can automatically identify the layout problems of handwritten Chinese characters and provide specific adjustment suggestions. By combining deep learning technology and image processing algorithms, this system will provide users with convenient and efficient guidance for Chinese character writing.

[0028] The system mainly consists of the following modules:

[0029] The data preprocessing module is used to generate a dataset of handwritten Chinese characters with layout problems and preprocess the dataset.

[0030] The Chinese character component reconstruction module is used for pre-training to learn the layout features of handwritten Chinese characters.

[0031] The layout feature extraction module is used to extract the layout features of the handwritten Chinese characters to be evaluated.

[0032] The layout evaluation module is used to evaluate the extracted layout features and give adjustment suggestions.

[0033] The adjustment execution module is used to adjust the handwritten Chinese characters according to the evaluation suggestions.

[0034] As Figure 1 , the component-based method for evaluating handwritten Chinese characters according to the embodiments of the present invention includes the following steps:

[0035] 1) Generate handwritten Chinese characters with layout problems: Adjust the components of the given handwritten Chinese characters by "random movement" and "size change" to obtain handwritten Chinese characters with layout problems:

[0036] Step 1: Read the SVG-formatted handwritten Chinese characters with stroke markings and the corresponding ideographic description sequences of the handwritten Chinese characters. These stroke-marked ideographic sequences are used for subsequent recognition and operation of Chinese character components. Figure 2 Taking the character "陷" as an example in

[0037] Step 2: Randomly select one of the components of the read handwritten Chinese character as the adjustment object.

[0038] Step 3: Move the selected component or adjust the size of the component. The moving distance and the proportion of size change should be randomly set within a reasonable range to ensure that the generated handwritten Chinese characters with layout problems have diversity and authenticity. (See Figure 2 b in Figure 2 (c in the text)

[0039] 2) Using the standard layout of handwritten Chinese characters without any changes and the data generated in step 1) as the dataset, the Chinese character component reconstruction module is pre-trained:

[0040] Step 1: Establish a Chinese character component reconstruction module. This module should be capable of recognizing and reconstructing handwritten Chinese characters in both standard and problematic layouts. For example... Figure 3 As shown, construct a layout refactoring encoder. Feature extraction is performed on handwritten Chinese characters with layout problems to obtain layout features. Using a pre-trained encoder Extract features from standard layout handwritten Chinese characters and problem layout handwritten Chinese characters. and In addition, a component classifier is used. To determine whether the components of a handwritten Chinese character are accurately included in the content features, a reconstruction decoder is used. Decoding the features of Chinese characters in a standard layout yields reconstructed standard Chinese characters. (Reconstruction encoder) Pre-trained encoder Component classifier Reconstructing the decoder The model structure can be referenced from the Convolutional Neural Network (CNN) in deep learning.

[0041] Step 2: Use standard layout handwritten Chinese characters and problem layout handwritten Chinese characters as training data. The problem layout handwritten Chinese characters are generated in step 1). Use character layout loss as the loss function to train the Chinese character component reconstruction module. The input includes standard layout handwritten Chinese character images, problem layout handwritten Chinese character images, the component sequence of the Chinese characters, and the name of the Chinese characters.

[0042] The calculation of character layout loss should comprehensively consider both character loss and component loss during the character encoding and decoding process to ensure that the features of handwritten characters during encoding are consistent with the features of their corresponding standard characters. A VGG16 network model is used as the feature extractor.

[0043] The formula for calculating character layout loss is as follows:

[0044]

[0045] in, This represents the character loss during the character encoding and decoding process; This represents component loss during the character encoding and decoding process; Used to penalize inconsistencies between VGG features and their corresponding standard character features during handwritten character encoding; , and For pre-defined hyperparameters, , , .

[0046] Character layout loss , and The calculation formula is:

[0047]

[0048]

[0049]

[0050] in, This represents the Chinese characters predicted by the Chinese character component reconstruction module. This represents the tag value of the Chinese character. This represents the sequence of Chinese character components predicted by the Chinese character component reconstruction module; The label value representing the sequence of components of the Chinese character; Represents the VGG16 network model; The layout of the question is represented by handwritten Chinese characters. Images representing standard layout handwritten Chinese characters.

[0051] 3) Using the pre-trained Chinese character component reconstruction module from step 2), extract the layout features of handwritten Chinese characters:

[0052] Step 1: Enter a handwritten Chinese character image and its name.

[0053] Step 2: Input the handwritten Chinese character image and name into the pre-trained Chinese character component reconstruction module.

[0054] Step 3: The Chinese character component reconstruction module extracts handwritten Chinese character images and processes them through the layout reconstruction encoder. Image features and its pre-trained encoder Image features obtained By subtracting the differences, we obtain the layout characteristics of the processed handwritten Chinese characters.

[0055] 4) Using the layout features of handwritten Chinese characters as input, and combining this with the layout evaluation module, the layout of handwritten Chinese characters is evaluated. Based on pre-set classification criteria, evaluation conclusions and modification suggestions are obtained (e.g., ...). Figure 4 ):

[0056] Step 1: Establish a layout relationship prediction network. (The following text appears to be a separate, unrelated section: "Will...") Figure 4As shown in b, the spatial features of the reconstructed handwritten Chinese characters after component layout are combined with the structural features of the middle layer of the graph neural network to form a message matrix of a node. This matrix is ​​used to represent the spatial relationships and structural features between Chinese character components, and is defined as follows:

[0057]

[0058] in, For structural features, For spatial features, The hyperparameter is set to 0.5. Component nodes in a graph neural network are defined as nodes. of Dimension indicates that the feature update of each node is related to its neighboring nodes. Correspondingly, features are updated using learnable weights and biases, and the update formula is:

[0059]

[0060]

[0061]

[0062] in, For message matrix, This is a hyperparameter, set to 0.4; For degree matrix, It is an adjacency matrix; Features of layer t; For learnable weights, As learnable biases, they are initialized with Gaussian randomness; functions Accumulate information passed from the graph neural network layers to update the radical representation; the graph network performs layout relationship prediction, such as... Figure 4 As shown in c.

[0063] Step 2: The layout relationship prediction network predicts the relationship categories between handwritten Chinese character components based on the layout features of the handwritten Chinese characters. These relationship categories include (1) the position is appropriate compared to other components; (2) the position is higher compared to other components; (3) the position is lower compared to other components; (4) the position is to the left compared to other components; (5) the position is to the right compared to other components; (6) the position is smaller compared to other components; and (7) the position is larger compared to other components.

[0064] Step 3: The layout evaluation module transforms the predicted category information according to a predefined evaluation method, converting it into specific evaluation conclusions and modification suggestions.

[0065] Step 4: Output evaluation conclusions and modification suggestions. The evaluation conclusions clearly point out the problems in the layout of handwritten Chinese characters, and the modification suggestions provide specific directions and steps for adjustment.

[0066] 5) Adjust handwritten Chinese characters based on layout evaluation suggestions:

[0067] Step 1: Using the evaluation conclusions and modification suggestions output by the layout evaluation module, adjust the position or size of the problematic components of handwritten Chinese characters. During the adjustment process, ensure that the relative position and size relationship between the components conforms to the Chinese character writing standards.

[0068] Step 2: Verify the adjusted handwritten Chinese characters to ensure the layout meets the requirements of the evaluation suggestions. If necessary, perform multiple adjustments and verifications until a satisfactory layout is achieved.

[0069] Figure 5 Example images of handwritten Chinese characters modified based on evaluation conclusions are provided. By comparing the handwritten characters before and after the adjustment, it is clear that the layout of the adjusted characters is more reasonable and aesthetically pleasing. Furthermore, Figure 5 The examples also demonstrate the effectiveness and accuracy of this method in recognizing and adjusting the layout of handwritten Chinese characters.

[0070] Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this invention. Furthermore, those skilled in the art will recognize that, based on the ideas of this invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. A component-based method for evaluating handwritten Chinese characters, characterized in that... include: 1) Adjust the components of a given handwritten Chinese character by "randomly moving" and "changing the size" to obtain a handwritten Chinese character with layout problems; 2) Using the standard layout of handwritten Chinese characters without any changes and the data generated in step 1) as the dataset, the Chinese character component reconstruction module is pre-trained; The input to the Chinese character component reconstruction module is a multi-channel image, which includes standard layout handwritten Chinese characters and problem layout handwritten Chinese characters. The network output is the features of the problem layout handwritten Chinese characters and the reconstructed standard layout handwritten Chinese characters. 3) Using the pre-trained Chinese character component reconstruction module from step 2), extract the layout features of handwritten Chinese characters; The extraction of layout features of handwritten Chinese characters refers to sending the input handwritten Chinese characters into the Chinese character component reconstruction module, obtaining the problematic layout features of handwritten Chinese characters, and then subtracting the layout features of standard handwritten Chinese characters to obtain the image features. 4) Using the layout features of handwritten Chinese characters as input, the layout evaluation module is used to evaluate the layout of handwritten Chinese characters. Based on the pre-set classification criteria, the evaluation conclusions and modification suggestions are obtained. The evaluation module is: to establish a layout relationship prediction network, and to combine the spatial features of handwritten Chinese characters reconstructed by component layout with the structural features of the middle layer of the graph neural network into a message matrix of a node. This matrix is ​​used to represent the spatial relationship and structural features between Chinese character components. The layout relationship prediction network predicts the relationship categories between components of handwritten Chinese characters based on the layout features of the handwritten Chinese characters. It classifies the positional relationships between each component of the handwritten Chinese character one by one until the classification of the relationships between all components is completed, and gives corresponding evaluation conclusions and text suggestions. 5) Adjust handwritten Chinese characters based on layout evaluation suggestions.

2. The component-based handwritten Chinese character evaluation method as described in claim 1, characterized in that... In step 1), the components of the given handwritten Chinese character refer to the 3755 commonly used Chinese characters at level 1 in GB2312_80, and the strokes of each handwritten Chinese character are saved independently.

3. The component-based handwritten Chinese character evaluation method as described in claim 1, characterized in that... In step 1), "random movement" refers to randomly moving one or more components of a handwritten Chinese character to change the layout of the components.

4. The component-based handwritten Chinese character evaluation method as described in claim 1, characterized in that... In step 1), "size change" refers to adjusting the size of one or more components in a handwritten Chinese character.

5. The component-based handwritten Chinese character evaluation method as described in claim 1, characterized in that... In step 2), the unaltered standard layout handwritten Chinese characters refer to the original handwritten Chinese characters that have not been adjusted by "random movement" and "size change".

6. The component-based handwritten Chinese character evaluation method as described in claim 1, characterized in that... In step 4), the input of the layout evaluation module is the layout features of the handwritten Chinese characters. In the layout evaluation module, the positional relationship between each component of the handwritten Chinese characters is classified one by one until the classification of the relationship between all components is completed, and corresponding evaluation conclusions and text suggestions are given.

7. The component-based handwritten Chinese character evaluation method as described in claim 1, characterized in that... In step 4), the pre-set classification criteria include: (1) the position is appropriate compared with other components; (2) the position is higher compared with other components; (3) the position is lower compared with other components; (4) the position is to the left compared with other components; (5) the position is to the right compared with other components; (6) the position is smaller compared with other components; (7) the position is larger compared with other components.

8. The component-based handwritten Chinese character evaluation method as described in claim 1, characterized in that... In step 5), the adjustment of handwritten Chinese characters based on the layout evaluation recommendations refers to the corresponding adjustment of handwritten Chinese characters according to the seven classification criteria pre-set in claim 7, including the adjustment of the position and size of components.

Images