Handwritten Chinese character image evaluation system based on deep learning
Through the handwritten Chinese character image evaluation system based on deep learning, the Chinese character stroke order is extracted and matched, the user can write the position, and the secondary input module can realize fast and correct writing input, which solves the problems of low recognition rate and slow input speed during continuous writing in the existing technology, and achieves efficient and coherent writing and recognition rate improvement.
Patent Information
- Application Number
- CN202311822261.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-12-27
- Publication Date
- 2025-06-27
AI Technical Summary
The existing handwriting recognition system has low recognition rate, slow input speed when writing continuously, and lacks brand effect, which cannot meet users' efficient and coherent writing needs.
The handwritten Chinese character image evaluation system based on deep learning is adopted. The Chinese character stroke sequence is extracted through the preparation processing module, the input step module is used to match the stroke order, the step correction module guides the user to write position, the secondary input module realizes quick and correct writing input, and the pen end fusion module integrates the pen segment to improve writing fluency and recognition rate.
It realizes fast and coherent writing by users, improves recognition rate and input speed, solves the problems of low recognition rate and slow input speed during continuous writing, and enhances market competitiveness through brand effect.
Smart Images
Figure HDA0004634330970000011 
Figure HDA0004634330970000021
Abstract
Description
Technical Field
[0001] The present invention relates to the field of Chinese character image evaluation, and specifically to a handwritten Chinese character image evaluation system based on deep learning. Background Art
[0002] With the continuous improvement of computer performance, human-computer interaction has become a major challenge in the Internet technology of the new era. Human-computer interaction is moving towards the goals of high efficiency, intelligence and standardization. At present, many intelligent interaction technologies such as handwritten input, voice input, and image recognition input have emerged. Among them, handwritten input does not require learning input method rules and does not require selecting candidate characters, and has the characteristics of low learning cost and fast input speed.
[0003] Most of the native handwritten recognition software on the market is for single-character input recognition. It is necessary to wait for the recognition of the previous character to complete before starting to write the next character, resulting in the disadvantages of long user input time and discontinuous writing. In recent years, the demand for continuous handwritten input has shown an increasing trend, while the supply side is slightly insufficient. In particular, there are not many enterprises with core intellectual property rights and excellent technologies, and the industry as a whole lacks brand effect. Summary of the Invention
[0004] The purpose of the present invention is to provide a handwritten Chinese character image evaluation system based on deep learning to solve the problems raised in the above background art.
[0005] To achieve the above purpose, the present invention provides the following technical solution: A handwritten Chinese character image evaluation system based on deep learning, including a preparation and processing module. The preparation and processing module performs an operation of extracting the stroke order of Chinese characters on the Chinese character image to be written, obtains a Chinese character skeleton image, and performs an operation of segmenting the Chinese character strokes to obtain an exploded image containing multiple strokes.
[0006] As a further optimization of this technical solution: The preparation and processing module is electrically connected to the input step module, the input step module is electrically connected to the step correction module, the step correction module is electrically connected to the secondary input module, and the secondary input module is electrically connected to the pen tip fusion module.
[0007] As a further optimization of this technical solution: The input step module uses the tree branch boundary method to perform a stroke order matching operation on the pen segments in the exploded image according to the strokes, forms a group of pen segments from the successfully matched pen segments, completes the first stage from the successfully matched pen segments to the strokes, and maps each pen segment to the successfully matched stroke order.
[0008] As a further optimization of this technical solution: The step correction module forms a failed pen segment set from the failed pen segments, corrects the strokes of the failed pen segment set, and then through the secondary input module, executes the following second stroke order.
[0009] As a further preference of this technical solution: The pen tip fusion module is used to merge the first stroke and the second stroke into a third stroke, form a set of pen segments to be fused for the same stroke order in the third stroke, perform pen segment fusion operations on each set of pen segments to be fused respectively, and fuse the pen segments in each set of pen segments to be fused;
[0010] As a further preference of this technical solution: The step correction module includes a stroke order calculation step, a handwriting recognition module, and a preprocessing module;
[0011] As a further preference of this technical solution: The preprocessing module performs an analysis operation on the Chinese character image being written, stores the incorrect stroke order, and waits for the handwriting recognition module to recognize it;
[0012] As a further preference of this technical solution: The handwriting recognition module mainly collects the position of the stylus, draws the corresponding stroke at the corresponding position on the screen. At the same time, during the user's writing process, according to the feedback of the stylus pressure, the written font will present different starting strokes and pen tips;
[0013] As a further preference of this technical solution: The stroke order calculation step completes the stroke order in sequence by the branch and bound method according to all the matching weights of each pen segment.
[0014] Compared with the prior art, the beneficial effects of the present invention are:
[0015] 1. In the present invention, through the step correction module to guide the user's writing position, the Chinese characters written by the user are more regular, which is convenient for the user and improves the recognition rate at the same time. The step correction module does not limit the user's writing range, effectively avoiding the problem that some handwriting is truncated, improving the writing fluency. Since the Chinese characters after guided writing are regular and orderly, through the preparation processing module, the segmentation operation of Chinese character strokes is pre-performed to obtain an exploded image containing multiple strokes, omitting the step of Chinese character cutting and not causing recognition errors due to incorrect cutting. The present invention can, due to the introduction of the secondary input module, allow the user to quickly perform secondary input of the correct writing method, so as to recognize the previous character when writing the next character, achieving the purpose of guiding the user to write quickly and continuously and pre-recognizing the written Chinese characters, and solving the problems of low recognition rate and slow input speed during continuous writing. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] Figure 1 is the flow of the handwritten Chinese character image evaluation system based on deep learning of the present invention Figure 1 ;
[0017] Figure 2Flow of the Handwritten Chinese Character Image Evaluation System Based on Deep Learning of the Present Invention Figure 2 。 Specific Embodiment
[0018] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.
[0019] Embodiment
[0020] Please refer to Figure 1 - Figure 2 As shown, the present invention provides a technical solution: a handwritten Chinese character image evaluation system based on deep learning, including a preparation processing module. The preparation processing module performs an operation of extracting the stroke order of Chinese characters on the Chinese character image to be written, obtains a Chinese character skeleton image, and performs an operation of segmenting the Chinese character strokes to obtain an explosion image containing multiple strokes.
[0021] In this embodiment, specifically: the preparation processing module is electrically connected to the input step module, the input step module is electrically connected to the step correction module, the step correction module is electrically connected to the secondary input module, and the secondary input module is electrically connected to the pen end fusion module.
[0022] In this embodiment, specifically: the input step module performs a stroke order matching operation on the pen segments in the explosion image according to the strokes by the tree branch boundary method, forms a group of pen segments from the successfully matched pen segments, completes the first stage from the successfully matched pen segments to the strokes, maps each pen segment to the successfully matched stroke order. The step correction module guides the user's writing position, making the Chinese characters written by the user more regular, which is convenient for the user and improves the recognition rate at the same time. The step correction module does not limit the user's writing range.
[0023] In this embodiment, specifically: the step correction module forms a failed pen segment set from the failed pen segments, corrects the strokes of the failed pen segment set, and then through the secondary input module, executes the following second stroke order. The step correction module guides the user's writing position, making the Chinese characters written by the user more regular, which is convenient for the user and improves the recognition rate at the same time. The step correction module does not limit the user's writing range.
[0024] In this embodiment, specifically: the pen end fusion module is used to merge the first stroke and the second stroke into the third stroke, form a set of pen segments to be fused for the same stroke order in the third stroke from two or more pen segments mapped to the same stroke order, perform a pen segment fusion operation on each set of pen segments to be fused respectively, and fuse the pen segments in each set of pen segments to be fused.
[0025] In this embodiment, specifically, the step correction module includes a stroke order calculation step, a handwriting recognition module, and a preprocessing module.
[0026] In this embodiment, specifically, the preprocessing module performs an analysis operation on the stroke order of the Chinese character image being written, stores the incorrect stroke order, and waits for the handwriting recognition module to identify it.
[0027] In this embodiment, specifically, the handwriting recognition module mainly collects the position of the stylus, draws the corresponding strokes at the corresponding positions on the screen. At the same time, during the user's writing process, according to the feedback of the stylus pressure, the written font will present different starting strokes and pen tips.
[0028] In this embodiment, specifically, the stroke order calculation step completes the stroke order in sequence through the branch and bound method according to the matching weight of each pen segment.
[0029] Working principle or structural principle: First, the preparation processing module performs an operation to extract the stroke order of the Chinese character image to be written, obtains the Chinese character skeleton image, performs a segmentation operation on the Chinese character strokes to obtain an exploded image containing multiple strokes. Then, the input step module uses the branch and bound method to perform a stroke order matching operation on the pen segments in the exploded image according to the strokes, forms a group of pen segments from the successfully matched pen segments, and completes the first stage from the successful pen segments to the strokes. Maps each pen segment to the successfully matched stroke order. When the user writes incorrectly, the step correction module forms a set of failed pen segments from the unmatched pen segments, stores the incorrect stroke order, and waits for the handwriting recognition module to identify it. Then, through the stroke order calculation step, according to all the matching weights of each pen segment, the stroke order is completed in sequence through the branch and bound method, so that the set of failed pen segments can be corrected for strokes. Then, through the secondary input module, the following second stroke order is executed and finally presented on the display screen for the user to learn.
[0030] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.
Claims
1. A handwritten Chinese character image evaluation system based on deep learning, including a preparation and processing module, characterized in that: The preparation processing module performs a Chinese character stroke order extraction operation on the Chinese character image to be written, obtains a Chinese character skeleton image, and performs a segmentation operation on the Chinese character strokes to obtain an exploded image containing multiple strokes.
2. The handwritten Chinese character image evaluation system based on deep learning according to claim 1, characterized in that: The preparation processing module is electrically connected to the input step module, the input step module is electrically connected to the step correction module, the step correction module is electrically connected to the secondary input module, and the secondary input module is electrically connected to the pen tip fusion module.
3. The handwritten Chinese character image evaluation system based on deep learning according to claim 2, characterized in that: The input step module performs a stroke order matching operation on the pen segments in the exploded image according to the strokes by the tree branch boundary method, forms a set of pen segments from the successfully matched pen segments, completes the first stage from the successfully matched pen segments to the strokes, and maps each pen segment to the successfully matched stroke order.
4. The handwritten Chinese character image evaluation system based on deep learning according to claim 2, wherein: The step correction module forms a set of failed pen segments from the failed pen segments, corrects the strokes of the set of failed pen segments, and then executes the following second stroke order through the secondary input module.
5. The handwritten Chinese character image evaluation system based on deep learning according to claim 2, characterized in that: The pen tip fusion module is used to merge the first stroke and the second stroke into a third stroke, form a set of pen segments to be fused for the stroke order in which more than 2 pen segments in the third stroke are mapped to the same stroke order, perform a pen segment fusion operation on each set of pen segments to be fused, and fuse the pen segments in each set of pen segments to be fused.
6. The handwritten Chinese character image evaluation system based on deep learning according to claim 1, wherein: The step correction module includes a stroke order calculation step, a handwriting recognition module, and a preprocessing module.
7. The handwritten Chinese character image evaluation system based on deep learning according to claim 6, characterized in that: The preprocessing module performs a Chinese character stroke order analysis operation on the Chinese character image being written, stores the incorrect stroke order, and waits for the handwriting recognition module to identify it.
8. The handwritten Chinese character image evaluation system based on deep learning according to claim 6, characterized in that: The handwriting recognition module mainly collects the position of the stylus, draws the corresponding strokes at the corresponding positions on the screen. At the same time, during the writing process of the user, according to the feedback of the stylus pressure, the written font will present different starting strokes and pen tips.
9. The handwritten Chinese character image evaluation system based on deep learning according to claim 6, wherein: The stroke order calculation step sequentially completes the stroke order by the tree branch boundary method according to all the matching weights of each pen segment.