CN110555403A - handwritten character evaluation method and system - Google Patents
handwritten character evaluation method and system Download PDFInfo
- Publication number
- CN110555403A CN110555403A CN201910805073.0A CN201910805073A CN110555403A CN 110555403 A CN110555403 A CN 110555403A CN 201910805073 A CN201910805073 A CN 201910805073A CN 110555403 A CN110555403 A CN 110555403A
- Authority
- CN
- China
- Prior art keywords
- stroke
- sequence
- character
- distance
- correct
- 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- Withdrawn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
Classifications
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of 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
- G06V30/32—Digital ink
-
- 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
- G06V30/32—Digital ink
- G06V30/333—Preprocessing; Feature extraction
- G06V30/347—Sampling; Contour coding; Stroke extraction
-
- 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
- G06V30/32—Digital ink
- G06V30/36—Matching; Classification
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Character Discrimination (AREA)
Abstract
The invention provides a handwritten character evaluation method and device. Recognizing the handwriting, and judging that the handwriting is irregular when the handwriting cannot be recognized or the confidence coefficient is smaller than a preset threshold value; carrying out stroke recognition on the handwritten character; comparing the character stroke sequence with the correct stroke sequence, and if the sequences are different, judging that the strokes are wrong; identifying the distance between the components; labeling irregular characters; the stroke order is corrected. The invention can identify whether the handwritten character is regularly written or not, identify whether the sequence of the written strokes is correct or not, highlight the written strokes and provide an error correction suggestion. The teacher can correct the errors of the students or the writers who find and correct the handwriting by themselves.
Description
Technical Field
The invention relates to the technical field of computer application, in particular to a handwritten character evaluation method and device.
Background
in the intelligent education of the scientific and technological era, teachers often make students use computers, ipads, handwriting boards and the like for teaching. The novel children learning device can be separated from the space-time limit, and teachers and children can learn in different spaces and time.
When the user mentions the job, the teacher does not just see if the child completed on time, but the word was written incorrectly. And also judge whether the strokes written by the children are regular and correct. The criteria that the teacher judges are basically only three dimensions, first, see first whether the word is wrongly written or not, and second see if the components of each word, i.e. radicals, etc., can not be merged together, whether far apart or too compact. Thirdly, it is determined whether the stroke order of writing each word by the child is correct. When a child writes many characters, the judgment of whether one character is correct and standard requires a relatively long time, and especially when a large number of students write simultaneously, the work of teachers is overwhelmed. Thus requiring the computer to make a quick judgment of the word. And meanwhile, a system prompt can be given to the wrong words, so that a teacher and a student can understand whether the mistakes occur in the writing process and where the mistakes are. This completes the evaluation.
the invention judges the normalization of the character through character recognition. And identifying the stroke sequence in the font to judge whether the children are standard in the writing process. The teacher can correct the wrongly written characters and the strokes with errors in time conveniently.
disclosure of Invention
The invention provides a handwritten character evaluation method, which mainly comprises the following steps:
Recognizing the handwriting, and judging that the handwriting is irregular when the handwriting cannot be recognized or the confidence coefficient is smaller than a preset threshold value;
carrying out stroke recognition on the handwritten character;
Comparing the character stroke sequence with the correct stroke sequence, and if the sequences are different, judging that the strokes are wrong;
identifying the distance between the components according to the strokes, and judging whether the font is standard or not according to the depth learning model prediction result of the font component distance;
And marking irregular characters, correcting stroke sequence and correcting irregular characters.
further optionally, as in the foregoing method, the recognizing a written word, and when the recognition is not successful or the confidence is smaller than a preset threshold, determining that the written word is irregular, mainly includes:
adopting a deep learning tool tensorflow to perform handwriting recognition, wherein a convolutional neural network is adopted in a deep learning handwritten character recognition algorithm, a first preset number of convolutional kernels are arranged on a first layer, the size of each convolutional kernel is 6 x 6, and the convolutional kernels are connected with an input layer in a full-connection mode to perform convolution to obtain a first layer of characteristic image; setting a second preset number of pooling on a second layer, wherein the size of a pooling kernel is 4 x 4, and pooling the first layer of feature images to obtain second layer of feature images; converting the second layer characteristic image by using a modified linear unit function at a third layer to obtain a third layer characteristic image; a fourth layer is provided with a third preset number of convolution kernels, each convolution kernel is 6 x 6 in size, and the fourth layer is connected with the input layer in a full-connection mode for convolution to obtain a fourth layer characteristic image; converting the fourth layer characteristic image by using a modified linear unit function to obtain a fifth layer characteristic image; setting a fourth preset number of pooling kernels on a sixth layer, wherein the size of each pooling kernel is 3 x 3, and pooling the fifth layer feature image to obtain a sixth layer feature image;
extracting sample features in a one-dimensional vector form from a prediction sample, converting the sample features in the one-dimensional vector form into sample features in a two-dimensional array form, and generalizing the sample features in the two-dimensional array by taking the two-dimensional array as a unit based on the deep learning convolutional neural network model to obtain a processing result; and outputs the learning prediction result and outputs the correct probability, i.e. confidence, of the result. A confidence threshold for the recognized word is preset. And when the handwriting can not be recognized by the handwriting recognition algorithm or the recognition confidence coefficient is smaller than a preset threshold value, judging that the character is written irregularly.
Further optionally, as in the method described above, the performing stroke recognition on the handwritten word mainly includes:
According to the authoritative classification method of basic strokes of Chinese characters in a stroke summary table of Chinese characters in encyclopedic, the method is divided into 32 strokes in total, stroke analysis is carried out according to the stroke direction and/or the inflection point of the strokes, when each character is written, stroke cutting is carried out according to the stroke falling time and the stroke starting time of each stroke, each character is cut into stroke sequences, and the stroke sequences are recorded; training a stroke classification model according to the stroke characteristics;
classifying each stroke according to the stroke classification model to obtain a stroke identification result;
Recognizing stroke sequences of the whole character and combining the stroke sequences into a stroke recognition sequence of the character;
The model classifies the stroke sequences to obtain the stroke recognition result of the character, and the method further comprises the following steps:
only stroke recognition is carried out on the characters which are regularly written;
further optionally, as in the method above, the comparing the word stroke order with the correct stroke order, and if the word stroke order is different, determining that the word stroke order is incorrect mainly includes:
obtaining a correct stroke sequence through a stroke query interface or through a dictionary;
the method for acquiring the correct stroke comprises the steps of inputting the stroke of a target word in a stroke query interface on the network, wherein the interface can convert the word into a correct sequence containing the correct stroke sequence; and comparing the sequence of the stroke sequence of the character with the sequence of the correct stroke sequence, and judging whether the sequence of the character completely conforms to the correct sequence of the character.
If the correct stroke order is identical to the word stroke recognition sequence, determining that the strokes are correct;
otherwise, the stroke is judged to be wrongly written.
Further optionally, in the method as described above, the performing distance identification between the components mainly includes:
the first training model is trained to read the character sequence, and splits the character sequence word by word into radical sequences corresponding to the character sequence; a second training model trained to read the radical sequence output by the first training model and map the radical sequence to a first input sequence corresponding to the radical sequence; the third training model is used for acquiring the distance between the radicals of each character in the character sequence, including the distance between the left radical and the right radical and the distance between the upper part and the lower part in the upper and lower structures, and mapping the distance between the radicals to a second input sequence corresponding to the distance between the radicals;
the neural network prediction model is configured to receive the first input sequence and the second input sequence, train by adopting a seq2seq end-to-end training model and obtain an output sequence, wherein the output sequence comprises the components of the character sequence and the distance values between the components.
further optionally, as in the foregoing method, the determining whether the font is normal according to the deep learning prediction result of the font component distance mainly includes:
And inputting the remote distance of the conventional printing font into the radical neural network prediction model to obtain a distance value between the radical sequence and the radical. Inputting the radical neural network prediction model by acquiring the handwritten word to be predicted to obtain a radical distance value of the handwritten word;
judging whether the distance value of the radical of the handwritten word is larger than or smaller than the correct distance value of the printed word or not and whether the distance value exceeds a preset distance threshold or not;
And if the preset threshold value is exceeded, determining that the font components are not in specification.
further optionally, in the method as described above, the labeling of irregular words mainly includes:
marking the characters with irregular writing and the characters with irregular components by changing the font color or the font size or underlining;
further optionally, as in the method described above, the correcting the stroke order mainly includes:
marking the stroke errors by adopting a color different from that of the irregular characters, and providing a correct stroke sequence within a preset area range of the stroke error characters;
The invention provides a handwritten character evaluation device, comprising:
the character regularity judging module is used for judging whether the characters are regular or not;
The stroke classifier training module is used for training a stroke classifier and identifying strokes written by a user;
the stroke sequence judging module is used for recording the stroke sequence written by the user and comparing the stroke sequence with the correct stroke sequence;
The component distance judging module is used for identifying the written components of the user and judging whether the distance between the components is proper;
the marking and error correcting module is used for marking and highlighting the irregular characters and highlighting the characters with incorrect strokes and performing error correction prompt on the correct strokes;
The technical scheme provided by the embodiment of the invention has the following beneficial effects:
the invention can identify whether the handwritten character is regularly written or not, identify whether the sequence of the written strokes is correct or not, highlight the written strokes and provide an error correction suggestion. The teacher or writer can conveniently find and correct the hand writing errors.
Drawings
FIG. 1 is a flow chart of an embodiment of a handwritten word evaluation method of the present invention;
Fig. 2 is a block diagram of an embodiment of a handwritten character evaluation apparatus according to the present invention.
Detailed Description
in order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in detail with reference to the accompanying drawings and specific embodiments.
Fig. 1 is a flowchart of a handwritten character evaluation method of the present invention. As shown in fig. 1, the handwritten character evaluation method in this embodiment may specifically include the following steps:
step 101, training a stroke classifier. The Chinese character is written according to basic strokes and divided into 32 strokes. According to the basic stroke table of Chinese characters recorded in Baidu encyclopedia, 32 types are recorded in total. Including horizontal, vertical, left falling, right falling, point, etc., and also horizontal folding, vertical hooking, etc., as well as vertical folding, horizontal folding, etc. By collecting data of strokes, a classifier is trained, and each stroke can be classified. The classifier adopts a deep learning convolutional neural network for training. And converting the stroke classification into an image classification problem to obtain a stroke classification model. The stroke classification model may identify to which stroke a user belongs each time the user writes a stroke.
the model classifies the stroke sequences to obtain the stroke recognition result of the character, and the method further comprises the following steps:
Only stroke recognition is carried out on the characters which are regularly written; because the error rate of stroke recognition increases if the writing is irregular, and an irregular word does not require further recognition of whether the strokes are written incorrectly, the teacher can directly judge that the word needs to be written again or to enhance the practice.
And 102, when the student writes, recognizing corresponding strokes of each stroke of the student according to the time sequence of the student in the process of writing, and recording the stroke sequence.
When a student writes each character, stroke cutting is carried out, each character is cut into stroke sequences according to the pen-down time and the pen-up time of each pen, and the stroke sequences are recorded.
When each character is written, stroke cutting is carried out according to the pen-down time and the pen-up time of each pen, each character is cut into a stroke sequence, and the stroke sequence is recorded; training a stroke classification model according to the stroke characteristics;
Classifying each stroke according to the stroke classification model to obtain a stroke identification result;
Recognizing stroke sequences of the whole character and combining the stroke sequences into a stroke recognition sequence of the character;
And 103, recognizing the characters written by the user, and obtaining the recognition technology of the handwritten character through six layers of feature extraction. And when the word cannot be recognized or the classification confidence coefficient is less than a certain threshold value, judging that the word is written irregularly.
adopting a deep learning tool tensorflow to perform handwriting recognition, wherein a convolutional neural network is adopted in a deep learning handwritten character recognition algorithm, a first preset number of convolutional kernels are arranged on a first layer, the size of each convolutional kernel is 6 x 6, and the convolutional kernels are connected with an input layer in a full-connection mode to perform convolution to obtain a first layer of characteristic image; setting a second preset number of pooling on a second layer, wherein the size of a pooling kernel is 4 x 4, and pooling the first layer of feature images to obtain second layer of feature images; converting the second layer characteristic image by using a modified linear unit function at a third layer to obtain a third layer characteristic image; a fourth layer is provided with a third preset number of convolution kernels, each convolution kernel is 6 x 6 in size, and the fourth layer is connected with the input layer in a full-connection mode for convolution to obtain a fourth layer characteristic image; converting the fourth layer characteristic image by using a modified linear unit function to obtain a fifth layer characteristic image; a fourth preset number of pooling cores are arranged on the sixth layer, the size of each pooling core is 3 x 3, and the fifth layer characteristic image is pooled to obtain a sixth layer characteristic image; through the drawing of above-mentioned many times characteristic, can make the model obtain maximum generalization, make even the illegible word of writing, can not discern because of the generalization not enough yet. And the device can better learn the confidence value of the neural network, so that the confidence value of recognition rate is reduced when irregular characters are written.
extracting sample features in a one-dimensional vector form from a prediction sample, converting the sample features in the one-dimensional vector form into sample features in a two-dimensional array form, and generalizing the sample features in the two-dimensional array by taking a two-dimensional subarray as a unit based on the deep learning convolutional neural network model to obtain a processing result; and outputs the learning prediction result and outputs the correct probability, i.e. confidence, of the result. A confidence threshold for the recognized word is preset. And when the handwriting can not be recognized by the trained handwriting recognition algorithm model or the recognition confidence coefficient is smaller than a preset threshold value, judging that the character is written irregularly.
and 104, identifying strokes of the recognized regular characters, and judging the correct stroke sequence.
And identifying the strokes of the characters which are identified as regular characters by using the stroke classification model through the stroke records of the characters to obtain the stroke sequence of the characters learned and written.
the stroke order of a word written by a student is compared with the correct stroke order of the word. When the sequence is different, marking is performed.
the correct stroke order may be obtained through an interface providing stroke queries over the web, or through a dictionary or the like. The correct stroke order is then compared to the word stroke order identified for that word. If each stroke is correct, the writing is considered to be correct, otherwise, the word is described as a stroke error.
Step 105, identifying the distance between the font components. The distance between the components refers to the distance width among a plurality of components forming the character;
as students write, especially pupils often have irregular matching of radicals. Such as left and right components being too far apart, or components being particularly close in proximity. Such writing is irregular. The word that needs to be corrected. If a machine is used to automatically identify whether a word conforms to the correct writing method, deep learning techniques are used to identify the components and to normalize the components.
the method mainly comprises the following steps of training a first training model to read a character sequence, splitting the character sequence word by word, and splitting the character sequence into a radical sequence corresponding to the character sequence; a second training model trained to read the radical sequence output by the first training model and map the radical sequence to a first input sequence corresponding to the radical sequence; the third training model is used for acquiring the distance between the radicals of each character in the character sequence, including the distance between the left radical and the right radical and the distance between the upper part and the lower part in the upper and lower structures, and mapping the distance between the radicals to a second input sequence corresponding to the distance between the radicals;
and training a radical neural network prediction model, wherein the neural network prediction model is configured to receive the first input sequence and the second input sequence, train the model by adopting a seq2seq end-to-end technology, and train for multiple times to obtain an output sequence, and the output sequence comprises the radicals of the character sequence and the distance values between the radicals.
and step 106, judging whether the font component writing is standard or not according to the depth learning prediction result of the font component distance.
And inputting the remote distance of the conventional printing font into the radical neural network prediction model to obtain a distance value between the radical sequence and the radical. Inputting the radical neural network prediction model by acquiring the handwritten word to be predicted to obtain a radical distance value of the handwritten word;
judging whether the distance value of the radical of the handwritten word is larger than or smaller than the correct distance value of the printed word or not and whether the distance value exceeds a preset distance threshold or not;
And if the preset threshold value is exceeded, determining that the font components are not in specification.
And step 107, identifying and labeling wrong characters. And the characters which cannot be recognized by the handwriting or the recognized characters with the confidence coefficient smaller than a certain threshold value are automatically marked as wrongly written characters. The method is characterized in that characters written irregularly and characters with irregular components are marked, marking methods can be used for marking through changing the color of the fonts or changing the size of the fonts or adding underlines and the like, and marking can be distinguished along with stroke errors and component marking and different styles can be carried out.
Step 108, correcting the wrong stroke sequence.
in the same way, for stroke errors, a color different from that of wrongly written characters is adopted, the characters are marked by the wrong strokes, and correct stroke sequences are provided near the characters. The teacher can correct the errors conveniently and the students can improve the errors conveniently.
Through the above description of the embodiments, it is clear to those skilled in the art that the above embodiments can be implemented by software, and can also be implemented by software plus a necessary general hardware platform. With this understanding, the technical solutions of the embodiments can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a usb disk, a removable hard disk, etc.), and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods according to the embodiments of the present invention.
the above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (9)
1. A handwritten word evaluation method and device are characterized in that the method comprises the following steps:
Recognizing the handwriting, and judging that the handwriting is irregular when the handwriting cannot be recognized or the confidence coefficient is smaller than a preset threshold value;
carrying out stroke recognition on the handwritten character;
Comparing the stroke sequence of the handwritten word with the correct stroke sequence, and if the sequences are different, judging that the strokes are wrong;
identifying the distance between the components, and judging whether the font is standard or not according to the deep learning prediction result of the font component distance;
and marking irregular characters, correcting stroke sequence and correcting irregular characters.
2. The method according to claim 1, wherein the recognition of the written word, and when the recognition is not performed or the confidence is smaller than a preset threshold, the writing is determined to be irregular, and the method mainly comprises:
adopting a deep learning tool tensorflow to perform handwriting recognition, wherein a convolutional neural network is adopted in a deep learning handwritten character recognition algorithm, a first preset number of convolutional kernels are arranged on a first layer, the size of each convolutional kernel is 6 x 6, and the convolutional kernels are connected with an input layer in a full-connection mode to perform convolution to obtain a first layer of characteristic image; setting a second preset number of pooling on a second layer, wherein the size of a pooling kernel is 4 x 4, and pooling the first layer of feature images to obtain second layer of feature images; converting the second layer characteristic image by using a modified linear unit function at a third layer to obtain a third layer characteristic image; a fourth layer is provided with a third preset number of convolution kernels, each convolution kernel is 6 x 6 in size, and the fourth layer is connected with the input layer in a full-connection mode for convolution to obtain a fourth layer characteristic image; converting the fourth layer characteristic image by using a modified linear unit function to obtain a fifth layer characteristic image; setting a fourth preset number of pooling kernels on a sixth layer, wherein the size of each pooling kernel is 3 x 3, and pooling the fifth layer feature image to obtain a sixth layer feature image;
extracting sample features in a one-dimensional vector form from a prediction sample, converting the sample features in the one-dimensional vector form into sample features in a two-dimensional array form, and generalizing the sample features in the two-dimensional array by taking the two-dimensional array as a unit based on the deep learning convolutional neural network model to obtain a processing result; and outputs the learning prediction result and outputs the correct probability, i.e. confidence, of the result. And when the handwriting can not be recognized by the handwriting recognition algorithm or the recognition confidence coefficient is smaller than the preset threshold value, judging that the character is written irregularly.
3. The method of claim 1, wherein the stroke recognition of the handwritten word comprises:
according to the authoritative classification method of basic strokes of Chinese characters in a stroke summary table of Chinese characters in encyclopedic, the method is divided into 32 strokes in total, stroke analysis is carried out according to the stroke direction and/or the inflection point of the strokes, when each character is written, stroke cutting is carried out according to the stroke falling time and the stroke starting time of each stroke, each character is cut into stroke sequences, and the stroke sequences are recorded; training a stroke classification model according to the stroke characteristics;
classifying each stroke according to the stroke classification model to obtain a stroke identification result;
recognizing stroke sequences of the whole character and combining the stroke sequences into a stroke recognition sequence of the character;
the model classifies the stroke sequences to obtain the stroke recognition result of the character, and the method further comprises the following steps:
Only stroke recognition is performed on regularly written characters.
4. the method of claim 1, wherein comparing the stroke order of the handwritten word with a correct stroke order and determining that the stroke is incorrect if the order is different comprises:
Obtaining a correct stroke sequence through a stroke query interface or through a dictionary;
the method for acquiring the correct stroke comprises the steps of inputting the stroke of a target word in a stroke query interface on the network, wherein the interface can convert the word into a correct sequence containing the correct stroke sequence; comparing the sequence of the character stroke sequence with the sequence of the correct stroke sequence, and judging whether the sequence of the character completely conforms to the correct sequence of the character;
if the correct stroke order is identical to the word stroke recognition sequence, determining that the strokes are correct;
otherwise, the stroke is judged to be wrongly written.
5. the method of claim 1, wherein said performing distance identification between components comprises:
the first training model is trained to read the character sequence, and splits the character sequence word by word into radical sequences corresponding to the character sequence; a second training model trained to read the radical sequence output by the first training model and map the radical sequence to a first input sequence corresponding to the radical sequence; the third training model is used for acquiring the distance between the radicals of each character in the character sequence, including the distance between the left radical and the right radical and the distance between the upper part and the lower part in the upper and lower structures, and mapping the distance between the radicals to a second input sequence corresponding to the distance between the radicals; the distance between the components refers to the distance width among a plurality of components forming the character;
the neural network prediction model is configured to receive the first input sequence and the second input sequence, train by adopting a seq2seq end-to-end training model and obtain an output sequence, wherein the output sequence comprises the components of the character sequence and the distance values between the components.
6. The method of claim 1, wherein the determining whether the font is normal according to the deep learning prediction result of the font radical distance mainly comprises:
Inputting the remote distance of a conventional printing font into the radical neural network prediction model to obtain a distance value between a radical sequence and a radical; inputting the radical neural network prediction model by acquiring the handwritten word to be predicted to obtain a radical distance value of the handwritten word;
judging whether the distance value of the radical of the handwritten word is larger than or smaller than the correct distance value of the printed word or not and whether the distance value exceeds a preset distance threshold or not;
and if the preset threshold value is exceeded, determining that the font components are not in specification.
7. The method of claim 1, wherein the labeling of irregular words comprises:
and marking the characters with irregular writing and the characters with irregular components by changing the font color or the font size or underlining.
8. The method of claim 1, wherein the correcting the stroke order comprises:
and marking the stroke errors by adopting a color different from that of the irregular character, and providing a correct stroke sequence within the preset area range of the stroke error character.
9. A handwritten word evaluation device, characterized in that said system comprises:
the character regularity judging module is used for judging whether the characters are regular or not;
The stroke classifier training module is used for training a stroke classifier and identifying strokes written by a user;
the stroke sequence judging module is used for recording the stroke sequence written by the user and comparing the stroke sequence with the correct stroke sequence;
The component distance judging module is used for identifying the written components of the user and judging whether the distance between the components is proper;
And the marking and error correction module is used for marking and highlighting the irregular characters and highlighting the characters with incorrect strokes and performing error correction prompt on the correct strokes.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910805073.0A CN110555403A (en) | 2019-08-29 | 2019-08-29 | handwritten character evaluation method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910805073.0A CN110555403A (en) | 2019-08-29 | 2019-08-29 | handwritten character evaluation method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110555403A true CN110555403A (en) | 2019-12-10 |
Family
ID=68738078
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910805073.0A Withdrawn CN110555403A (en) | 2019-08-29 | 2019-08-29 | handwritten character evaluation method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110555403A (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111428623A (en) * | 2020-03-20 | 2020-07-17 | 郑州工程技术学院 | Chinese blackboard-writing style analysis system based on big data and computer vision |
| US20200251217A1 (en) * | 2019-12-12 | 2020-08-06 | Renee CASSUTO | Diagnosis Method Using Image Based Machine Learning Analysis of Handwriting |
| CN111680761A (en) * | 2020-06-17 | 2020-09-18 | 北京字节跳动科技有限公司 | Information feedback method and device and electronic equipment |
| CN111695539A (en) * | 2020-06-17 | 2020-09-22 | 北京一起教育信息咨询有限责任公司 | Evaluation method and device for handwritten Chinese characters and electronic equipment |
| CN111695537A (en) * | 2020-06-17 | 2020-09-22 | 北京一起教育信息咨询有限责任公司 | Method and device for stroke recognition and electronic equipment |
| CN111797822A (en) * | 2020-07-03 | 2020-10-20 | 北京字节跳动网络技术有限公司 | Character object evaluation method and device and electronic equipment |
| CN111814743A (en) * | 2020-07-30 | 2020-10-23 | 深圳壹账通智能科技有限公司 | Handwriting recognition method and device and computer readable storage medium |
| CN111898588A (en) * | 2020-08-24 | 2020-11-06 | 河南羲和网络科技股份有限公司 | Method and device for standardizing writing strokes |
| CN112215061A (en) * | 2020-08-27 | 2021-01-12 | 拓尔思信息技术股份有限公司 | Detection method and device for copying on handwriting screen, electronic equipment and storage medium |
| CN112446349A (en) * | 2020-12-09 | 2021-03-05 | 北京有竹居网络技术有限公司 | Handwriting detection method and device |
| CN112633950A (en) * | 2021-01-06 | 2021-04-09 | 上海归程网络科技有限公司 | Mobile digital marketing effect evaluation system |
| CN113627260A (en) * | 2021-07-12 | 2021-11-09 | 科大讯飞股份有限公司 | Method, system and computing device for recognizing stroke order of handwritten Chinese characters |
| CN114140799A (en) * | 2021-08-11 | 2022-03-04 | 杭州师范大学 | Real-time feedback and evaluation method for digital writing exercise of children |
| KR102468713B1 (en) * | 2022-07-07 | 2022-11-21 | 주식회사 에이치투케이 | AI- based Device and Method for Stroke Order Recognition of Korean Handwriting of Student |
-
2019
- 2019-08-29 CN CN201910805073.0A patent/CN110555403A/en not_active Withdrawn
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200251217A1 (en) * | 2019-12-12 | 2020-08-06 | Renee CASSUTO | Diagnosis Method Using Image Based Machine Learning Analysis of Handwriting |
| CN111428623A (en) * | 2020-03-20 | 2020-07-17 | 郑州工程技术学院 | Chinese blackboard-writing style analysis system based on big data and computer vision |
| CN111680761B (en) * | 2020-06-17 | 2022-05-10 | 北京字节跳动网络技术有限公司 | Information feedback method and device and electronic equipment |
| CN111680761A (en) * | 2020-06-17 | 2020-09-18 | 北京字节跳动科技有限公司 | Information feedback method and device and electronic equipment |
| CN111695539A (en) * | 2020-06-17 | 2020-09-22 | 北京一起教育信息咨询有限责任公司 | Evaluation method and device for handwritten Chinese characters and electronic equipment |
| CN111695537A (en) * | 2020-06-17 | 2020-09-22 | 北京一起教育信息咨询有限责任公司 | Method and device for stroke recognition and electronic equipment |
| CN111797822A (en) * | 2020-07-03 | 2020-10-20 | 北京字节跳动网络技术有限公司 | Character object evaluation method and device and electronic equipment |
| CN111797822B (en) * | 2020-07-03 | 2024-01-23 | 北京字节跳动网络技术有限公司 | Text object evaluation method and device and electronic equipment |
| CN111814743A (en) * | 2020-07-30 | 2020-10-23 | 深圳壹账通智能科技有限公司 | Handwriting recognition method and device and computer readable storage medium |
| CN111898588A (en) * | 2020-08-24 | 2020-11-06 | 河南羲和网络科技股份有限公司 | Method and device for standardizing writing strokes |
| CN112215061A (en) * | 2020-08-27 | 2021-01-12 | 拓尔思信息技术股份有限公司 | Detection method and device for copying on handwriting screen, electronic equipment and storage medium |
| CN112446349A (en) * | 2020-12-09 | 2021-03-05 | 北京有竹居网络技术有限公司 | Handwriting detection method and device |
| CN112633950A (en) * | 2021-01-06 | 2021-04-09 | 上海归程网络科技有限公司 | Mobile digital marketing effect evaluation system |
| CN113627260A (en) * | 2021-07-12 | 2021-11-09 | 科大讯飞股份有限公司 | Method, system and computing device for recognizing stroke order of handwritten Chinese characters |
| CN114140799A (en) * | 2021-08-11 | 2022-03-04 | 杭州师范大学 | Real-time feedback and evaluation method for digital writing exercise of children |
| CN114140799B (en) * | 2021-08-11 | 2024-09-06 | 杭州师范大学 | Real-time feedback and judgment method for digital writing exercise of children |
| KR102468713B1 (en) * | 2022-07-07 | 2022-11-21 | 주식회사 에이치투케이 | AI- based Device and Method for Stroke Order Recognition of Korean Handwriting of Student |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110555403A (en) | handwritten character evaluation method and system | |
| CN111753767B (en) | Method and device for automatically correcting operation, electronic equipment and storage medium | |
| US10853638B2 (en) | System and method for extracting structured information from image documents | |
| CN101326518B (en) | Method and device for script recognition for ink notes | |
| US20070286486A1 (en) | System and method for automated reading of handwriting | |
| CN1226696C (en) | Explanatory and search for handwriting sloppy Chinese characters based on shape of radicals | |
| CN1770174A (en) | Parsing hierarchical lists and outlines | |
| CN108664975B (en) | Uyghur handwritten letter recognition method and system and electronic equipment | |
| CN110598686A (en) | Invoice identification method, system, electronic equipment and medium | |
| JP6170860B2 (en) | Character recognition device and identification function generation method | |
| CN111932418B (en) | Student learning condition identification method and system, teaching terminal and storage medium | |
| CN114730241B (en) | Gesture and stroke recognition in touch user interface input | |
| CN113011412A (en) | Character recognition method, device, equipment and storage medium based on stroke order and OCR (optical character recognition) | |
| CN115359500A (en) | Intelligent pen writing method and device with Chinese character error correction function and storage medium | |
| CN110414622B (en) | Classifier training method and device based on semi-supervised learning | |
| CN115661183B (en) | Intelligent scanning management system and method based on edge calculation | |
| CN104951811A (en) | Handwriting trace recognition method and device for writing with writing brush | |
| CN117095414A (en) | Handwriting recognition system and recognition method based on dot matrix paper pen | |
| JP2020115175A (en) | Information processor, method for processing information, and program | |
| CN107209862A (en) | Program, information storage medium and identifying device | |
| K Jabde et al. | A Comprehensive Literature Review on Air-written Online Handwritten Recognition | |
| Al-Barhamtoshy et al. | A toolkit for teaching Arabic handwriting | |
| CN117523578B (en) | Classroom paper pen interaction method and device based on artificial intelligence | |
| Nisa et al. | Annotation of struck-out text in handwritten documents | |
| JP7338203B2 (en) | Character recognition device and program |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20191210 |