CN111428623A - Chinese blackboard-writing style analysis system based on big data and computer vision - Google Patents
Chinese blackboard-writing style analysis system based on big data and computer vision Download PDFInfo
- Publication number
- CN111428623A CN111428623A CN202010203565.5A CN202010203565A CN111428623A CN 111428623 A CN111428623 A CN 111428623A CN 202010203565 A CN202010203565 A CN 202010203565A CN 111428623 A CN111428623 A CN 111428623A
- Authority
- CN
- China
- Prior art keywords
- writing
- blackboard
- handwriting
- pen
- acquisition unit
- 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.)
- Granted
Links
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
- G06V30/32—Digital ink
- G06V30/333—Preprocessing; Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- 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
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/20—Education
- G06Q50/205—Education administration or guidance
-
- 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
Abstract
The invention discloses a Chinese blackboard writing style analysis system based on big data and computer vision. The system comprises: the method comprises the steps that a data acquisition unit acquires a blackboard writing handwriting image and a pen-down state sequence; extracting the pen-falling state characteristics of the unit through the pen-falling state characteristics; obtaining a handwriting segmentation heat map by using a handwriting segmentation heat map acquisition unit; the writing characteristic extraction unit extracts the characteristics of the handwriting segmentation heat map; the blackboard writing comprehensive characteristic acquisition unit performs fusion analysis on the blackboard writing characteristic and the handwriting state characteristic to obtain a blackboard writing comprehensive characteristic; and finally, extracting comprehensive features of the blackboard writing to be analyzed and performing similarity matching to obtain a matched blackboard writing style. The method realizes the board writing style analysis in the field of education and teaching, and has the advantages of more objectivity of analysis results, high analysis efficiency and low implementation difficulty.
Description
Technical Field
The invention belongs to the field of computer vision and big data, and particularly relates to a Chinese blackboard-writing style analysis system based on big data and computer vision.
Background
The blackboard writing is an auxiliary teaching mode of conventional classroom education widely applied to education places such as colleges and universities, enterprise training, vocational training, primary and secondary school education and the like. Specifically, the instructor synchronously performs board description auxiliary explanation when using the presentation to explain the courseware. Along with the development of science and technology, electronic blackboard writing gradually enters the education industry. The appearance of the electronic blackboard-writing brings convenience to classroom education, and teachers and students are free from dust troubles, so that the health of the teachers and the students is guaranteed. At present, in classroom education, electronic blackboard books have accumulated a large amount of data and have not been used for analysis. Moreover, due to the lack of a corresponding data system, teachers cannot pay attention to blackboard-writing data similar to or different from own styles.
The writing style and writing rhythm are key indexes for relation teaching quality and teaching result. At present, the analysis and review of blackboard writing is mainly carried out in a mode that a reviewer listens to courses at any place and then carries out subjective scoring according to a scoring table and the like. The conventional blackboard-writing analysis and review mode is lack of objectivity, difficult to implement and low in review efficiency.
Therefore, the existing blackboard-writing analysis field has the problems of lack of objectivity in analysis mode, high implementation difficulty and low analysis efficiency.
Disclosure of Invention
The invention aims to provide a Chinese blackboard-writing style analysis system based on big data and computer vision aiming at the defects in the prior art, so that blackboard-writing style analysis is realized, the objectivity of an analysis result is ensured, the implementation difficulty is reduced, and the analysis efficiency is improved.
A Chinese writing style analysis system based on big data and computer vision, the system comprising:
the data acquisition unit is used for acquiring the blackboard writing handwriting image and the pen-down state sequence;
the pen-down state feature extraction unit is used for inputting the pen-down state sequence into the first fully-connected neural network to obtain pen-down state features;
the handwriting segmentation heat map acquisition unit is used for inputting the blackboard writing handwriting image obtained from the data acquisition unit into a handwriting semantic segmentation deep neural network to obtain a handwriting segmentation heat map;
the blackboard writing feature extraction unit is used for inputting the handwriting segmentation heat map into a blackboard writing feature extraction network;
the blackboard-writing comprehensive characteristic acquisition unit is used for performing joint operation on the blackboard-writing comprehensive characteristics and the handwriting state characteristics and inputting the blackboard-writing comprehensive characteristics into a third fully-connected neural network to obtain the blackboard-writing comprehensive characteristics;
and the blackboard-writing style analysis unit is used for extracting comprehensive blackboard-writing features of the blackboard writing to be analyzed, and performing similar matching on the extracted features and the features in the blackboard-writing feature database to obtain a matched blackboard-writing style.
Further, the handwriting semantic segmentation deep neural network comprises:
the handwriting characteristic encoder is used for extracting the characteristics of the blackboard writing handwriting image obtained from the data acquisition unit;
and the handwriting characteristic decoder is used for decoding the output of the handwriting characteristic encoder, obtaining a handwriting segmentation heat map through convolution and pooling, and representing the confidence coefficient of the handwriting by hot spots in the handwriting segmentation heat map.
Further, the network for extracting writing characteristics of the blackboard comprises:
the blackboard writing feature encoder is used for extracting features of the handwriting segmentation heat map;
and the second full-connection neural network is used for carrying out characteristic weighting on the output of the writing characteristic encoder to obtain the writing characteristics of the blackboard.
Furthermore, the pen-falling state feature extraction unit and the handwriting segmentation heat map acquisition unit are arranged at a local end to complete calculation, and the blackboard writing feature extraction unit, the blackboard writing comprehensive feature acquisition unit and the blackboard writing style analysis unit are arranged at a cloud end to complete calculation.
Furthermore, the blackboard writing handwriting image is a color image, and RGB three-channel data of the blackboard writing handwriting image is acquired as input of the handwriting segmentation heat map acquisition unit.
And further, processing the pen-down state sequence into a fixed-length time sequence signal through interpolation.
Compared with the prior art, the invention has the following beneficial effects:
1. the invention designs the neural network structure to analyze the blackboard writing based on the computer vision technology and the deep learning technology, and compared with the traditional blackboard writing analysis and review mode, the invention ensures the objectivity of the analysis and review result, and has low implementation difficulty and high analysis efficiency.
2. The invention analyzes the blackboard writing image and the pen-falling state sequence, not only can obtain the spatial domain characteristic of the blackboard writing, but also can obtain the time domain rhythm characteristic of the blackboard writing, and the analysis result is more reasonable.
3. The invention designs a handwriting semantic segmentation deep neural network to extract character masks, and can isolate differences under different working conditions.
4. The invention integrates the electronic blackboard-writing data and designs the blackboard-writing feature database, so that customers can obtain blackboard-writing data with similar styles or different styles.
5. According to the invention, the neural network structure is divided into a first stage and a second stage, the first stage is arranged at the local end, and the calculation of the second stage is arranged at the cloud end, so that on one hand, the calculation amount of the cloud end can be greatly reduced through the configuration, and on the other hand, the confidentiality of the cloud end post-processing neural network weight can be ensured.
Drawings
FIG. 1 is a diagram of a neural network architecture of the system of 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 further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
The invention provides a Chinese blackboard-writing style analysis system based on big data and computer vision. FIG. 1 is a diagram of a neural network architecture of the system of the present invention. The following description will be made by way of specific examples.
The first embodiment is as follows:
the Chinese blackboard-writing style analysis system based on big data and computer vision comprises:
and the data acquisition unit is used for acquiring the blackboard writing handwriting image and the pen-down state sequence.
The method comprises the steps of analyzing the style of a blackboard writing, firstly obtaining the content of the blackboard writing, including a blackboard writing image and a pen-down state sequence. The blackboard writing image is a color image, and RGB three-channel data can be acquired.
Acquiring a blackboard writing handwriting image and pen-down information, and a recommendation method comprises the following steps: and calling by using the API to respectively obtain a writing-on state sequence S of the writing-on handwriting layer of the blackboard writing and a writing-off state sequence S of the current page written by the user by using the electromagnetic pen. Namely, the implementer acquires corresponding data from the online classroom education platform by using the program interface. In addition, the blackboard-writing content of the electronic whiteboard can be acquired through means such as window handle resource capture.
The content of the obtained writing image B, B of the blackboard writing is writing image information written by using an electromagnetic pen, and is not PPT content, and specifically, the content is a color image. The typical blackboard-writing image B should contain writing color information, and the types of strokes may be different, such as writing with thickness simulation or writing with uniform thickness. For some working conditions with poor software conditions, handwriting color information may not exist, and in such a case, even if the obtained image is black and white, the image is converted into a three-channel image to be processed next step.
And the pen-down state feature extraction unit is used for inputting the pen-down state sequence into the first fully-connected neural network to obtain the pen-down state features.
Since the analysis system of the present invention is analyzed page by page, it is described herein with respect to a single sheet of board.
Firstly, the calculation of a first stage is carried out, the calculation of the first stage is carried out at a local end, and the calculation of the first stage comprises a pen-down state feature extraction unit and a handwriting segmentation heat map acquisition unit. The local end is the client end using the electronic blackboard book. The main advantage of placing the first stage of computation on the client side is that the computation load of the cloud server can be greatly reduced by using the computation resources of the client side. According to the writing contents of the current page, the pen-down time state is subjected to feature extraction, and is sampled to be a fixed-length time sequence signal, such as S' 1: 4096: 1024 of 4096: 1024 sample points. It is worth noting that the width of the input of the first fully connected neural network FC1 responsible for extracting the features of the fixed-length signal S' should be determined as a practical matter, where the recommended value is 4096 × 1024 sample inputs, resulting in a high-dimensional feature value V1.
In fact, the pen-down status information S is not fixed, and since the pen-down feature of the blackboard writing at any time needs to be identified uniformly, the one-dimensional signal S needs to be resampled to be the fixed-length signal S'. As for the pen-down status information S, it is characterized in that S is a binary sequence, i.e. S is in the pen-down state at the time of t1t1When the pen is lifted at time t2, S is 1t20. For resampling, it should be noted that, since S and S' are both binary sequences, a nearest neighbor interpolation method or a linear interpolation method may be used, and a linear interpolation method is suggested.
And the handwriting segmentation heat map acquisition unit is used for inputting the blackboard writing image acquired from the data acquisition unit into a handwriting semantic segmentation deep neural network to acquire a handwriting segmentation heat map, and the handwriting semantic segmentation deep neural network comprises a handwriting characteristic encoder and a handwriting characteristic decoder. The handwriting characteristic encoder is used for extracting the characteristics of the blackboard writing handwriting image obtained from the data acquisition unit; and the handwriting characteristic decoder is used for decoding the output of the handwriting characteristic encoder, obtaining a handwriting segmentation heat map through convolution and pooling, and representing the confidence coefficient of the handwriting by hot spots in the handwriting segmentation heat map.
In order to overcome the difference between the writing information of the blackboard under various working conditions, such as the difference between the blackboard and the whiteboard, the difference under different illumination, and the stroke difference under different ink marks, common characteristics need to be extracted to isolate the influence of the working conditions and objects on the writing style description. In order to isolate other parts irrelevant to handwriting, the invention uses a semantic segmentation network to segment the character part.
Because the image B needs to be input into the semantic segmentation deep neural network to obtain the part belonging to the character, in order to ensure the accuracy of segmentation, the image B needs to be scaled to a uniform size.
The handwriting semantic segmentation depth neural network is based on a semantic segmentation depth neural network, comprises a handwriting characteristic encoder EncA and a handwriting characteristic decoder DecA, and is used for end-to-end semantic segmentation. Where EncA and DecA are both conventional semantic segmentation encoder-decoders, implementers can refer to the presently disclosed deep neural networks, such as SegNet, ENet, etc.
The confidence of a character part is expressed in the thermodynamic diagram H with a single channel, the result of semantic segmentation generally needs to be thresholded, but in order to avoid losing detailed information, the thermodynamic diagram H is not subjected to post-processing, and the result is directly sent to a next layer of network for analysis. So far, writing characters of the blackboard writing under different working conditions are uniformly described in a heat map form.
After the handwriting segmentation processing is performed on the blackboard writing, the acquired handwriting segmentation heat map and the pen-down state also need to be subjected to blackboard writing post-processing, namely, the calculation of the second stage of the invention. And the second stage comprises the steps of extracting the characteristics of the handwriting segmentation heat map and further obtaining a characteristic descriptor through a third fully-connected neural network by combining the characteristics of the pen-down state. Due to the consideration of confidentiality of the blackboard-book post-processing deep neural network and the weight thereof and the need of a large amount of data access during matching, the computing position at the second stage is a cloud server or a server cluster. Therefore, the handwriting segmentation heatmap and the pen-down status features need to be uploaded to the cloud for further processing.
And the writing characteristic extraction unit is used for inputting the handwriting semantic heat map into a writing characteristic extraction network, and the writing characteristic extraction network comprises a writing characteristic encoder and a second full-connection neural network.
After the cloud obtains the handwriting segmentation heat map, the potential spatial domain styles such as writing composition, stroke, size and the like of the handwriting need to be analyzed continuously, so that the spatial domain feature description needs to be further performed on the handwriting segmentation.
Specifically, the obtained single-channel handwriting segmentation heat map H is input into a blackboard writing feature extraction deep convolution neural network, and the network feature is that a single-channel image H is input, and a group of high-dimensional feature values V2, namely blackboard writing features, are output through convolution calculation and full-connection neural network.
Thus, the pen-down status feature V1 and the writing on blackboard feature V2 are obtained. The pen-down status feature and the writing feature need to be integrated to obtain the writing integrated feature. Therefore, the invention also comprises a blackboard-writing comprehensive characteristic acquisition unit.
And the blackboard-writing comprehensive characteristic acquisition unit is used for performing channel-by-channel joint operation on the blackboard-writing comprehensive characteristics and the handwriting state characteristics and inputting the blackboard-writing comprehensive characteristics and the handwriting state characteristics into the third fully-connected neural network to obtain the blackboard-writing comprehensive characteristics.
The blackboard-writing integrated feature obtaining unit performs a splicing/merging (merging) operation on the V1 and the V2, and obtains a high-dimensional feature value V, namely the blackboard-writing integrated feature through a third fully-connected neural network, wherein the dimension of the V is the sum of the V1 dimension and the V2 dimension.
Therefore, the blackboard-writing comprehensive characteristics for objectively describing the spatial domain style and the time domain rhythm of the blackboard-writing can be obtained.
The blackboard-writing style analysis system of the invention is based on the deep neural network of fig. 1, which needs to be trained with a prepared sample set to obtain better performance. The data preparation and training of the overall network architecture of the present invention is described in detail below.
As shown in FIG. 1, the solid line part is a handwriting semantic segmentation deep neural network containing EncA and DecA. The handwriting semantic segmentation deep neural network is based on the semantic segmentation deep neural network, and the network is trained by using the original image and the labeled heat map. For EncA and DecA training data preparation, the recommended implementer extracts the blackboard writing text mask by combining a proper machine vision preprocessing method and partial manual marking, and the specific method can be a marking method of self-adaptive threshold value and manual correction.
In fig. 1, the dotted line part is a sub-network of the whole DNN, a training form of a twin network is adopted, training data are a pen-down sequence S' and a handwriting segmentation heat map H obtained in an early stage, and training can be performed independently from a semantic segmentation network.
The method includes the steps of firstly, simply classifying book styles based on manual labeling, dividing similar single pages into one group, and also adopting a semi-supervised mode, wherein an implementer can reduce labeling cost by using clustering or other means, and the semi-supervised mode is out of the discussion range of the invention.
And the blackboard-writing style analysis unit is used for extracting comprehensive blackboard-writing features of the blackboard writing to be analyzed, and performing similar matching on the extracted features and the features in the blackboard-writing feature database to obtain a matched blackboard-writing style.
And after analyzing all the blackboard-writing data, storing the obtained comprehensive blackboard-writing characteristics into a blackboard-writing characteristic database. And after the system receives the blackboard-writing data to be analyzed, extracting the comprehensive blackboard-writing features of the blackboard-writing data to obtain comprehensive blackboard-writing features, and performing similar matching on the features in the rest blackboard-writing feature databases to obtain the matched blackboard-writing style. Furthermore, blackboard writing contents with similar styles can be pushed to teachers for teachers to teach and refer. In addition, blackboard-writing contents with different styles, namely blackboard-writing contents with low similarity can be pushed, so that teachers can open up new ideas. Teaching and research, teacher and user can learn and comment on the book contents page by using the system of the invention, thus realizing the combination of electric teaching and big data. In order to match and distinguish the types of the blackboard writing, a cosine similarity mode is recommended to be used for searching and classifying. And sequencing the finally matched single-page results according to the similarity, and selecting a proper number of single-page blackboard-writing to provide for teachers, so that the teachers can conveniently learn and comment on blackboard-writing contents with similar styles. By analogy, teaching and research, teacher and user use the system of the invention to study and comment on the contents of the board page by page, thus realizing the combination of electronic teaching and big data. It should be noted that, if the blackboard-writing is performed by multiple pages, the style matching results of the multiple pages are integrated, and recommendation of the blackboard-writing content meeting the corresponding similarity requirement is obtained. For example, style matching results of each page are obtained, style types of each page are obtained, and the content of the blackboard-writing with the style most similar to that of each page is pushed to the client.
The invention designs the neural network structure to analyze the blackboard writing based on the computer vision technology and the deep learning technology, and compared with the traditional blackboard writing analysis and review mode, the invention ensures the objectivity of the analysis and review result, and has low implementation difficulty and high analysis efficiency. The invention analyzes the blackboard writing image and the pen-falling state sequence, not only can obtain the airspace characteristic of the blackboard writing, but also can obtain the rhythm characteristic of the blackboard writing, and the analysis result is more reasonable. The invention designs a handwriting semantic segmentation deep neural network to extract character masks, and can isolate differences under different working conditions. The invention integrates the electronic blackboard-writing data and designs the blackboard-writing feature database, so that customers can obtain blackboard-writing data with similar styles or different styles. According to the invention, the neural network structure is divided into a first stage and a second stage, the first stage is arranged at the local end, and the calculation of the second stage is arranged at the cloud end, so that on one hand, the calculation amount of the cloud end can be greatly reduced through the configuration, and on the other hand, the confidentiality of the cloud end post-processing neural network weight can be ensured.
The above embodiments are merely preferred embodiments of the present invention, which should not be construed as limiting the present invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (6)
1. A Chinese writing style analysis system based on big data and computer vision, the system includes:
the data acquisition unit is used for acquiring the blackboard writing handwriting image and the pen-down state sequence;
the pen-down state feature extraction unit is used for inputting the pen-down state sequence into the first fully-connected neural network to obtain pen-down state features;
the handwriting segmentation heat map acquisition unit is used for inputting the blackboard writing handwriting image obtained from the data acquisition unit into a handwriting semantic segmentation deep neural network to obtain a handwriting segmentation heat map;
the blackboard writing feature extraction unit is used for inputting the handwriting segmentation heat map into a blackboard writing feature extraction network;
the blackboard-writing comprehensive characteristic acquisition unit is used for performing joint operation on the blackboard-writing comprehensive characteristics and the handwriting state characteristics and inputting the blackboard-writing comprehensive characteristics into a third fully-connected neural network to obtain the blackboard-writing comprehensive characteristics;
and the blackboard-writing style analysis unit is used for extracting comprehensive blackboard-writing features of the blackboard writing to be analyzed, and performing similar matching on the extracted features and the features in the blackboard-writing feature database to obtain a matched blackboard-writing style.
2. The system of claim 1, wherein the handwriting semantic segmentation deep neural network comprises:
the handwriting characteristic encoder is used for extracting the characteristics of the blackboard writing handwriting image obtained from the data acquisition unit;
and the handwriting characteristic decoder is used for decoding the output of the handwriting characteristic encoder, obtaining a handwriting segmentation heat map through convolution and pooling, and representing the confidence coefficient of the handwriting by hot spots in the handwriting segmentation heat map.
3. The system of claim 1, wherein the network of blackboard writing feature extractions comprises:
the blackboard writing feature encoder is used for extracting features of the handwriting segmentation heat map;
and the second full-connection neural network is used for carrying out characteristic weighting on the output of the writing characteristic encoder to obtain the writing characteristics of the blackboard.
4. The system according to any one of claims 1 to 3, wherein the pen-down state feature extraction unit and the handwriting segmentation heatmap acquisition unit are arranged on a local end to complete calculation, and the blackboard writing feature extraction unit, the blackboard writing comprehensive feature acquisition unit and the blackboard writing style analysis unit are arranged on a cloud end to complete calculation.
5. The system of claim 4, wherein the blackboard-writing image is a color image, and RGB three-channel data thereof is acquired as an input of the handwriting segmentation heatmap acquisition unit.
6. The system of claim 4, wherein the sequence of pen-down states is processed by interpolation as a fixed-length timing signal.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010203565.5A CN111428623B (en) | 2020-03-20 | 2020-03-20 | Chinese blackboard-writing style analysis system based on big data and computer vision |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010203565.5A CN111428623B (en) | 2020-03-20 | 2020-03-20 | Chinese blackboard-writing style analysis system based on big data and computer vision |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111428623A true CN111428623A (en) | 2020-07-17 |
| CN111428623B CN111428623B (en) | 2020-12-18 |
Family
ID=71548450
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010203565.5A Active CN111428623B (en) | 2020-03-20 | 2020-03-20 | Chinese blackboard-writing style analysis system based on big data and computer vision |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111428623B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114821399A (en) * | 2022-04-07 | 2022-07-29 | 厦门大学 | Intelligent classroom-oriented blackboard writing automatic extraction method |
Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1373432A (en) * | 2001-02-28 | 2002-10-09 | 曾立彬 | Method and system for recognizing personal characteristics of scrip |
| US20050185842A1 (en) * | 2000-01-06 | 2005-08-25 | Williams David R. | Pen-based handwritten character recognition and storage system |
| US20070065003A1 (en) * | 2005-09-21 | 2007-03-22 | Lockheed Martin Corporation | Real-time recognition of mixed source text |
| CN101253513A (en) * | 2005-08-29 | 2008-08-27 | 微软公司 | Style aware use of writing input |
| CN101295371A (en) * | 2008-04-18 | 2008-10-29 | 浙江大学 | Computer estimation method of Chinese character writing morphology beauty |
| CN101465074A (en) * | 2008-10-21 | 2009-06-24 | 华南理工大学 | Method for evaluating hand-written Hanzi layout |
| CN102609735A (en) * | 2012-02-06 | 2012-07-25 | 安徽科大讯飞信息科技股份有限公司 | Method and apparatus for assessing standard fulfillment of character writing |
| CN104050453A (en) * | 2014-06-24 | 2014-09-17 | 南京师范大学 | Evaluation method for handwritten Chinese character handwriting |
| WO2015150911A2 (en) * | 2014-04-04 | 2015-10-08 | Myscript | System and method for superimposed handwriting recognition technology |
| CN106384094A (en) * | 2016-09-18 | 2017-02-08 | 北京大学 | Chinese word stock automatic generation method based on writing style modeling |
| CN106407874A (en) * | 2016-03-25 | 2017-02-15 | 东南大学 | Handwriting recognition method based on handwriting coordinate sequence |
| CN108154136A (en) * | 2018-01-15 | 2018-06-12 | 众安信息技术服务有限公司 | For identifying the method, apparatus of writing and computer-readable medium |
| CN110555403A (en) * | 2019-08-29 | 2019-12-10 | 韶关市启之信息技术有限公司 | handwritten character evaluation method and system |
| CN110674671A (en) * | 2018-07-03 | 2020-01-10 | 富士施乐株式会社 | System, method and computer readable medium for capturing stroke ink |
-
2020
- 2020-03-20 CN CN202010203565.5A patent/CN111428623B/en active Active
Patent Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050185842A1 (en) * | 2000-01-06 | 2005-08-25 | Williams David R. | Pen-based handwritten character recognition and storage system |
| CN1373432A (en) * | 2001-02-28 | 2002-10-09 | 曾立彬 | Method and system for recognizing personal characteristics of scrip |
| CN101253513A (en) * | 2005-08-29 | 2008-08-27 | 微软公司 | Style aware use of writing input |
| US20070065003A1 (en) * | 2005-09-21 | 2007-03-22 | Lockheed Martin Corporation | Real-time recognition of mixed source text |
| CN101295371A (en) * | 2008-04-18 | 2008-10-29 | 浙江大学 | Computer estimation method of Chinese character writing morphology beauty |
| CN101465074A (en) * | 2008-10-21 | 2009-06-24 | 华南理工大学 | Method for evaluating hand-written Hanzi layout |
| CN102609735A (en) * | 2012-02-06 | 2012-07-25 | 安徽科大讯飞信息科技股份有限公司 | Method and apparatus for assessing standard fulfillment of character writing |
| WO2015150911A2 (en) * | 2014-04-04 | 2015-10-08 | Myscript | System and method for superimposed handwriting recognition technology |
| CN104050453A (en) * | 2014-06-24 | 2014-09-17 | 南京师范大学 | Evaluation method for handwritten Chinese character handwriting |
| CN106407874A (en) * | 2016-03-25 | 2017-02-15 | 东南大学 | Handwriting recognition method based on handwriting coordinate sequence |
| CN106384094A (en) * | 2016-09-18 | 2017-02-08 | 北京大学 | Chinese word stock automatic generation method based on writing style modeling |
| CN108154136A (en) * | 2018-01-15 | 2018-06-12 | 众安信息技术服务有限公司 | For identifying the method, apparatus of writing and computer-readable medium |
| CN110674671A (en) * | 2018-07-03 | 2020-01-10 | 富士施乐株式会社 | System, method and computer readable medium for capturing stroke ink |
| CN110555403A (en) * | 2019-08-29 | 2019-12-10 | 韶关市启之信息技术有限公司 | handwritten character evaluation method and system |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114821399A (en) * | 2022-04-07 | 2022-07-29 | 厦门大学 | Intelligent classroom-oriented blackboard writing automatic extraction method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111428623B (en) | 2020-12-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110110581B (en) | Test paper correcting method and system based on artificial intelligence | |
| CN112115736B (en) | Job correction method and system based on image recognition and intelligent terminal | |
| CN109389061A (en) | Paper recognition methods and system | |
| CN105427696A (en) | Method for distinguishing answer to target question | |
| CN110490081A (en) | A kind of remote sensing object decomposition method based on focusing weight matrix and mutative scale semantic segmentation neural network | |
| CN106846961A (en) | The treating method and apparatus of electronic test paper | |
| CN110837793A (en) | Intelligent recognition handwriting mathematical formula reading and amending system | |
| CN111627088A (en) | Sample automatic generation method for mathematical test paper image recognition | |
| CN109902683A (en) | Paper core point and Data Input method and system | |
| CN112115252A (en) | Intelligent auxiliary writing processing method and device, electronic equipment and storage medium | |
| CN110689018A (en) | Intelligent marking system and processing method thereof | |
| CN112183143A (en) | Reading and amending method and teaching system | |
| CN111428623B (en) | Chinese blackboard-writing style analysis system based on big data and computer vision | |
| CN112116840B (en) | Job correction method and system based on image recognition and intelligent terminal | |
| CN112966518A (en) | High-quality answer identification method for large-scale online learning platform | |
| CN113888757A (en) | Examination paper intelligent analysis method, examination paper intelligent analysis system and storage medium based on benchmarking evaluation | |
| CN113011154A (en) | Job duplicate checking method based on deep learning | |
| CN111353397B (en) | Big data and OCR (optical character recognition) based structured sharing system for Chinese blackboard-writing in online classroom | |
| Khan | Arabic handwritten alphabets, words and paragraphs per user (AHAWP) dataset | |
| CN116010569A (en) | Online answering method, system, electronic equipment and storage medium | |
| CN111666474B (en) | Whole page question searching method and terminal | |
| CN110826330B (en) | Name recognition method and device, computer equipment and readable storage medium | |
| Rajesh et al. | Digitized exam paper evaluation | |
| CN108985175B (en) | Handwritten Chinese sentence set recognition method based on standard peripheral outline and deep learning | |
| Jones | Integrating machine learning in secondary geometry |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |