CN110866499A - Handwritten text recognition method, system, device and medium - Google Patents
Handwritten text recognition method, system, device and medium Download PDFInfo
- Publication number
- CN110866499A CN110866499A CN201911122436.7A CN201911122436A CN110866499A CN 110866499 A CN110866499 A CN 110866499A CN 201911122436 A CN201911122436 A CN 201911122436A CN 110866499 A CN110866499 A CN 110866499A
- Authority
- CN
- China
- Prior art keywords
- motion signal
- character recognition
- character
- motion
- user
- 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
- 238000000034 method Methods 0.000 title claims abstract description 105
- 230000033001 locomotion Effects 0.000 claims abstract description 200
- 230000008569 process Effects 0.000 claims abstract description 45
- 230000009471 action Effects 0.000 claims abstract description 42
- 238000012549 training Methods 0.000 claims description 81
- 238000003860 storage Methods 0.000 claims description 17
- 238000000605 extraction Methods 0.000 claims description 10
- 239000013598 vector Substances 0.000 claims description 10
- 241001422033 Thestylus Species 0.000 claims description 9
- 238000009826 distribution Methods 0.000 claims description 9
- 238000000354 decomposition reaction Methods 0.000 claims description 7
- 238000010606 normalization Methods 0.000 claims description 6
- 239000011159 matrix material Substances 0.000 claims description 5
- 239000000203 mixture Substances 0.000 claims description 4
- 238000001514 detection method Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 238000001228 spectrum Methods 0.000 claims description 3
- 230000007547 defect Effects 0.000 description 25
- 238000005516 engineering process Methods 0.000 description 12
- 238000012545 processing Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 6
- 238000012015 optical character recognition Methods 0.000 description 5
- 230000006870 function Effects 0.000 description 4
- 230000003287 optical effect Effects 0.000 description 4
- 238000013473 artificial intelligence Methods 0.000 description 3
- 238000012550 audit Methods 0.000 description 3
- 230000011218 segmentation Effects 0.000 description 3
- 230000001133 acceleration Effects 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 238000005111 flow chemistry technique Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 239000013307 optical fiber Substances 0.000 description 2
- 230000000644 propagated effect Effects 0.000 description 2
- 238000012552 review Methods 0.000 description 2
- 206010063385 Intellectualisation Diseases 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 230000036544 posture Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 238000013519 translation 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
- G06V30/32—Digital ink
-
- 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
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
- G06F2218/04—Denoising
- G06F2218/06—Denoising by applying a scale-space analysis, e.g. using wavelet analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
-
- 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
Abstract
The invention provides a method, a system, equipment and a medium for identifying handwritten texts, wherein the method comprises the following steps: acquiring a motion signal of a handwriting action in the writing process of a user, wherein the motion signal comprises a time sequence signal of at least one motion parameter; searching a character recognition model corresponding to a user, wherein the input of the character recognition model is a motion signal in the character handwriting process, and the output of the character recognition model is a recognized character; inputting a motion signal of the handwriting action into a character recognition model to obtain an output character recognition result; and obtaining and recording a text corresponding to the motion signal of the handwriting action according to the character recognition result. The invention provides a scheme for automatic handwritten text recognition and entry, which performs character recognition according to a motion signal in the process of handwriting characters, has high character recognition accuracy compared with the method of performing character recognition only according to track coordinates, and can customize a character recognition model according to the writing habit of each user, thereby further improving the accuracy of character recognition.
Description
Technical Field
The present invention relates to the field of text recognition technologies, and in particular, to a method, a system, a device, and a medium for recognizing handwritten text.
Background
With the development of the science and technology level, informatization and intellectualization are gradually applied to various work fields. Information and data have penetrated into every industry and business function field today, and become important production factors. In some cases, a large amount of handwritten information needs to be timely recorded into a computer and other terminal systems for data sorting and statistical analysis in the next stage.
For example, in the entire vehicle audio workflow, it is necessary to write a defect code to a corresponding defect position on the vehicle body and record the defect information into a computer. At the present stage, the defect information recording mode is mainly divided into two steps: recording with paper pen on site, and manually inputting into computer. The indoor static audio operation process can be briefly summarized as follows: firstly, measuring or observing a vehicle body or a certain part at a special evaluation station, searching whether a certain defect exists or not, and evaluating the grade of the defect; for each defect, after the defect is found, a special pen is needed to mark a defect code (generally, the defect code is an English letter, and the defect grade is an English letter or an English letter-digit combination) at the defect position on the vehicle body or the workpiece, meanwhile, a common pen is used to record information such as the defect position, the defect code, the defect grade and the like in a portable notebook computer, and after the recording is completed, the next position or the next part is continuously checked. And after all defects are checked, returning to the computer in the office area, and inputting each piece of defect information into the system one by using a keyboard according to a defect list recorded on the notebook by handwriting. Because the missing item type that needs the inspection among the whole car audio process is as many as hundreds of thousands of items, and is various, above mode of operation can lead to (1) defect information input process consuming time longer, influences the review efficiency, and (2) when artifical the entering, there is the entry mistake risk of difficult perception, influences the review result.
For the application scenes of the type, the automatic character input technology can greatly improve the input efficiency and reduce the error risk. Technologies available at present for automatic text entry mainly include Optical Character Recognition (OCR), voice recognition technology, and some automatic entry systems based on a stylus.
The OCR technology is to perform optical scanning on characters on paper and perform character recognition through an artificial intelligence technology. However, the OCR technology does not fully utilize stroke sequence information, and has relatively poor recognition accuracy for handwritten characters, and can achieve better recognition accuracy only for print characters. Meanwhile, an extra scanning step is required in the OCR operation process, and under the condition that the recognition accuracy cannot meet the requirement, the effect of improving the input efficiency cannot be achieved, and even extra workload is increased.
The voice recognition technology is used for dictating information to be input by a user, acquiring voice signals through a microphone, performing voice recognition by using an artificial intelligence technology, and converting the voice signals into characters. However, speech recognition is very susceptible to environmental noise and speaker accent, and in a factory production workshop and other noisy environments, the recognition accuracy is often poor, and practical application is difficult.
With the development of sensor technology and the wide application of intelligent terminals, some automatic entry systems based on a stylus are put into use in succession. The technology collects or calculates the movement track of the pen point through optics, pressure, a movement sensor and the like, and stores the handwritten content as electronic version information by tracking and recording the pen point track. However, in this type of system, most handwriting is only recorded in the form of pictures, and the handwriting content is not subjected to character recognition, and cannot be used for subsequent work such as data analysis and operation; in a small number of automatic input systems with a character recognition function, a character recognition algorithm is matched with handwritten fonts in a database only according to pen point track coordinates and an empirical mapping relation so as to achieve the recognition purpose. Considering that writing habits and handwriting of different users have certain differences, the matching algorithm is simple, identification model parameters are not customized for different users, the identification accuracy is poor, and the recording efficiency is low instead because wrong character identification results need to be repeatedly corrected in the using process.
Disclosure of Invention
Aiming at the problems in the prior art, the invention aims to provide a handwritten text recognition method, a system, equipment and a medium based on deep learning, which are convenient to operate, free from noise and other factors and high in recognition accuracy.
The embodiment of the invention provides a handwritten text recognition method, which comprises the following steps:
acquiring a motion signal of a handwriting action in the writing process of a user, wherein the motion signal comprises a time sequence signal of at least one motion parameter;
searching a character recognition model corresponding to a user, wherein the input of the character recognition model is a motion signal in the character handwriting process, and the output of the character recognition model is a recognized character;
inputting the motion signal of the handwriting action into the character recognition model to obtain a character recognition result output by the character recognition model;
and obtaining and recording a text corresponding to the motion signal of the handwriting action according to the character recognition result.
Optionally, the method further includes training a character recognition model corresponding to the user by using the following steps:
collecting a motion signal of a user in the character handwriting process, carrying out character marking on the motion signal, and forming each training sample in a training set by the marked motion signal;
and training the character recognition model by adopting the training set to obtain and store the character recognition model corresponding to the user.
Optionally, the motion signal comprises a time series signal of motion parameters in at least one specific direction;
after the labeled motion signals are formed into each training sample in the training set, the method further comprises the following steps:
and respectively turning or rotating the motion signal of each training sample along each specific direction by a certain angle to obtain different training samples and adding the different training samples into the training set.
Optionally, after the labeled motion signals are configured into each training sample in the training set, the method further includes preprocessing the motion signals of each training sample by using the following steps:
integrating the motion signals of each training sample into a fixed length N to obtain an M multiplied by N motion signal matrix, wherein M is the number of motion parameters;
and carrying out integral normalization on the motion signal of each training sample, so that the motion signal is in accordance with normal distribution.
Optionally, the training of the character recognition model using the training set includes the following steps:
encoding the motion signal of the training sample by using an encoder;
carrying out feature extraction on the coded motion signal to obtain a feature vector of a training sample;
and training a classifier by using the feature vectors of the training samples.
Optionally, after obtaining and recording the text corresponding to the motion signal according to the character recognition result, the method further includes the following steps:
receiving feedback information of a user on a recorded text, if the feedback information comprises characters which are identified incorrectly and corrected, marking a motion signal corresponding to the characters according to the feedback information, and adding the marked motion signal into the training set;
and retraining the character recognition model corresponding to the user by adopting the updated training set and storing the character recognition model.
Optionally, after obtaining the motion signal of the handwriting action in the writing process of the user, the method further includes segmenting the motion signal of the handwriting action to obtain a motion signal segment corresponding to each character;
and inputting the motion signal of the handwriting action into the character recognition model, wherein the step of inputting the motion signal segment corresponding to each character into the character recognition model respectively to obtain the character recognition result of each character, and integrating the character recognition result into the text corresponding to the motion signal.
Optionally, the segmenting the motion signal of the handwriting motion includes:
carrying out integral normalization on the motion signals of the handwriting action to enable the motion signals to be in accordance with normal distribution;
decomposing the motion signal into a multi-level scale space and a wavelet space by adopting a wavelet decomposition algorithm to obtain a decomposed wavelet coefficient and wavelet energy distribution;
according to the frequency information of the motion signal on the multilevel scale after wavelet decomposition, a spectrum discontinuity point detection algorithm is applied to divide the motion signal into a plurality of motion signal segments;
and clustering the motion signal segments by adopting a Gaussian mixture model algorithm, and clustering the motion signal segments belonging to the same character to obtain the motion signal segment corresponding to each character.
Optionally, the obtaining a motion signal of a handwriting action in a writing process of the user includes obtaining the motion signal of the handwriting action from an accessory arranged on the stylus pen.
Optionally, before searching for the character recognition model corresponding to the user, determining the recognition information of the user according to the recognition information of the accessory of the stylus pen.
The embodiment of the invention also provides a handwritten text recognition system which is applied to the handwritten text recognition method, and the system comprises the following components:
the signal acquisition module is used for acquiring a motion signal of a handwriting action in the writing process of a user, wherein the motion signal comprises a time sequence signal of at least one motion parameter;
the model searching module is used for searching a character recognition model corresponding to a user, wherein the character recognition model is input as a motion signal in the character handwriting process and output as a recognized character;
the character recognition module is used for inputting the motion signal of the handwriting action into the character recognition model to obtain a character recognition result output by the character recognition model;
and the text recording module is used for obtaining and recording the text corresponding to the motion signal of the handwriting action according to the character recognition result.
An embodiment of the present invention further provides a handwritten text recognition apparatus, including:
a processor;
a memory having stored therein executable instructions of the processor;
wherein the processor is configured to perform the steps of the handwritten text recognition method via execution of the executable instructions.
The embodiment of the invention also provides a computer readable storage medium for storing a program, and the program realizes the steps of the handwritten text recognition method when being executed.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure.
The handwritten text recognition method, the system, the equipment and the medium provided by the invention have the following advantages:
the invention solves the problems in the prior art, provides a scheme for automatic handwritten text recognition and entry, obtains the motion signal in the process of handwriting characters according to the motion mode of a pen in the writing process, and performs character recognition according to the motion signal of each character, thereby improving the accuracy of character recognition compared with the character recognition only according to the track coordinate; furthermore, the recognition model parameters adopted by the invention can be customized by learning the writing habits and handwriting characteristics of the users, so that the character recognition model is customized according to the writing habits of each user, and the character recognition accuracy is further improved.
Drawings
Other features, objects and advantages of the present invention will become more apparent upon reading of the following detailed description of non-limiting embodiments thereof, with reference to the accompanying drawings.
FIG. 1 is a flow chart of a method of handwritten text recognition in one embodiment of the invention;
FIG. 2 is a flow diagram of a training encoder according to an embodiment of the present invention;
FIG. 3 is a flow chart of a method of handwritten text recognition in accordance with an embodiment of the present invention;
FIG. 4 is a flow chart of motion signal segmentation according to an embodiment of the present invention;
FIG. 5 is a flow diagram of character recognition according to an embodiment of the present invention;
FIG. 6 is a block diagram of a handwritten text recognition system in accordance with an embodiment of the invention;
FIG. 7 is a schematic diagram of a handwritten text recognition device in accordance with an embodiment of the invention;
fig. 8 is a schematic diagram of a computer-readable storage medium according to an embodiment of the present invention.
Detailed Description
Example embodiments will now be described more fully with reference to the accompanying drawings. Example embodiments may, however, be embodied in many different forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of example embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.
Furthermore, the drawings are merely schematic illustrations of the present disclosure and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and thus their repetitive description will be omitted. Some of the block diagrams shown in the figures are functional entities and do not necessarily correspond to physically or logically separate entities. These functional entities may be implemented in the form of software, or in one or more hardware modules or integrated circuits, or in different networks and/or processor devices and/or microcontroller devices.
As shown in fig. 1, in order to solve the technical problems of the prior art, the present invention provides a handwritten text recognition method, which includes the following steps:
s100: acquiring a motion signal of a handwriting action in the writing process of a user, wherein the motion signal comprises a time sequence signal of at least one motion parameter;
s200: searching a character recognition model corresponding to a user, wherein the input of the character recognition model is a motion signal in the character handwriting process, and the output of the character recognition model is a recognized character;
s300: inputting the motion signal of the handwriting action into the character recognition model to obtain a character recognition result output by the character recognition model;
s400: and obtaining and recording a text corresponding to the motion signal of the handwriting action according to the character recognition result, wherein the text can comprise one or more characters.
Therefore, the handwritten text recognition method of the present invention obtains the motion signal in the process of handwriting the character according to the motion mode of the pen in the writing process through step S100, and then performs character recognition according to the motion signal of each character according to step S300, which improves the accuracy of character recognition compared with performing character recognition only according to the trajectory coordinate, and then automatically records the recognized text according to step S400. Further, the invention adopts the character recognition model customized according to different writing habits of each user through the step S200, and the recognition model parameters can be customized by learning the writing habits and handwriting characteristics of the users, thereby further improving the accuracy of character recognition according to the different writing habits of each user.
In this embodiment, the step S100 of acquiring the motion signal of the handwriting action during the writing process of the user may include acquiring the motion signal of the handwriting action from an accessory provided to the stylus pen. For example in the form of a cap or other accessory (pen casing, ornament, etc.) for the stylus in place. A corresponding motion sensor can be arranged in the fitting. In this embodiment, the motion sensor may include a six-axis motion sensor, a gyroscope, or the like, for collecting the motion signal of the stylus. In the writing process of a user, the motion sensor measures motion information such as displacement and rotation of the handwriting pen, transmits a motion signal to the computer terminal through Bluetooth or WIFI, and then the computer terminal analyzes and identifies the motion signal by adopting the handwritten text identification method, and automatically writes the identified text into a document.
Further, before searching for the character recognition model corresponding to the user, determining the recognition information of the user according to the recognition information of the accessory of the stylus pen may be further included. That is, the mapping relationship between the identification information of the accessory of the stylus pen and the identification information of the user is stored in advance. When the user needs to use the handwriting pen, the power switch on the accessory of the handwriting pen is turned on, and the motion sensor of the accessory starts to work. When the user writes, the motion signal is transmitted to the computer terminal, and the computer terminal can search the identification information of the user according to the identification information of the accessory, determine the identity of the user, and search the character recognition model corresponding to the identification information of the user in the step S200. Here, the identification information of the accessory may include an ID of the accessory, the identification information of the user may include an ID of the user, and the like, but the present invention is not limited thereto. The method of acquiring the identification information of the user is not limited to this method, and the user may log in the computer and match the accessories before formal use.
In other alternative embodiments, the motion sensor is not limited to be disposed in an accessory of the pen, and other devices may be used, for example, a smart band worn by the user, and the like, and the corresponding motion signal may be detected along with the handwriting action of the user, which all fall within the protection scope of the present invention.
As shown in fig. 2, in this embodiment, the method for recognizing a handwritten text further includes training a character recognition model corresponding to a user by using the following steps:
(1) collecting a motion signal of handwriting action of a user, carrying out character marking on the motion signal, and forming each training sample in a training set by the marked motion signal;
because the fonts and writing modes of different users are different, the font of a specific user is identified by the training system at this stage, and the accuracy rate of identifying the written characters of the user is improved. The concrete form can be as follows: computer software randomly displays 26 English letters and 10 Arabic numerals (suitable for recognizing letters and numerals, if the Chinese characters need to be recognized, a plurality of common Chinese characters can be randomly displayed), each character appears at least 2 times, a user uses a handwriting pen and writes the characters one by one according to the prompt of the computer, and the computer collects the motion signals of the pen in the writing process of the user as a training set. The content randomly displayed by the computer is related to the application field of the handwritten text recognition method, and some characters commonly used in the application field, including numbers, letters, Chinese characters and the like, can be displayed generally.
In this embodiment, the motion signal includes 9 channels, which respectively represent the changes of the movement acceleration, the rotational acceleration, and the attitude angle of the handwriting motion in three directions of the X axis, the Y axis, and the Z axis with time, as 9-dimensional time-series signals. The three directions of the X-axis, Y-axis and Z-axis here may be three directions in a pre-constructed accessory coordinate system, or three directions of a pre-constructed stylus coordinate system.
(2) Preprocessing the training set data such as denoising and normalization. The specific pretreatment comprises the following steps:
in the process of collecting the training set, since there may be a difference in writing speed of the user and a difference in time sequence length corresponding to each character in the training set, after filtering each channel to eliminate noise influence, integrating the motion signal of each training sample into a fixed length N, for example, N equals to 128, by an interpolation algorithm, to obtain an mxn motion signal matrix, where M is the number of motion parameters, and for the 9-dimensional time sequence signal, for example, into a 9 × 128 motion signal matrix;
and carrying out integral normalization on the motion signal of each training sample, so that the motion signal is in accordance with normal distribution.
(3) After the training set is processed, the training sample can be further expanded, so that different pen holding postures and writing angles of a user can be applied in the character recognition process, and certain robustness such as translation, rotation invariance and the like can be realized. Specifically, in this embodiment, the motion signal of each training sample is respectively turned or rotated by a certain angle along each specific direction to obtain different training samples, and the different training samples are added into the training set. The specific directions may include the above-mentioned X-axis, Y-axis, and Z-axis directions, i.e., the motion signal in each training sample is flipped and rotated by different angles along the X-axis, Y-axis, and Z-axis directions, respectively, and added to the training set.
(4) And training the character recognition model by adopting the training set to obtain and store the character recognition model corresponding to the user. In this embodiment, the character recognition model is trained using the following steps:
encoding the motion signal of the training sample by using an encoder;
carrying out feature extraction on the coded motion signal to obtain a feature vector of a training sample;
and training a classifier by using the feature vectors of the training samples.
In the present invention, as an example, a convolutional auto-Encoders (CAE) in machine learning is applied to encode the original data (in other embodiments, other models or methods may be selected). By training the encoder, the initial feature extraction can be performed on the motion signal of the handwriting action, and the signal dimension is reduced, for example, the original 9 × 128 motion signal matrix is encoded to be 9 × 9 dimensions, and the reduction of the data dimension is beneficial to the training of the classifier at the next stage. Here, the encoder performs preliminary feature extraction on the motion signal, and then inputs the motion signal into the classifier, and the classifier performs feature extraction on the motion signal again to obtain deeper features. The feature extraction result is represented as a fixed-length (e.g., 64-dimensional) feature vector. The feature extraction algorithm may use a Convolutional Neural Network (CNN), a residual network (ResNet), a sparse coding algorithm, and so on. And training a classifier according to the extracted feature vectors so as to obtain a character recognition model, wherein the character recognition model can comprise the encoder and the classifier. The classifier can adopt a support vector machine, a Bayesian classifier, a multi-layer perceptron and the like.
As shown in fig. 3, in this embodiment, after obtaining and recording the text corresponding to the motion signal according to the character recognition result, the method further includes the following steps:
after the text is input, the user can manually correct a small number of wrongly recognized characters in the text at the computer. After the user modification is completed, the character with the recognition error can be added into the training set as new data, and the encoder and the classifier are trained again by using the new data.
Specifically, after receiving feedback information of a user on a recorded text, if the text is correct, automatically entering a document, if the feedback information comprises characters which are identified incorrectly and corrected, marking a motion signal corresponding to the characters according to the feedback information, and adding the marked motion signal into the training set;
and retraining the character recognition model corresponding to the user by adopting the updated training set and storing the character recognition model. Here, the character recognition model may be updated at intervals, or when the number of samples in the error sample set reaches a certain number. Therefore, the character recognition model can be continuously optimized in the using process, and the recognition accuracy is improved.
When the user uses the device, the user generally writes a plurality of characters continuously, so that the signal received by the computer end is a motion signal in the process of writing the plurality of characters by the stylus pen, the motion signal is divided according to different characters, and the motion signal of each character is respectively encoded and recognized in the next stage. Specifically, the step S100: after obtaining the motion signal of the handwriting action in the writing process of the user, the method also comprises the step of dividing the motion signal of the handwriting action to obtain a motion signal segment corresponding to each character.
In the step S300, inputting the motion signal of the handwriting action to the character recognition model, including inputting the motion signal segment corresponding to each character into the character recognition model respectively, to obtain a character recognition result of each character, and then continuing to step S400: and integrating the text corresponding to the motion signal according to the character recognition result and recording the text.
As shown in fig. 4, in this embodiment, a Wavelet Transform (WT) algorithm is used to perform multi-scale decomposition on a motion signal of a handwriting action, detect a breakpoint occurring in a writing process according to information such as a Wavelet coefficient and a Wavelet energy value and complete pre-segmentation, and finally, a Gaussian Mixture Model (GMM) is used to implement signal segmentation, so as to obtain motion signals of a pen when writing different characters respectively. Specifically, in this embodiment, the segmenting the motion signal of the handwriting motion includes the following steps:
firstly, filtering each channel to eliminate noise influence, and then integrally normalizing the motion signals of the handwriting action to ensure that the motion signals conform to normal distribution;
decomposing the motion signal into a multi-level scale space and a wavelet space by adopting a wavelet decomposition algorithm to obtain information such as decomposed wavelet coefficients and wavelet energy distribution;
according to the frequency information of the motion signal on the multilevel scale after wavelet decomposition, a spectrum discontinuity point detection algorithm is applied to divide the motion signal into a plurality of motion signal segments;
and clustering the plurality of pre-divided motion signal segments by adopting a Gaussian mixture model algorithm, clustering the motion signal segments belonging to the same character to obtain a motion signal segment corresponding to each character, and finally realizing the division of motion signals of different characters.
As shown in fig. 5, a flow chart of character recognition according to this embodiment is adopted to perform character recognition by using a character recognition model including an encoder and a classifier trained according to fig. 2 and 3. Firstly, encoding collected motion signals by using an encoder, performing primary feature extraction, then performing feature extraction again by using a classifier to obtain feature vectors, and classifying by using the classifier according to the feature vectors to obtain output character recognition results.
The handwritten text recognition method can be applied to automatic recognition and entry of defect codes in the whole vehicle audit process, and solves the problem of complicated information entry steps. The application scenario of the present invention is not limited to the above scenario. Taking the application in the finished automobile auditing process as an example, when an inspector finds a defect, a defect code is marked on an automobile body, at the moment, the system collects the motion information (movement and rotation) of the pen, analyzes the motion mode of the handwriting pen by using a related artificial intelligence algorithm, thereby identifying the defect code written by the inspector on the automobile body and automatically recording the defect code into the system. The method is applied, so that the steps of paper pen recording and computer input in the operation flow at the current stage are omitted, the workload of registering defect information by an audit inspector is reduced, and the audit work efficiency is improved.
As shown in fig. 6, an embodiment of the present invention further provides a handwritten text recognition system, which is applied to the handwritten text recognition method, and the system includes:
the signal acquisition module M100 is used for acquiring a motion signal of a handwriting action in the writing process of a user, wherein the motion signal comprises a time sequence signal of at least one motion parameter;
the model searching module M200 is used for searching a character recognition model corresponding to a user, wherein the character recognition model is input as a motion signal in the character handwriting process and output as a recognized character;
the character recognition module M300 is configured to input the motion signal of the handwriting motion to the character recognition model, so as to obtain a character recognition result output by the character recognition model;
and the text recording module M400 is used for obtaining and recording a text corresponding to the motion signal of the handwriting action according to the character recognition result.
Therefore, the handwritten text recognition system of the present invention obtains the motion signal of the handwritten character in the process of writing according to the motion mode of the pen in the process of writing through the signal acquisition module M100, and then performs character recognition according to the motion signal of each character according to the character recognition module M300, which improves the accuracy of character recognition compared with performing character recognition only according to the trajectory coordinate, and then automatically records the recognized text according to step S400. Furthermore, the invention adopts the character recognition model customized according to different writing habits of each user through the model searching module M200, and the recognition model parameters can be customized by learning the writing habits and handwriting characteristics of the users, thereby further improving the accuracy of character recognition according to the different writing habits of each user.
In this embodiment, the functions of the modules may be implemented by implementing the steps in the handwritten text recognition method, for example, the signal acquisition module M100 may adopt a specific implementation of step S100, the model search module M200 may adopt a specific implementation of step S200, the character recognition module M300 may adopt a specific implementation of step S300, and the text recording module M400 may adopt a specific implementation of step S400, which is not described herein again.
In this embodiment, the handwritten text recognition system may further include a model training module, configured to train a character recognition model corresponding to the user, specifically, the model training module may train by using the user font learning method shown in fig. 2 and fig. 3, as described above, and may optimize the character recognition model by using the classifier optimization method shown in fig. 3.
The embodiment of the invention also provides handwritten text recognition equipment, which comprises a processor; a memory having stored therein executable instructions of the processor; wherein the processor is configured to perform the steps of the handwritten text recognition method via execution of the executable instructions.
As will be appreciated by one skilled in the art, aspects of the present invention may be embodied as a system, method or program product. Thus, various aspects of the invention may be embodied in the form of: an entirely hardware embodiment, an entirely software embodiment (including firmware, microcode, etc.) or an embodiment combining hardware and software aspects that may all generally be referred to herein as a "circuit," module "or" platform.
An electronic device 600 according to this embodiment of the invention is described below with reference to fig. 7. The electronic device 600 shown in fig. 7 is only an example, and should not bring any limitation to the functions and the scope of use of the embodiments of the present invention.
As shown in fig. 7, the electronic device 600 is embodied in the form of a general purpose computing device. The combination of the electronic device 600 may include, but is not limited to: at least one processing unit 610, at least one memory unit 620, a bus 630 connecting different platform combinations (including memory unit 620 and processing unit 610), a display unit 640, etc.
Wherein the storage unit stores program code executable by the processing unit 610 to cause the processing unit 610 to perform steps according to various exemplary embodiments of the present invention described in the above-mentioned electronic prescription flow processing method section of the present specification. For example, the processing unit 610 may perform the steps as shown in fig. 1.
The storage unit 620 may include readable media in the form of volatile memory units, such as a random access memory unit (RAM)6201 and/or a cache memory unit 6202, and may further include a read-only memory unit (ROM) 6203.
The memory unit 620 may also include a program/utility 6204 having a set (at least one) of program modules 6205, such program modules 6205 including, but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which, or some combination thereof, may comprise an implementation of a network environment.
The electronic device 600 may also communicate with one or more external devices 700 (e.g., keyboard, pointing device, bluetooth device, etc.), with one or more devices that enable a user to interact with the electronic device 600, and/or with any devices (e.g., router, modem, etc.) that enable the electronic device 600 to communicate with one or more other computing devices. Such communication may occur via an input/output (I/O) interface 650. Also, the electronic device 600 may communicate with one or more networks (e.g., a Local Area Network (LAN), a Wide Area Network (WAN), and/or a public network such as the Internet) via the network adapter 660. The network adapter 660 may communicate with other modules of the electronic device 600 via the bus 630. It should be appreciated that although not shown in the figures, other hardware and/or software modules may be used in conjunction with the electronic device 600, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage platforms, to name a few.
The embodiment of the invention also provides a computer readable storage medium for storing a program, and the program realizes the steps of the handwritten text recognition method when being executed. In some possible embodiments, aspects of the present invention may also be implemented in the form of a program product comprising program code for causing a terminal device to perform the steps according to various exemplary embodiments of the present invention described in the above-mentioned electronic prescription flow processing method section of this specification, when the program product is run on the terminal device.
Referring to fig. 8, a program product 800 for implementing the above method according to an embodiment of the present invention is described, which may employ a portable compact disc read only memory (CD-ROM) and include program code, and may be run on a terminal device, such as a personal computer. However, the program product of the present invention is not limited in this regard and, in the present document, a readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.
The program product may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium include: an electrical connection having one or more wires, a portable disk, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
The computer readable storage medium may include a propagated data signal with readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A readable storage medium may also be any readable medium that is not a readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a readable storage medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.
Program code for carrying out operations for aspects of the present invention may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computing device (e.g., through the internet using an internet service provider).
In summary, compared with the prior art, the handwritten text recognition method, system, device and medium provided by the present invention have the following advantages:
the invention solves the problems in the prior art, provides a scheme for automatic handwritten text recognition and entry, obtains the motion signal in the process of handwriting characters according to the motion mode of a pen in the writing process, and performs character recognition according to the motion signal of each character, thereby improving the accuracy of character recognition compared with the character recognition only according to the track coordinate; furthermore, the recognition model parameters adopted by the invention can be customized by learning the writing habits and handwriting characteristics of the users, so that the character recognition model is customized according to the writing habits of each user, and the character recognition accuracy is further improved.
The foregoing is a more detailed description of the invention in connection with specific preferred embodiments and it is not intended that the invention be limited to these specific details. For those skilled in the art to which the invention pertains, several simple deductions or substitutions can be made without departing from the spirit of the invention, and all shall be considered as belonging to the protection scope of the invention.
Claims (13)
1. A method for handwritten text recognition, said method comprising the steps of:
acquiring a motion signal of a handwriting action in the writing process of a user, wherein the motion signal comprises a time sequence signal of at least one motion parameter;
searching a character recognition model corresponding to a user, wherein the input of the character recognition model is a motion signal in the character handwriting process, and the output of the character recognition model is a recognized character;
inputting the motion signal of the handwriting action into the character recognition model to obtain a character recognition result output by the character recognition model;
and obtaining and recording a text corresponding to the motion signal of the handwriting action according to the character recognition result.
2. The method of claim 1, further comprising training a character recognition model corresponding to the user by:
collecting a motion signal of a user in the character handwriting process, carrying out character marking on the motion signal, and forming each training sample in a training set by the marked motion signal;
and training the character recognition model by adopting the training set to obtain and store the character recognition model corresponding to the user.
3. The handwritten text recognition method according to claim 2, characterized in that said motion signal comprises a time series signal of motion parameters in at least one specific direction;
after the labeled motion signals are formed into each training sample in the training set, the method further comprises the following steps:
and respectively turning or rotating the motion signal of each training sample along each specific direction by a certain angle to obtain different training samples and adding the different training samples into the training set.
4. The method of claim 2, wherein after constructing the labeled motion signals into the training samples in the training set, the method further comprises preprocessing the motion signals of the training samples by:
integrating the motion signals of each training sample into a fixed length N to obtain an M multiplied by N motion signal matrix, wherein M is the number of motion parameters;
and carrying out integral normalization on the motion signal of each training sample, so that the motion signal is in accordance with normal distribution.
5. The method of claim 2, wherein training a character recognition model using the training set comprises the steps of:
encoding the motion signal of the training sample by using an encoder;
carrying out feature extraction on the coded motion signal to obtain a feature vector of a training sample;
and training a classifier by using the feature vectors of the training samples.
6. The method of claim 2, wherein after obtaining and recording the text corresponding to the motion signal according to the character recognition result, the method further comprises the following steps:
receiving feedback information of a user on a recorded text, if the feedback information comprises characters which are identified incorrectly and corrected, marking a motion signal corresponding to the characters according to the feedback information, and adding the marked motion signal into the training set;
and retraining the character recognition model corresponding to the user by adopting the updated training set and storing the character recognition model.
7. The method according to claim 1, wherein after obtaining the motion signal of the handwriting action in the writing process of the user, the method further comprises segmenting the motion signal of the handwriting action to obtain a motion signal segment corresponding to each character;
and inputting the motion signal of the handwriting action into the character recognition model, wherein the step of inputting the motion signal segment corresponding to each character into the character recognition model respectively to obtain the character recognition result of each character, and integrating the character recognition result into the text corresponding to the motion signal.
8. The method of claim 7, wherein the step of segmenting the motion signal of the handwritten gesture comprises the steps of:
carrying out integral normalization on the motion signals of the handwriting action to enable the motion signals to be in accordance with normal distribution;
decomposing the motion signal into a multi-level scale space and a wavelet space by adopting a wavelet decomposition algorithm to obtain a decomposed wavelet coefficient and wavelet energy distribution;
according to the frequency information of the motion signal on the multilevel scale after wavelet decomposition, a spectrum discontinuity point detection algorithm is applied to divide the motion signal into a plurality of motion signal segments;
and clustering the motion signal segments by adopting a Gaussian mixture model algorithm, and clustering the motion signal segments belonging to the same character to obtain the motion signal segment corresponding to each character.
9. The method of claim 1, wherein the obtaining the motion signal of the handwriting action during the writing process of the user comprises obtaining the motion signal of the handwriting action from an accessory arranged on a handwriting pen.
10. The method of claim 9, wherein before searching for the character recognition model corresponding to the user, determining the identification information of the user according to the identification information of the accessory of the stylus.
11. A handwritten text recognition system, applied to the handwritten text recognition method according to any one of claims 1 to 10, said system comprising:
the signal acquisition module is used for acquiring a motion signal of a handwriting action in the writing process of a user, wherein the motion signal comprises a time sequence signal of at least one motion parameter;
the model searching module is used for searching a character recognition model corresponding to a user, wherein the character recognition model is input as a motion signal in the character handwriting process and output as a recognized character;
the character recognition module is used for inputting the motion signal of the handwriting action into the character recognition model to obtain a character recognition result output by the character recognition model;
and the text recording module is used for obtaining and recording the text corresponding to the motion signal of the handwriting action according to the character recognition result.
12. A handwritten text recognition device, comprising:
a processor;
a memory having stored therein executable instructions of the processor;
wherein the processor is configured to perform the steps of the handwritten text recognition method of any of claims 1 to 10 via execution of the executable instructions.
13. A computer-readable storage medium storing a program, wherein the program when executed implements the steps of the handwritten text recognition method of any of claims 1 to 10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911122436.7A CN110866499B (en) | 2019-11-15 | 2019-11-15 | Handwritten text recognition method, system, device and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911122436.7A CN110866499B (en) | 2019-11-15 | 2019-11-15 | Handwritten text recognition method, system, device and medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110866499A true CN110866499A (en) | 2020-03-06 |
| CN110866499B CN110866499B (en) | 2022-12-13 |
Family
ID=69653762
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911122436.7A Active CN110866499B (en) | 2019-11-15 | 2019-11-15 | Handwritten text recognition method, system, device and medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110866499B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112364782A (en) * | 2020-11-13 | 2021-02-12 | 深圳振科智能科技有限公司 | Method, device, equipment and medium for identifying high-altitude writing |
| CN113128486A (en) * | 2021-03-31 | 2021-07-16 | 河北师范大学 | Construction method and device of handwritten mathematical formula sample library and terminal equipment |
| CN113176830A (en) * | 2021-04-30 | 2021-07-27 | 北京百度网讯科技有限公司 | Recognition model training method, recognition device, electronic equipment and storage medium |
| CN113723571A (en) * | 2021-10-13 | 2021-11-30 | 湖州师范学院 | Method for judging pen holding posture in dot matrix pen writing |
Citations (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1571979A (en) * | 2001-10-15 | 2005-01-26 | 西尔弗布鲁克研究有限公司 | A method and apparatus for decoding handwritten characters |
| CN101178629A (en) * | 2006-11-07 | 2008-05-14 | 汉王科技股份有限公司 | Character cutting method |
| CN101390107A (en) * | 2006-03-01 | 2009-03-18 | Zi德库玛股份公司 | A method for additive character recognition and an apparatus thereof |
| CN101491101A (en) * | 2006-07-18 | 2009-07-22 | 汤姆森许可贸易公司 | Methods and apparatus for adaptive reference filtering |
| CN103593673A (en) * | 2013-10-27 | 2014-02-19 | 沈阳建筑大学 | On-line handwritten signature authentication method based on dynamic threshold |
| US20140136970A1 (en) * | 2011-07-14 | 2014-05-15 | Tencent Technology (Shenzhen) Company Limited | Text inputting method, apparatus and system |
| CN104205126A (en) * | 2012-03-23 | 2014-12-10 | 微软公司 | Rotation-free recognition of handwritten characters |
| CN104899594A (en) * | 2014-03-06 | 2015-09-09 | 中国科学院沈阳自动化研究所 | Hand action identification method based on surface electromyography decomposition |
| CN105550643A (en) * | 2015-12-08 | 2016-05-04 | 小米科技有限责任公司 | Medical term recognition method and device |
| CN105919584A (en) * | 2016-06-23 | 2016-09-07 | 电子科技大学 | Heart rate estimation method and device for wearable heart rate monitoring equipment |
| CN106339104A (en) * | 2016-08-24 | 2017-01-18 | 广州市香港科大霍英东研究院 | Text input method and device for smart watch |
| CN106408038A (en) * | 2016-09-09 | 2017-02-15 | 华南理工大学 | Rotary Chinese character identifying method based on convolution neural network model |
| CN106774839A (en) * | 2016-11-23 | 2017-05-31 | 童迎伟 | A kind of intelligence wearing key board unit and its input method |
| CN107316067A (en) * | 2017-05-27 | 2017-11-03 | 华南理工大学 | A kind of aerial hand-written character recognition method based on inertial sensor |
| CN108805194A (en) * | 2018-06-04 | 2018-11-13 | 上海交通大学 | A kind of hand-written recognition method and system based on WIFI channel state informations |
| CN109086654A (en) * | 2018-06-04 | 2018-12-25 | 平安科技（深圳）有限公司 | Handwriting model training method, text recognition method, device, equipment and medium |
| CN110135414A (en) * | 2019-05-16 | 2019-08-16 | 京北方信息技术股份有限公司 | Corpus update method, device, storage medium and terminal |
| US20190318148A1 (en) * | 2016-02-16 | 2019-10-17 | Secuve Co., Ltd. | System and method of authenticating handwritten signature based on dynamic movement tracking of spatial-division segments |
-
2019
- 2019-11-15 CN CN201911122436.7A patent/CN110866499B/en active Active
Patent Citations (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1571979A (en) * | 2001-10-15 | 2005-01-26 | 西尔弗布鲁克研究有限公司 | A method and apparatus for decoding handwritten characters |
| CN101390107A (en) * | 2006-03-01 | 2009-03-18 | Zi德库玛股份公司 | A method for additive character recognition and an apparatus thereof |
| CN101491101A (en) * | 2006-07-18 | 2009-07-22 | 汤姆森许可贸易公司 | Methods and apparatus for adaptive reference filtering |
| CN101178629A (en) * | 2006-11-07 | 2008-05-14 | 汉王科技股份有限公司 | Character cutting method |
| US20140136970A1 (en) * | 2011-07-14 | 2014-05-15 | Tencent Technology (Shenzhen) Company Limited | Text inputting method, apparatus and system |
| CN104205126A (en) * | 2012-03-23 | 2014-12-10 | 微软公司 | Rotation-free recognition of handwritten characters |
| CN103593673A (en) * | 2013-10-27 | 2014-02-19 | 沈阳建筑大学 | On-line handwritten signature authentication method based on dynamic threshold |
| CN104899594A (en) * | 2014-03-06 | 2015-09-09 | 中国科学院沈阳自动化研究所 | Hand action identification method based on surface electromyography decomposition |
| CN105550643A (en) * | 2015-12-08 | 2016-05-04 | 小米科技有限责任公司 | Medical term recognition method and device |
| US20190318148A1 (en) * | 2016-02-16 | 2019-10-17 | Secuve Co., Ltd. | System and method of authenticating handwritten signature based on dynamic movement tracking of spatial-division segments |
| CN105919584A (en) * | 2016-06-23 | 2016-09-07 | 电子科技大学 | Heart rate estimation method and device for wearable heart rate monitoring equipment |
| CN106339104A (en) * | 2016-08-24 | 2017-01-18 | 广州市香港科大霍英东研究院 | Text input method and device for smart watch |
| CN106408038A (en) * | 2016-09-09 | 2017-02-15 | 华南理工大学 | Rotary Chinese character identifying method based on convolution neural network model |
| CN106774839A (en) * | 2016-11-23 | 2017-05-31 | 童迎伟 | A kind of intelligence wearing key board unit and its input method |
| CN107316067A (en) * | 2017-05-27 | 2017-11-03 | 华南理工大学 | A kind of aerial hand-written character recognition method based on inertial sensor |
| CN108805194A (en) * | 2018-06-04 | 2018-11-13 | 上海交通大学 | A kind of hand-written recognition method and system based on WIFI channel state informations |
| CN109086654A (en) * | 2018-06-04 | 2018-12-25 | 平安科技（深圳）有限公司 | Handwriting model training method, text recognition method, device, equipment and medium |
| CN110135414A (en) * | 2019-05-16 | 2019-08-16 | 京北方信息技术股份有限公司 | Corpus update method, device, storage medium and terminal |
Non-Patent Citations (4)
| Title |
|---|
| FEI YI等: "Dual Model Medical Invoices Recognition", 《SENSORS》 * |
| SHASHIDHAR PATIL等: "Handwriting Recognition in Free Space Using WIMU-Based Hand Motion Analysis", 《JOURNAL OF SENSORS》 * |
| 张亚磊等: "基于置信测度的GMM模型适配算法在音乐分割中的应用", 《多媒体学术会议》 * |
| 郑能恒等: "基于模型在线更新和平滑处理的音乐分割算法", 《深圳大学学报(理工版)》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112364782A (en) * | 2020-11-13 | 2021-02-12 | 深圳振科智能科技有限公司 | Method, device, equipment and medium for identifying high-altitude writing |
| CN113128486A (en) * | 2021-03-31 | 2021-07-16 | 河北师范大学 | Construction method and device of handwritten mathematical formula sample library and terminal equipment |
| CN113176830A (en) * | 2021-04-30 | 2021-07-27 | 北京百度网讯科技有限公司 | Recognition model training method, recognition device, electronic equipment and storage medium |
| CN113723571A (en) * | 2021-10-13 | 2021-11-30 | 湖州师范学院 | Method for judging pen holding posture in dot matrix pen writing |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110866499B (en) | 2022-12-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110866499B (en) | Handwritten text recognition method, system, device and medium | |
| CN110046350B (en) | Grammar error recognition method, device, computer equipment and storage medium | |
| US9524440B2 (en) | System and method for superimposed handwriting recognition technology | |
| CA2969593C (en) | Method for text recognition and computer program product | |
| US10007859B2 (en) | System and method for superimposed handwriting recognition technology | |
| Garain et al. | Recognition of online handwritten mathematical expressions | |
| US9904847B2 (en) | System for recognizing multiple object input and method and product for same | |
| US20080008387A1 (en) | Method and apparatus for recognition of handwritten symbols | |
| CN111125317A (en) | Model training, classification, system, device and medium for conversational text classification | |
| CN110929802A (en) | Information entropy-based subdivision identification model training and image identification method and device | |
| CN112149663A (en) | RPA and AI combined image character extraction method and device and electronic equipment | |
| CN113657098B (en) | Text error correction method, device, equipment and storage medium | |
| CN110414622B (en) | Classifier training method and device based on semi-supervised learning | |
| CN113486178B (en) | Text recognition model training method, text recognition method, device and medium | |
| CN111178254A (en) | Signature identification method and device | |
| Sundaram et al. | Bigram language models and reevaluation strategy for improved recognition of online handwritten Tamil words | |
| CN113407719B (en) | Text data detection method and device, electronic equipment and storage medium | |
| CN115373982A (en) | Test report analysis method, device, equipment and medium based on artificial intelligence | |
| CN110059180B (en) | Article author identity recognition and evaluation model training method and device and storage medium | |
| Urala et al. | Recognition of open vocabulary, online handwritten pages in Tamil script | |
| CN115346223A (en) | Method and device for evaluating written information, electronic equipment and storage medium | |
| CN111160245A (en) | Dynamic signature identification method and device | |
| EP3295292B1 (en) | System and method for superimposed handwriting recognition technology | |
| Zhang et al. | An Auto-Grading Oriented Approach for Off-Line Handwritten Organic Cyclic Compound Structure Formulas Recognition. | |
| Ma | Machine compatible script for fast text entry of text using handwriting |
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 | ||
| PP01 | Preservation of patent right |
Effective date of registration: 20230131 Granted publication date: 20221213 |
|
| PP01 | Preservation of patent right | ||
| PD01 | Discharge of preservation of patent |
Date of cancellation: 20240108 Granted publication date: 20221213 |
|
| PD01 | Discharge of preservation of patent | ||
| PP01 | Preservation of patent right |
Effective date of registration: 20240227 Granted publication date: 20221213 |
|
| PP01 | Preservation of patent right |