CN116129446A - Handwritten Chinese character recognition method based on deep learning - Google Patents
Handwritten Chinese character recognition method based on deep learning Download PDFInfo
- Publication number
- CN116129446A CN116129446A CN202310097533.5A CN202310097533A CN116129446A CN 116129446 A CN116129446 A CN 116129446A CN 202310097533 A CN202310097533 A CN 202310097533A CN 116129446 A CN116129446 A CN 116129446A
- Authority
- CN
- China
- Prior art keywords
- stroke
- strokes
- handwritten chinese
- recognition
- handwritten
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
- 238000000034 method Methods 0.000 title claims abstract description 23
- 238000013135 deep learning Methods 0.000 title claims abstract description 19
- 238000012549 training Methods 0.000 claims abstract description 48
- 238000013527 convolutional neural network Methods 0.000 claims description 15
- 238000012163 sequencing technique Methods 0.000 claims description 9
- 238000002372 labelling Methods 0.000 claims description 7
- 238000013528 artificial neural network Methods 0.000 description 5
- 238000012986 modification Methods 0.000 description 5
- 230000004048 modification Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 230000003321 amplification Effects 0.000 description 1
- 235000021167 banquet Nutrition 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000012552 review Methods 0.000 description 1
- 238000011282 treatment Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/22—Character recognition characterised by the type of writing
- G06V30/226—Character recognition characterised by the type of writing of cursive writing
- G06V30/2268—Character recognition characterised by the type of writing of cursive writing using stroke segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/19—Recognition using electronic means
- G06V30/191—Design or setup of recognition systems or techniques; Extraction of features in feature space; Clustering techniques; Blind source separation
- G06V30/19147—Obtaining sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Abstract
The application discloses a handwritten Chinese font recognition method based on deep learning, which comprises the following steps: establishing a stroke recognition model; training the stroke recognition model by adopting a training set; performing stroke recognition on the handwritten Chinese to be recognized by using the trained stroke recognition model to obtain corresponding strokes and stroke sequences; comparing the recognized strokes with the strokes of the printed characters in the comparison library to determine matched candidate printed characters; and comparing the stroke sequence obtained by recognition with the stroke sequence of the candidate printed text, and determining a final recognition result. The handwriting Chinese character recognition method based on the strokes and the stroke sequences can recognize handwriting Chinese characters through the strokes and the stroke sequences, is high in recognition speed when the character complexity is low, can adapt to handwriting habits of different people, and improves recognition accuracy.
Description
Technical Field
The application relates to the technical field of image recognition, in particular to a handwritten Chinese font recognition method based on deep learning.
Background
Handwritten Chinese characters are contacted by most domestic personnel every day, and workers in many related industries need to sign files, record information and the like in a manner of handwritten Chinese characters, and paper files are usually converted into electronic files in a manner of scanning after handwriting is finished, so that the electronic files are convenient to store and review.
Generally, an electronic file needs to identify its content to facilitate copying, for example, after meeting content is recorded by handwriting, it needs to be converted into an electronic file as soon as possible, and handwritten Chinese characters in the electronic file need to be identified to be converted into correct printed characters, so as to facilitate reading and understanding of more people, and therefore, an identification technology of handwritten Chinese characters needs to be adopted. At present, many recognition technologies of handwritten Chinese characters adopt a neural network, after training the neural network through a training set with a large number of handwritten Chinese characters, the trained neural network is adopted to recognize the handwritten Chinese characters to be recognized, and the printed characters corresponding to the handwritten Chinese characters, such as CN114581922A and CN111652332A, are determined. However, these existing recognition technologies are all used for integrally recognizing handwritten Chinese characters and pushing results according to the similarity probability of recognition, and as the handwritten Chinese characters are extremely different when written by different people, different handwritten Chinese characters of different people may have higher similarity, so that the handwritten Chinese characters are easily and mistakenly recognized, and the recognition accuracy is not high enough.
Disclosure of Invention
The embodiment of the application provides a handwritten Chinese font recognition method based on deep learning, which is used for solving the problem that the recognition accuracy is not high enough in the mode of integrally recognizing handwritten Chinese characters in the prior art.
In one aspect, an embodiment of the present application provides a method for recognizing a handwritten chinese font based on deep learning, including:
establishing a stroke recognition model;
training the stroke recognition model by adopting a training set;
performing stroke recognition on the handwritten Chinese to be recognized by using the trained stroke recognition model to obtain corresponding strokes and stroke sequences;
comparing the recognized strokes with the strokes of the printed characters in the comparison library to determine matched candidate printed characters;
and comparing the stroke sequence obtained by recognition with the stroke sequence of the candidate printed text, and determining a final recognition result.
In one possible implementation, the stroke recognition model employs a convolutional neural network, and training the stroke recognition model using a training set includes: acquiring a handwritten Chinese training image; marking the printed characters corresponding to the handwritten Chinese training images; inputting the handwritten Chinese training image with the labels into a convolutional neural network, and outputting a corresponding training recognition result; and adjusting parameters of the convolutional neural network according to errors of the labeling and the training recognition results, inputting the handwritten Chinese training image with the labeling into the adjusted convolutional neural network again until the set training times are reached, and obtaining a stroke recognition model after training.
In one possible implementation, the handwritten chinese training image includes a background image and foreground handwritten text superimposed on the background image, and obtaining the handwritten chinese training image includes: acquiring a plurality of groups of handwritten Chinese images and a plurality of background images; extracting handwritten Chinese in the handwritten Chinese image to obtain foreground handwritten characters; and combining the foreground handwritten characters with the background image, and superposing the foreground handwritten characters in the combination in the background image to obtain the handwritten Chinese training image.
In one possible implementation manner, performing stroke recognition on the handwritten Chinese to be recognized by using the trained stroke recognition model to obtain corresponding strokes and stroke sequences, including: splitting independent areas in handwriting to obtain a plurality of stroke areas; splitting each stroke area according to the stroke trend to obtain a plurality of strokes; recording the position relation of each stroke; and sequencing the strokes according to the set direction and the position relation of each stroke to obtain the stroke sequence.
In one possible implementation, splitting independent regions in handwriting to obtain a plurality of stroke regions includes: analyzing the space structure of the handwritten Chinese, determining whether the handwritten Chinese can be divided into two or more main stroke areas, if so, dividing the handwritten Chinese into two or more main stroke areas, and recording the position relation of each main stroke area; the spatial structure of each main stroke area is analyzed to determine whether the main stroke area can be divided into two or more sub-stroke areas, if so, the main stroke area is divided into two or more sub-stroke areas, and the position relation of each sub-stroke area is recorded.
In one possible implementation, after splitting each stroke area to obtain a plurality of strokes, a plurality of strokes with connection relations are combined according to the stroke requirement, whether the combined strokes are basic strokes is determined, and if so, the combined strokes are reserved.
In one possible implementation, the identified strokes include a number of each stroke and a number of each stroke, the printed text in the comparison library also includes the number of each stroke and the number of each stroke, the identified strokes are compared with the printed text strokes in the comparison library to determine matched candidate printed text, including: comparing the total number of strokes with the total number of strokes of the printed characters in the comparison library, and reserving the printed characters with the same total number of strokes as the total number of strokes in the comparison library; and comparing the serial numbers of the reserved printed characters with the serial numbers of the strokes, and reserving the printed characters with the serial numbers of the strokes matched with the serial numbers of the strokes as candidate printed characters.
In one possible implementation, comparing the recognized stroke order with the stroke order of the candidate printed text to determine a final recognition result includes: sequencing the numbers of the strokes according to the stroke sequence to obtain the stroke number sequence; sequencing the stroke numbers of the candidate printed characters according to the sequence to obtain the serial numbers of the printed characters; and comparing the stroke number sequence with the printing character number sequence, and reserving candidate printing characters with the printing character number sequence matched with the stroke number sequence as a recognition result.
The handwritten Chinese font recognition method based on deep learning in the application has the following advantages:
the handwriting Chinese is identified through the strokes and the stroke sequences, the identification speed is high when the character complexity is low, and the identification method based on the strokes and the stroke sequences can adapt to handwriting habits of different people, so that the identification accuracy is improved.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present application, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of a method for recognizing a handwritten chinese font based on deep learning according to an embodiment of the present application.
Detailed Description
The following description of the embodiments of the present application will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are only some, but not all, of the embodiments of the present application. All other embodiments, which can be made by one of ordinary skill in the art without undue burden from the present disclosure, are within the scope of the present disclosure.
Fig. 1 is a flowchart of a method for recognizing a handwritten chinese font based on deep learning according to an embodiment of the present application. The embodiment of the application provides a handwritten Chinese font recognition method based on deep learning, which comprises the following steps:
s100, establishing a stroke recognition model.
Illustratively, the stroke recognition model employs a convolutional neural network, CNN (Convolutional Neural Network) for short, which is widely used in computer vision processing. The neural network includes an input layer, a convolution layer, a pooling layer, a full connection layer, and an output layer, which require initialization of its various parameters prior to use of the neural network. The convolutional neural network is adopted to perform deep learning on various handwritten Chinese characters, so that the accuracy of the recognition result is effectively improved.
S110, training the stroke recognition model by adopting a training set.
Illustratively, S110 specifically includes: acquiring a handwritten Chinese training image; marking the printed characters corresponding to the handwritten Chinese training images; inputting the handwritten Chinese training image with the labels into a convolutional neural network, and outputting a corresponding training recognition result; and adjusting parameters of the convolutional neural network according to errors of the labeling and the training recognition results, inputting the handwritten Chinese training image with the labeling into the adjusted convolutional neural network again until the set training times are reached, and obtaining a stroke recognition model after training.
In an embodiment of the present application, the handwritten chinese training image includes a background image and foreground handwritten text superimposed on the background image, and obtaining the handwritten chinese training image includes: acquiring a plurality of groups of handwritten Chinese images and a plurality of background images; extracting handwritten Chinese in the handwritten Chinese image to obtain foreground handwritten characters; and combining the foreground handwritten characters with the background image, and superposing the foreground handwritten characters in the combination in the background image to obtain the handwritten Chinese training image.
Specifically, the handwritten Chinese image and the background image can be crawled on the network by adopting a crawler program, and when the foreground handwritten characters and the background image are combined, the combination can be performed in a set sequence, and a random combination mode can also be adopted. When the foreground handwritten characters are superimposed in the background image, the foreground handwritten characters can be subjected to angle rotation, amplification, reduction and other treatments, so that the richness of the handwritten Chinese training image is improved.
And S120, performing stroke recognition on the handwritten Chinese to be recognized by using the trained stroke recognition model to obtain corresponding strokes and stroke sequences.
Illustratively, S120 specifically includes: splitting independent areas in handwriting to obtain a plurality of stroke areas; splitting each stroke area according to the stroke trend to obtain a plurality of strokes; recording the position relation of each stroke; and sequencing the strokes according to the set direction and the position relation of each stroke to obtain the stroke sequence.
In the splitting process of the handwritten Chinese, the space structure of the handwritten Chinese is analyzed to determine whether the handwritten Chinese can be divided into two or more main stroke areas, if so, the handwritten Chinese is divided into two or more main stroke areas, and the position relation of each main stroke area is recorded; the spatial structure of each main stroke area is analyzed to determine whether the main stroke area can be divided into two or more sub-stroke areas, if so, the main stroke area is divided into two or more sub-stroke areas, and the position relation of each sub-stroke area is recorded. For example, the "character" may be split into three main stroke areas, namely "", "dog" and "", respectively, and the upper and lower "" areas may be split into two sub-stroke areas, namely "mouth", respectively, and the sub-stroke areas obtained by splitting may be split according to the trend of the strokes, so as to obtain corresponding strokes.
For handwritten Chinese characters which only have an upper-lower structure, an upper-middle-lower structure, a left-right structure or a left-middle-right structure, such as 'put', 'banquet', 'position', 'tree', and the like, only two or three main stroke areas exist in the Chinese characters, and no sub-stroke area exists, so that only the main stroke area needs to be divided, and the divided main stroke area is split into strokes. For individual words, such as "country", "large", etc., there is neither a main stroke area nor a sub-stroke area, so that splitting of strokes can be directly performed without splitting of stroke areas.
In the embodiment of the application, the strokes are ordered from top to bottom and from left to right, which accords with the general writing habit. Specifically, for handwritten Chinese having two upper and lower or three upper, middle and lower main stroke areas, strokes are ordered in the up-to-down direction, and for handwritten Chinese having two left, right or three left, middle and right main stroke areas, strokes are ordered in the left-to-right direction. If the split main stroke area also has sub-stroke areas, the strokes are ordered in corresponding directions according to the structure of the sub-stroke areas.
In the process of splitting the stroke area, the positional relationship of each stroke area including the main stroke area and the sub-stroke area needs to be recorded, for example, the character comprises three main stroke areas from top to bottom, and the three main stroke areas are arranged in the sequence from top to bottom: after the positional relationship of the stroke areas is recorded, the split strokes can be ordered according to the positional relationship, so that the correct stroke sequence is obtained.
Further, after splitting each stroke area to obtain a plurality of strokes, combining the plurality of strokes with connection relation according to the stroke requirement to determine whether the combined strokes are basic strokes, and if so, reserving the combined strokes.
Since some base strokes require multiple strokes, for example, a "Chinese" word contains two base strokes, a "Chinese" and a "Chinese" respectively
Wherein->
The basic strokes which are easy to split into the combination of the folding and the hook are needed to be combined according to the connection relation, if the combined strokes are still basic strokes, excessive splitting is indicated, and therefore the combined basic strokes are needed to be reserved.
S130, comparing the recognized strokes with the strokes of the printed characters in the comparison library to determine matched candidate printed characters.
Illustratively, the identified strokes include a number of each stroke and a number of each stroke, and the printed text in the comparison library also includes a number of each stroke and a number of each stroke. And S130 specifically includes: comparing the total number of strokes with the total number of strokes of the printed characters in the comparison library, and reserving the printed characters with the same total number of strokes as the total number of strokes in the comparison library; and comparing the serial numbers of the reserved printed characters with the serial numbers of the strokes, and reserving the printed characters with the serial numbers of the strokes matched with the serial numbers of the strokes as candidate printed characters.
The comparison library stores the printing plate characters of the common Chinese characters, the fonts of the printing plate characters can be selected by a user, and the comparison library not only needs to store the printing characters, but also needs to record the related information of each printing character, including the stroke number, the stroke sequence and the like of each printing character.
In the embodiment of the application, the basic strokes of the Chinese character are more, and if the stroke sequence of the basic strokes is described according to the basic strokes, more information needs to be described. In order to reduce the data size, the application adopts Arabic numerals to represent each basic stroke, for example, 1, 2, 3 and 4 … respectively represent the horizontal, vertical, left-falling and right-falling … in the basic strokes, so that Arabic numerical codes corresponding to the basic stroke sequence can be used for representing each Chinese character, including the stroke sequence of handwriting Chinese and printing characters, and the data size is effectively reduced.
And S140, comparing the stroke sequence obtained by recognition with the stroke sequence of the candidate printed text, and determining a final recognition result.
Illustratively, S140 specifically includes: sequencing the numbers of the strokes according to the stroke sequence to obtain the stroke number sequence; sequencing the stroke numbers of the candidate printed characters according to the sequence to obtain the serial numbers of the printed characters; and comparing the stroke number sequence with the printing character number sequence, and reserving candidate printing characters with the printing character number sequence matched with the stroke number sequence as a recognition result.
The method and the device can identify the handwritten characters from two aspects of strokes and stroke sequences, but because different Chinese characters with the same stroke and stroke sequences exist, such as 'heaven' and 'Fu', if the strokes and the stroke sequences are only relied on, a plurality of identification results can be generated, and the identification accuracy is seriously influenced, so that after the identification results are obtained, the identification results are corrected by adopting the recorded position relation of each stroke, for example, when the handwritten Chinese character is 'heaven', the first stroke 'one' is positioned above the third stroke 'and the first stroke' one 'and the third stroke' of the 'Fu' are crossed, the 'Fu' character in the identification results can be eliminated, and the accuracy of the identification results is improved.
While preferred embodiments of the present application have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. It is therefore intended that the following claims be interpreted as including the preferred embodiments and all such alterations and modifications as fall within the scope of the application.
It will be apparent to those skilled in the art that various modifications and variations can be made in the present application without departing from the spirit or scope of the application. Thus, if such modifications and variations of the present application fall within the scope of the claims and the equivalents thereof, the present application is intended to cover such modifications and variations.
Claims (8)
1. The handwritten Chinese character recognition method based on deep learning is characterized by comprising the following steps of:
establishing a stroke recognition model;
training the stroke recognition model by adopting a training set;
performing stroke recognition on the handwritten Chinese to be recognized by using the trained stroke recognition model to obtain corresponding strokes and stroke sequences;
comparing the recognized strokes with the strokes of the printed characters in the comparison library to determine matched candidate printed characters;
and comparing the stroke sequence obtained by recognition with the stroke sequence of the candidate printed text, and determining a final recognition result.
2. The deep learning based handwritten chinese font recognition method of claim 1, wherein the stroke recognition model employs a convolutional neural network, and wherein training the stroke recognition model using a training set comprises:
acquiring a handwritten Chinese training image;
labeling the printed characters corresponding to the handwritten Chinese training images;
inputting the handwritten Chinese training image with the labels into the convolutional neural network, and outputting a corresponding training recognition result;
and adjusting parameters of the convolutional neural network according to errors of the labeling and training recognition results, and inputting the handwritten Chinese training image with the labeling into the adjusted convolutional neural network again until the set training times are reached, so as to obtain a stroke recognition model after training.
3. The deep learning based handwritten chinese font recognition method of claim 2, wherein the handwritten chinese training image comprises a background image and foreground handwritten text superimposed on the background image, the obtaining the handwritten chinese training image comprising:
acquiring a plurality of groups of handwritten Chinese images and a plurality of background images;
extracting the handwritten Chinese in the handwritten Chinese image to obtain the foreground handwritten characters;
and combining the foreground handwritten characters with the background image, and superposing the foreground handwritten characters in the combination in the background image to obtain the handwritten Chinese training image.
4. The deep learning-based handwritten chinese font recognition method of claim 1, wherein the performing stroke recognition on the handwritten chinese to be recognized using the trained stroke recognition model to obtain corresponding strokes and stroke sequences comprises:
splitting independent areas in the handwriting to obtain a plurality of stroke areas;
splitting each stroke area according to the stroke trend to obtain a plurality of strokes;
recording the position relation of each stroke;
and sequencing the strokes according to the set direction and the position relation of each stroke to obtain the stroke sequence.
5. The method for recognition of deep learning based handwritten chinese fonts of claim 4, wherein said splitting individual regions in said handwriting to obtain a plurality of stroke regions comprises:
analyzing the space structure of the handwritten Chinese, determining whether the handwritten Chinese can be divided into two or more main stroke areas, if so, dividing the handwritten Chinese into two or more main stroke areas, and recording the position relation of each main stroke area;
and analyzing the space structure of each main stroke area to determine whether the main stroke area can be divided into two or more sub-stroke areas, if so, dividing the main stroke area into two or more sub-stroke areas, and recording the position relation of each sub-stroke area.
6. The deep learning based handwritten chinese font recognition method of claim 4, wherein after splitting each of the stroke regions to obtain a plurality of strokes, a plurality of strokes having a connection relationship are further combined according to a stroke requirement, and whether the combined strokes are basic strokes is determined, and if so, the combined strokes are reserved.
7. The deep learning based handwritten chinese font recognition method of claim 1, wherein the recognized strokes include a number of each stroke and a number of each stroke, the printed text in the comparison library also includes a number of each stroke and a number of each stroke, the comparing the recognized strokes with the printed text strokes in the comparison library to determine matched candidate printed text, comprising:
comparing the total number of strokes with the total number of strokes of the printed characters in the comparison library, and reserving the printed characters with the total number of strokes being the same as the total number of strokes in the comparison library;
and comparing the reserved stroke numbers of the printed characters with the stroke numbers, and reserving the printed characters with the stroke numbers matched with the stroke numbers as the candidate printed characters.
8. The deep learning based handwritten chinese font recognition method of claim 7, wherein comparing the recognized stroke order with the stroke order of the candidate printed text to determine a final recognition result comprises:
sequencing the numbers of the strokes according to the stroke sequence to obtain a stroke number sequence;
sequencing the stroke numbers of the candidate printed characters according to the sequence to obtain the serial numbers of the printed characters;
and comparing the stroke number sequence with the printing text number sequence, and reserving the candidate printing text with the printing text number sequence matched with the stroke number sequence as the identification result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310097533.5A CN116129446A (en) | 2023-02-10 | 2023-02-10 | Handwritten Chinese character recognition method based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310097533.5A CN116129446A (en) | 2023-02-10 | 2023-02-10 | Handwritten Chinese character recognition method based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116129446A true CN116129446A (en) | 2023-05-16 |
Family
ID=86297101
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310097533.5A Withdrawn CN116129446A (en) | 2023-02-10 | 2023-02-10 | Handwritten Chinese character recognition method based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116129446A (en) |
-
2023
- 2023-02-10 CN CN202310097533.5A patent/CN116129446A/en not_active Withdrawn
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US7742642B2 (en) | System and method for automated reading of handwriting | |
| US7302099B2 (en) | Stroke segmentation for template-based cursive handwriting recognition | |
| Naz et al. | The optical character recognition of Urdu-like cursive scripts | |
| Nakagawa et al. | Collection of on-line handwritten Japanese character pattern databases and their analyses | |
| US7885464B2 (en) | Apparatus, method, and program for handwriting recognition | |
| US7860313B2 (en) | Methods and apparatuses for extending dynamic handwriting recognition to recognize static handwritten and machine generated text | |
| JP3452774B2 (en) | Character recognition method | |
| JP2713622B2 (en) | Tabular document reader | |
| US7929769B2 (en) | Script recognition for ink notes | |
| CN1099500A (en) | Estimation of baseline, line spacing and character height for handwriting recognition | |
| JP3294995B2 (en) | Form reader | |
| CN1324068A (en) | Explanatory and search for handwriting sloppy Chinese characters based on shape of radicals | |
| Bai et al. | Keyword spotting in document images through word shape coding | |
| Dongre et al. | Development of comprehensive devnagari numeral and character database for offline handwritten character recognition | |
| JPWO2007094078A1 (en) | Character string search method and apparatus | |
| Pechwitz et al. | Handwritten Arabic word recognition using the IFN/ENIT-database | |
| US10586133B2 (en) | System and method for processing character images and transforming font within a document | |
| Suresh et al. | Telugu Optical Character Recognition Using Deep Learning | |
| WO2019185245A2 (en) | An image processing system and an image processing method | |
| Marinai | Text retrieval from early printed books | |
| CN111259888A (en) | Image-based information comparison method and device and computer-readable storage medium | |
| CN116129446A (en) | Handwritten Chinese character recognition method based on deep learning | |
| CN109522892A (en) | The character image information labeling method of neural network aiding | |
| Kumar et al. | Line based robust script identification for indianlanguages | |
| CN114241486A (en) | Method for improving accuracy rate of identifying student information of test paper |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20230516 |