CN114237484A - Handwriting input recognition method and device, electronic equipment and medium - Google Patents
Handwriting input recognition method and device, electronic equipment and medium Download PDFInfo
- Publication number
- CN114237484A CN114237484A CN202010942894.1A CN202010942894A CN114237484A CN 114237484 A CN114237484 A CN 114237484A CN 202010942894 A CN202010942894 A CN 202010942894A CN 114237484 A CN114237484 A CN 114237484A
- Authority
- CN
- China
- Prior art keywords
- word
- radical
- frequency
- stroke data
- mapping
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 51
- 238000013507 mapping Methods 0.000 claims abstract description 77
- 238000003062 neural network model Methods 0.000 claims abstract description 56
- 238000012549 training Methods 0.000 claims description 48
- 238000005516 engineering process Methods 0.000 claims description 18
- 238000012545 processing Methods 0.000 claims description 16
- 230000000295 complement effect Effects 0.000 description 17
- 238000010586 diagram Methods 0.000 description 14
- 238000004891 communication Methods 0.000 description 11
- 230000008569 process Effects 0.000 description 7
- 238000000605 extraction Methods 0.000 description 6
- 238000012360 testing method Methods 0.000 description 6
- 238000012795 verification Methods 0.000 description 6
- 238000004590 computer program Methods 0.000 description 5
- 238000012986 modification Methods 0.000 description 5
- 230000004048 modification Effects 0.000 description 5
- 230000006870 function Effects 0.000 description 4
- 230000003287 optical effect Effects 0.000 description 4
- 230000005236 sound signal Effects 0.000 description 4
- 230000008859 change Effects 0.000 description 3
- 230000003993 interaction Effects 0.000 description 3
- 230000001133 acceleration Effects 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 230000002085 persistent effect Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 239000000779 smoke Substances 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000001052 transient effect Effects 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
- G06F3/048—Interaction techniques based on graphical user interfaces [GUI]
- G06F3/0487—Interaction techniques based on graphical user interfaces [GUI] using specific features provided by the input device, e.g. functions controlled by the rotation of a mouse with dual sensing arrangements, or of the nature of the input device, e.g. tap gestures based on pressure sensed by a digitiser
- G06F3/0488—Interaction techniques based on graphical user interfaces [GUI] using specific features provided by the input device, e.g. functions controlled by the rotation of a mouse with dual sensing arrangements, or of the nature of the input device, e.g. tap gestures based on pressure sensed by a digitiser using a touch-screen or digitiser, e.g. input of commands through traced gestures
- G06F3/04883—Interaction techniques based on graphical user interfaces [GUI] using specific features provided by the input device, e.g. functions controlled by the rotation of a mouse with dual sensing arrangements, or of the nature of the input device, e.g. tap gestures based on pressure sensed by a digitiser using a touch-screen or digitiser, e.g. input of commands through traced gestures for inputting data by handwriting, e.g. gesture or text
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
- G06F3/02—Input arrangements using manually operated switches, e.g. using keyboards or dials
- G06F3/023—Arrangements for converting discrete items of information into a coded form, e.g. arrangements for interpreting keyboard generated codes as alphanumeric codes, operand codes or instruction codes
- G06F3/0233—Character input methods
- G06F3/0236—Character input methods using selection techniques to select from displayed items
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Abstract
The embodiment of the invention discloses a handwriting input recognition method, which comprises the steps of acquiring stroke data of handwriting input of a user in real time; identifying the stroke data acquired in real time by using a radical model to obtain an identification result; if the recognition result represents that the stroke data corresponds to the target radical, acquiring an associated word set corresponding to the target radical according to a pre-established radical mapping word table; performing completion prediction on the stroke data according to a deep neural network model to obtain a completion word set corresponding to the stroke data; determining candidate characters to be displayed according to the associated character set and the completion character set, and displaying the candidate characters; therefore, the associated characters and the completion character set are fused, so that the matching degree of the displayed candidate characters and the characters which the user wants to input is higher, and the accuracy of candidate characters which are predicted and recommended by handwriting input can be improved.
Description
Technical Field
The embodiment of the invention relates to the technical field of handwriting input, in particular to a handwriting input recognition method, a handwriting input recognition device, electronic equipment and a handwriting input medium.
Background
With the continuous development of science and technology, portable touch devices are rapidly developed and popularized, man-machine interaction becomes more and more frequent, and application scenes of handwriting input are more and more, for example, in the aspects of signatures of various electronic files, character input and the like.
In the prior art, when handwriting is input on a touch device, because each character usually has a plurality of strokes, the handwriting input speed is relatively low, and in order to solve the problem of low handwriting input speed, prediction recommendation is usually performed according to the stroke sequence input by a user, but handwriting written by the user may be written in one stroke without one stroke of sequence information, and an inverted pen phenomenon may exist in the writing process of the user, such as left and right structural fonts, writing the right half part first and then writing the left half part, and after the stroke sequence is completely disordered, the recommendation based on the stroke sequence is invalid, so that the problem of relatively low accuracy of candidate characters for prediction recommendation of handwriting input in the prior art is solved.
Disclosure of Invention
The embodiment of the invention provides a method, a device, electronic equipment and a medium for identifying handwriting input, which can improve the accuracy of candidate characters for predicting and recommending the handwriting input.
The first aspect of the embodiments of the present invention provides a method for recognizing a handwriting input, including:
acquiring stroke data input by a user through handwriting in real time;
identifying the stroke data acquired in real time by using a radical model to obtain an identification result;
if the recognition result represents that the stroke data corresponds to the target radical, acquiring an associated word set corresponding to the target radical according to a pre-established radical mapping word table; performing completion prediction on the stroke data according to a deep neural network model to obtain a completion word set corresponding to the stroke data; and determining candidate characters to be displayed according to the associated character set and the completion character set, and displaying the candidate characters.
Optionally, the creating step of the radical mapping word table includes:
acquiring a high-frequency word list and a non-high-frequency word list according to historical input data, wherein the frequency of each word in the written high-frequency word list is not less than a first preset frequency, and the frequency of each word in the non-high-frequency word list is not less than the first preset frequency;
and creating the radical mapping word table according to the high-frequency word table, the non-high-frequency word table and the radical.
Optionally, after the stroke data acquired in real time is recognized by using the radical model to obtain a recognition result, the method further includes: if the recognition result represents that the stroke data corresponds to the non-radical, performing completion prediction on the stroke data according to a deep neural network model to obtain a completion word set corresponding to the stroke data; and determining candidate words to be displayed according to the completion word set, and displaying the candidate words.
Optionally, the deep neural network model is created based on DropStroke technology.
Optionally, the training sample set of the deep neural network model includes a complete word set and an incomplete word set, wherein for each word in the complete word set, the strokes constituting the word are complete strokes, and for each word in the incomplete word set, the strokes constituting the word lack at least one stroke compared to the complete strokes of the word.
Optionally, the obtaining of the associated word set corresponding to the target radical according to the pre-established radical mapping word table includes:
acquiring a mapping word set of which the radical is the target radical from the radical mapping word table;
and acquiring the associated word set from the mapping word set according to a second preset frequency, wherein the frequency of the associated word set is greater than the second preset frequency.
A second aspect of an embodiment of the present invention provides a handwriting input recognition apparatus, including:
the stroke data acquisition module is used for acquiring stroke data input by a user through handwriting in real time;
the stroke identification module is used for identifying the stroke data acquired in real time by using the component model to obtain an identification result;
the display module is used for acquiring an associated character set corresponding to the target radical according to a pre-established radical mapping character table if the identification result represents that the stroke data corresponds to the target radical; performing completion prediction on the stroke data according to a deep neural network model to obtain a completion word set corresponding to the stroke data; determining candidate characters to be displayed according to the associated character set and the completion character set, and displaying the candidate characters;
optionally, the apparatus further comprises:
the system comprises a radical mapping table creating module, a radical mapping table creating module and a radical mapping table creating module, wherein the radical mapping table creating module is used for acquiring a high-frequency word table and a non-high-frequency word table according to historical input data, the frequency of each word in the written high-frequency word table is not less than a first preset frequency, and the frequency of each word in the non-high-frequency word table is not less than the first preset frequency; and creating the radical mapping word table according to the high-frequency word table, the non-high-frequency word table and the radical.
Optionally, if the recognition result represents that the stroke data corresponds to a non-radical, the display module is configured to perform completion prediction on the stroke data according to a deep neural network model to obtain a completion word set corresponding to the stroke data; and determining candidate words to be displayed according to the completion word set, and displaying the candidate words.
Optionally, the deep neural network model is created based on DropStroke technology.
Optionally, the training sample set of the deep neural network model includes a complete word set and an incomplete word set, wherein for each word in the complete word set, the strokes constituting the word are complete strokes, and for each word in the incomplete word set, the strokes constituting the word lack at least one stroke compared to the complete strokes of the word.
Optionally, the display module further includes:
an associated word set obtaining unit, configured to obtain, from the radical mapping word table, a mapping word set in which a radical is the target radical; and acquiring the associated word set from the mapping word set according to a second preset frequency, wherein the frequency of the associated word set is greater than the second preset frequency.
A third aspect of embodiments of the present invention provides an apparatus for data processing, comprising a memory, and one or more programs, wherein the one or more programs are stored in the memory, and the one or more programs configured to be executed by the one or more processors comprise the steps of the recognition method for handwriting input as described above.
A fourth aspect of embodiments of the present invention provides a machine-readable medium having stored thereon instructions, which, when executed by one or more processors, cause an apparatus to perform a method of recognition of a handwritten input as described above.
The embodiment of the invention has the following beneficial effects:
based on the technical scheme, stroke data input by a user through handwriting are acquired in real time; identifying the stroke data acquired in real time by using a radical model, and acquiring an associated character set corresponding to a target radical from a pre-established radical mapping character table if the target radical corresponding to the stroke data is identified; performing completion prediction on the stroke data according to a deep neural network model to obtain a completion word set corresponding to the stroke data; determining candidate characters to be displayed according to the associated character set and the completion character set, and displaying the candidate characters; therefore, when the stroke data input in real time is identified to correspond to the target radicals, the associated word set can be predicted in advance through the radical mapping word table, namely the prediction speed of the associated word set is higher; the stroke data is subjected to completion prediction, so that the matching degree of the obtained completion word set and the user is higher, namely the accuracy of the predicted completion word set is higher; on the basis, determining candidate words based on the associated word set and the completion word set, so that the determined candidate words improve the prediction speed on the basis of ensuring the accuracy; furthermore, the user can find the words or phrases which the user wants to input at a higher speed, and the input efficiency is improved.
Drawings
FIG. 1 is a first flowchart of a method for recognizing a handwriting input according to an embodiment of the present invention;
FIG. 2 is a second flowchart of a method for recognizing handwriting input according to an embodiment of the present invention;
FIG. 3 is a flowchart illustrating the overall steps of a handwriting recognition method according to an embodiment of the present invention;
FIG. 4 is a schematic structural diagram of a recognition apparatus for handwriting input according to an embodiment of the present invention;
FIG. 5 is a block diagram of a handwriting input recognition apparatus as a device according to an embodiment of the present invention;
fig. 6 is a block diagram of a server in some embodiments of the inventions.
Detailed Description
In order to better understand the technical solutions, the technical solutions of the embodiments of the present invention are described in detail below with reference to the drawings and the specific embodiments, and it should be understood that the specific features of the embodiments and the embodiments of the present invention are detailed descriptions of the technical solutions of the embodiments of the present invention, and are not limitations of the technical solutions of the present invention, and the technical features of the embodiments and the embodiments of the present invention may be combined with each other without conflict.
Aiming at the technical problem that the accuracy of candidate words predicted and recommended by handwriting input is low, the embodiment of the invention provides a handwriting input recognition scheme, which is used for recognizing stroke data of the handwriting input in real time and specifically comprises the following steps: acquiring stroke data input by a user through handwriting in real time; identifying the stroke data acquired in real time by using a radical model to obtain an identification result; if the recognition result represents that the stroke data corresponds to the target radical, acquiring an associated character set corresponding to the target radical according to a pre-established radical mapping character table; performing completion prediction on the stroke data according to the deep neural network model to obtain a completion word set corresponding to the stroke data; and determining candidate characters to be displayed according to the associated character set and the completion character set, and displaying the candidate characters.
In the embodiment of the invention, the stroke data can be stroke data of Chinese characters, in the actual process, some Chinese characters have components and some Chinese characters do not have components, for example, Chinese character ' text ' does not have components, Chinese character ' smoke ' has components and the components are fire '.
In the embodiment of the invention, after the stroke data are acquired in real time, the stroke data acquired in real time are identified by using the radical model, so that the radical model needs to be trained before the stroke data acquired in real time are identified.
When the component model is trained, the training sample set may include, for example, stroke data of each component in a chinese character and stroke data of a character corresponding to each component. At this time, the training sample set contains 13100 samples in total, and the training sample set can be randomly divided into 3 groups: 12500 samples for training, 300 samples for validation and 300 samples for testing. Then 12500 samples are used for model training, and after the training is finished, 300 verification samples are used for verifying the trained model; after the verification is met, the model is tested by using 300 test samples, and if the test meets the test condition, a radical model is obtained.
If the verification of the model does not meet the verification requirement, training the model again by using the training sample until the trained model meets the verification requirement; and testing the model which is verified to meet the requirements until the trained model meets the verification requirements and the testing conditions, and taking the trained model as a final model, namely the radical model.
And when a radical model of the output radicals is trained, inputting the stroke characteristics of each training sample into the radical model for each training sample in the training sample set, and outputting the radicals of the training samples. Thus, after the radical model is trained, the stroke data acquired in real time can be recognized by using the radical model, and whether the stroke data corresponds to the recognition result of the target radical can be recognized.
If the recognition result represents that the stroke data corresponds to the target radical, acquiring an associated character set corresponding to the target radical according to a pre-established radical mapping character table; performing completion prediction on the stroke data according to the deep neural network model to obtain a completion word set corresponding to the stroke data; and determining candidate characters to be displayed according to the associated character set and the completion character set, and displaying the candidate characters.
Specifically, when stroke data acquired in real time is recognized using the radical model, if the stroke features of the stroke data are recognized by the radical model and the similarity of the stroke features to a certain radical is greater than the preset similarity, the target radical for which the stroke data is recognized is determined to be a radical having a similarity greater than the preset similarity.
In the embodiment of the present invention, the preset similarity may be set manually or by a device, or may be set according to actual requirements, and the preset similarity is usually set to a value not less than 60%, for example, 65%, 75%, 95%, or the like; of course, the preset similarity may also be set to a value less than 60%, and the present invention is not particularly limited.
For example, stroke data written by a user in a writing box of a smart phone is acquired in real time, then the character radical model is used for identifying the stroke data acquired in real time, and if the similarity between the stroke features of the stroke data and the character radicals of 'wide' is greater than the preset similarity, the character radicals of the target character radical corresponding to the stroke data are identified as 'wide'.
In the embodiment of the invention, before acquiring the associated character set corresponding to the target radical according to the pre-established radical mapping character table, the radical mapping character table is required to be pre-established; after creating the radical mapping word table, a search is performed from the radical mapping word table using the target radical as a keyword to obtain an associated word set corresponding to the target radical.
The creating steps of the radical mapping word table in the embodiment of the invention are as follows:
a1, acquiring a high-frequency word list and a non-high-frequency word list according to historical input data, wherein the frequency of writing each word in the high-frequency word list is not less than a first preset frequency, and the frequency of each word in the non-high-frequency word list is not less than the first preset frequency;
step A2, creating a radical mapping word table according to the high frequency word table, the non-high frequency word table and the radical.
In step a1, the historical input data is first obtained, and at this time, a high frequency word table and a non-high frequency word table may be created from the historical input data by mining a large amount of user input data as the historical input data.
Specifically, the historical input data may include input data input by a user in a search engine, query data input by the user in an application, such as query data input by the user on an e-commerce platform, document data input by the user, such as input characters in blogs and forums, and the like, so that a large amount of historical input data is collected in various ways, and thus, the accuracy of the created radical mapping word list is higher in the case of collecting a large amount of historical input data.
In the embodiment of the present invention, the first preset frequency is determined according to an average input frequency of words in the history input data, wherein a value range of the first preset frequency is between an unequaled input frequency and a highest input frequency of the words in the history input data, for example, if the average input frequency is p1 and the highest input frequency is p2, a frequency p3 is taken as the first preset frequency, wherein p1> p3> p 2.
For example, taking words in the history input data including "yell", "library", "text", "summer", as an example, if the frequency of "yell" and "summer" is less than p3, the "yell" and "summer" are added to the non-high frequency word list; if the frequency of the 'library' and 'text' is not less than p3, the 'library' and 'text' are added to the high frequency word list.
After creating the high frequency word table and the non-high frequency word table, a radical mapping word table is created based on the radical of each of the two word tables, the high frequency word table and the non-high frequency word table. Of course, the radical mapping word table may be created only according to the radicals of each word in the high frequency word table, and the present invention is not particularly limited.
In the embodiment of the invention, before the stroke data is completely predicted according to the deep neural network model, the deep neural network model needs to be trained; and after the deep neural network model is trained, performing complementary prediction on the stroke data by using the deep neural network model.
Specifically, the deep neural network model may be a VGG model, a google lenet model, a ResNet model, an inclusion-ResNet-v 2 model, or the like, and the present invention is not particularly limited.
Specifically, in the model training process, the training samples of the deep neural network model may be the number of samples with complete notes, in this case, the training sample set of the deep neural network model includes a complete word set, and the strokes of each word in the complete word set are complete strokes
And after the training sample set is obtained, firstly, performing model training on the data in the training sample set generally comprises the steps of data preprocessing, feature extraction, recognition and the like.
The method comprises the steps of extracting features, wherein the features in eight directions of each training sample in a training sample set are usually extracted in the feature extraction step, and specifically, an 8-dimensional feature vector is obtained by calculating the projection of each point on a writing track in 8 directions; of course, the deep neural network model may also adopt other feature extraction methods, such as Path Signature, to extract stroke features of the training samples; and after the stroke features of the training samples are obtained, recognizing according to the stroke features.
Therefore, when the stroke data input in real time is identified to correspond to the target radicals, the associated character set can be predicted in advance through the radical mapping character table, namely the prediction speed of the associated character set is higher; the stroke data is subjected to completion prediction, so that the matching degree of the obtained completion word set and the user is higher; on the basis, determining candidate words based on the associated word set and the completion word set, so that the determined candidate words improve the prediction speed on the basis of ensuring the accuracy; furthermore, the user can find the words or phrases which the user wants to input at a higher speed, and the input efficiency is improved.
Method embodiment one
Referring to fig. 1, a flowchart illustrating steps of a first embodiment of a method for recognizing a handwriting input according to the present invention is shown, which may specifically include the following steps:
s101, acquiring stroke data input by a user through handwriting in real time;
s102, identifying the stroke data acquired in real time by using a radical model to obtain an identification result;
s103, if the identification result represents that the stroke data corresponds to the target radical, acquiring an associated word set corresponding to the target radical according to a pre-established radical mapping word table; performing completion prediction on the stroke data according to the deep neural network model to obtain a completion word set corresponding to the stroke data; and determining candidate characters to be displayed according to the associated character set and the completion character set, and displaying the candidate characters.
The identification method in the embodiment of the present invention may be applied to a client and may also be applied to a server, and the present invention is not particularly limited.
The client can be an electronic device with or externally connected with a touch device, such as a smart phone, a tablet personal computer, a notebook computer, a smart watch and the like, and the touch device can be a touch pad, a touch screen and the like; the server can be a desktop computer, a notebook computer, a tablet computer, an all-in-one machine, a smart phone and the like.
In step S101, when the recognition method in the embodiment of the present invention is applied to a client, the client executes steps S101 to S103, and at this time, the client may obtain stroke data input by a user through handwriting in real time, specifically, obtain the stroke data input by the user through a touch device on the client or connected to the client.
When the identification method in the embodiment of the present invention is applied to a server, the client also obtains the stroke data handwritten by the user in real time, and then uploads the obtained stroke data to the server in real time, or the server reads the stroke data obtained by the client from the client in real time, so that the server can obtain the stroke data in real time, thereby executing steps S101 to S103.
Further, the client obtains the stroke data input by the user through handwriting in real time, and the stroke data may be input by the user through a finger or an electronic pen, and the invention is not limited in particular.
After the stroke data is obtained in real time in step S101, step S102 is executed, in which a radical model is obtained by training, and the training process is described in detail with reference to the above-mentioned training procedure for the radical model.
Step S102, identifying stroke data acquired in real time according to a radical model obtained by pre-training to obtain an identification result; if the recognition result represents that the stroke data corresponds to the target radical, executing step S103; if the recognition result represents that the stroke data corresponds to the non-radical, executing step S104, and performing completion prediction on the stroke data according to the deep neural network model to obtain a completion word set corresponding to the stroke data; and determining candidate words to be displayed according to the completion word set, and displaying the candidate words.
Specifically, when the stroke data acquired in real time is identified by using the radical model, if the stroke data is identified to correspond to a certain radical, the corresponding certain radical is taken as a target radical; and if the stroke data does not correspond to any one of the radicals in the radical model, determining that the stroke data represented by the recognition result corresponds to the non-radical.
For example, if it is acquired that the stroke data input by a certain user in real time is "death", the stroke data input in real time is recognized through the radical model, and the radical corresponding to the stroke data is recognized as "death", then the target radical is determined as "death", and then step S103 is executed; if the stroke data input by a user in real time is acquired as 'five', and the stroke data is identified not to correspond to any one of the radicals, the stroke data is determined to correspond to the non-radical, and step S104 is executed.
If the stroke data is identified to correspond to the target radical, step S103 is executed. In step S103, a radical mapping word table may be used to obtain a related word set, and then a deep neural network model is used to predict a completion word set; the complete word set can be obtained by predicting by using a deep neural network model, and then the associated word set is obtained by using a radical mapping word table; of course, the deep neural network model and the radical mapping word table may be used together to obtain the complete word set and the associated word set, and the present invention is not limited in particular.
And after the associated character set and the completion character set are obtained, determining candidate characters and displaying the candidate characters according to the associated character set and the completion character set.
Specifically, when the related word set is acquired using the radical mapping word table, a plurality of words with a top frequency order may be searched from the radical mapping word table as the related word set according to the target radical. For example, taking "wide" as an example, if the radicals in the radical mapping word table are "dead" and the frequencies are sequentially ranked from large to small as "dead", "busy", "forget", "blind", "hope", "win", "dell" and "gadget", if the word with the frequency ranking top 4 is taken as the associated word set, the associated word set is determined as dead "," busy "," forget "and" blind ", and if the word with the frequency ranking top 3 is taken as the associated word set, the associated word set is determined as" dead "," busy "and" forget ".
Specifically, when a radical mapping word table is used to obtain a related word set, firstly, a mapping word set with the radical as a target radical is obtained from the radical mapping word table; then acquiring an associated word set from the mapping word set according to a second preset frequency, wherein the frequency of the associated word set is greater than the second preset frequency; the second preset frequency and the first preset frequency may be the same or different.
And when the second preset frequency is the same as the first preset frequency, the selected associated character sets are all high-frequency characters, namely the selected associated character sets are more consistent with the input habit of the user, so that the matching degree of the selected associated character sets and the characters which the user wants to input is higher. And when the second preset frequency is less than the first preset frequency, if the high-frequency word with the radical as the target radical is not found in the radical mapping word table, selecting one or more words from the low-frequency words with the radical as the target radical as an associated word set.
Specifically, after the mapping word set is obtained, one or more words with a frequency greater than a second preset frequency may be randomly selected from the mapping word set as an associated word set; one or more words with a higher rank and a frequency greater than a second preset frequency may also be selected from the mapping word set as the associated word set according to the frequency of the words in the mapping word set, which is not limited in the present invention.
Preferably, when selecting the related word set, in the mapping word set in which the radical is the target radical in the radical mapping word table, the probability that the word with higher frequency is selected as the related word set is higher.
And when the stroke data is subjected to completion prediction according to the deep neural network model to obtain a completion word set, inputting stroke authentication of the stroke data into the deep neural network model, and acquiring one or more output words as the completion word set. In this case, the training data set of the deep neural network model may only include the complete word set, and of course, may also include the complete word set and the incomplete word set, and the present invention is not particularly limited.
For example, also taking "death" as an example, the stroke features of "death" are input into the neural network model, and the complete word set obtained by prediction is sequentially "death", "it", "win", and "blind".
After the associated character set and the completion character set are obtained, the associated character set and the completion character set are fused to determine candidate characters to be displayed; specifically, according to the number of candidate words, a related word set and a completion word set are fused; of course, the associated word set and the complementary word set may be fused according to the frequency, and one or more words with the highest frequency may be selected from the associated word set and the complementary word set as candidate words.
Of course, if the associated word set and the complementary word set contain the same word, the same word contained in the associated word set and the complementary word set is used as a candidate word, and at this time, the matching degree between the candidate word and the word that the user wants to input is made higher.
Specifically, according to the number of the candidate words, at least one associated word is selected from the associated word set and at least one complementary word is selected from the complementary word set to serve as the candidate word, wherein the number of the selected at least one associated word and the selected at least one complementary word is not larger than the number of the candidate words. If the number of the candidate words is n, selecting m words from the associated word set, and selecting (n-m) words from the completion word set as candidate words, where m may be greater than (n-m), and of course, m may also be not greater than (n-m), and the present invention is not limited specifically.
In the embodiment of the invention, the associated word set comprises one or more words, and correspondingly, the completion word set comprises one or more words.
And after the candidate character is obtained, displaying the candidate character on a display screen of the client so as to facilitate the selection of a user.
If the recognition result represents that the stroke data corresponds to the non-radical, executing step S104, and performing completion prediction on the stroke data according to the deep neural network model to obtain a completion word set corresponding to the stroke data; and selecting one or more characters from the completion character set as candidate characters, and displaying the selected candidate characters.
Specifically, when a candidate word is selected from the completion word set, a candidate word can be selected from the completion word set according to the number of candidate words and the number of words in the completion word set; if the number of the characters in the complete character set is not more than the number of the candidate words, all the characters in the complete character set can be used as the candidate characters; if the number of the characters in the completion character set is larger than that of the candidate words, the characters with the top ranking frequency in the completion character set can be selected as the candidate characters. For example, the completion word set includes words a1, a2, a3, and a4, and the number of candidate words is 2, 2 words a1 and a2 with the highest frequency are selected from a1, a2, a3, and a4 as candidate words.
In the embodiment, when the stroke data input in real time is identified to correspond to the target radical, the related word set can be predicted in advance through the radical mapping word table, namely, the prediction speed of the related word set is faster; the stroke data is subjected to completion prediction, so that the matching degree of the obtained completion word set and the user is higher, namely the accuracy of the predicted completion word set is higher; on the basis, determining candidate words based on the associated word set and the completion word set, so that the determined candidate words improve the prediction speed on the basis of ensuring the accuracy; furthermore, the user can find the words or phrases which the user wants to input at a higher speed, and the input efficiency is improved.
Example two
Referring to fig. 2, a flowchart illustrating steps of a second embodiment of a handwriting input recognition method according to the present invention is shown, which may specifically include the following steps:
s201, acquiring stroke data input by a user through handwriting in real time;
s202, identifying the stroke data acquired in real time by using a radical model to obtain an identification result;
s203, if the identification result represents that the stroke data corresponds to the target radical, acquiring an associated word set corresponding to the target radical according to a pre-established radical mapping word table; performing completion prediction on the stroke data according to a deep neural network model to obtain a completion word set corresponding to the stroke data, wherein the deep neural network model is created based on a DropStroke technology; and determining candidate characters to be displayed according to the associated character set and the completion character set, and displaying the candidate characters.
In the embodiment of the invention, before the stroke data is completely predicted according to the deep neural network model, the deep neural network model needs to be trained; and after the deep neural network model is trained, performing complementary prediction on the stroke data by using the deep neural network model.
Specifically, the deep neural network model is created based on DropStroke technology. The DropStroke technique is a method of data enhancement on existing samples, namely, random stroke removal. The deep neural network model may be a VGG model, a google lenet model, a ResNet model, an inclusion-ResNet-v 2 model, and the like, and the present invention is not particularly limited.
Specifically, when the deep neural network model based on the DropStroke technology is used, in the model training process, sample data of partial handwriting which is discarded randomly and sample data of complete handwriting are used for training together, so that the deep neural network model obtained through training has a good completion prediction effect on incomplete fonts, and the completion prediction accuracy can be higher.
Specifically, due to the adoption of the DropStroke technology, a training sample set of the deep neural network model comprises a complete character set and an incomplete character set, wherein for each character in the complete character set, the strokes forming the character are complete strokes; for each word in the incomplete set of words, the strokes making up the word lack at least one stroke as compared to the complete strokes of the word.
Specifically, for each word in the incomplete word set, the strokes that make up the word are obtained by randomly discarding some strokes of the complete strokes of the word. And aiming at each character in the incomplete character set, the strokes forming the word change can comprise continuous strokes, wrongly written strokes and the like, and the wrongly written strokes can write 'left falling' into 'right falling' and the like.
Specifically, DropStroke's method is to discard a part of strokes of a Chinese character randomly without changing the order of the remaining strokes to obtain a new symbol. According to the combination formula, assuming that the number of strokes of a Chinese character is n, and m strokes are discarded, the number of combinations, i.e. the number of new characters, can be obtained as shown in formula 1.
Wherein, the value range of m belongs to [1, n-1] and m belongs to n, so for any character, the total number of new characters and original characters can be generated by DropStroke technology as formula 2, which is specifically as follows:
thus, as can be seen from formula 1 and formula 2, the DropStroke technology can be used to expand the training data set, and generate massive training data; therefore, on the basis of enhancing the training data, the accuracy of the deep neural network model completion prediction obtained by training the training data is higher.
Specifically, after the training data set can be extended by the DropStroke technology, after the training sample set is obtained, firstly, performing model training on the data in the training sample set generally includes steps of data preprocessing, feature extraction, recognition and the like.
The method comprises the steps of extracting features, wherein the features in eight directions of each training sample in a training sample set are usually extracted in the feature extraction step, and specifically, an 8-dimensional feature vector is obtained by calculating the projection of each point on a writing track in 8 directions; of course, the deep neural network model may also adopt other feature extraction methods, such as Path Signature, to extract stroke features of the training samples; and after the stroke features of the training samples are obtained, recognizing according to the stroke features.
Therefore, when the stroke data input in real time is identified to correspond to the target radicals, the associated character set can be predicted in advance through the radical mapping character table, namely the prediction speed of the associated character set is higher; the stroke data is subjected to complement prediction, so that the matching degree of the obtained complement word set and a user is higher, and the accuracy of the predicted complement word set can be higher due to the adoption of the DropStroke technology; on the basis, determining candidate words based on the associated word set and the completion word set, so that the determined candidate words improve the prediction speed on the basis of ensuring the accuracy; furthermore, the user can find the words or phrases which the user wants to input at a higher speed, and the input efficiency is improved.
The overall steps of the recognition method of handwriting input in the embodiment of the invention are specifically shown in fig. 3. Firstly, step 30 is executed to obtain stroke data input by a user in real time; next, executing step 31, using the radical model for identification; next, step 32 is executed to determine whether the stroke data is a radical; if yes, go to step 33 and step 34; step 33, mapping the associated word set, wherein each word in the associated word set is usually a high-frequency word; step 34, complement prediction is carried out, and a deep neural network model is used for complement prediction to obtain a complement word set; step 35, result fusion is performed, that is, the associated character set and the completion character set are fused to obtain candidate words; after the candidate word is obtained, step 36 is executed to display the candidate word.
If the stroke data is determined not to be the radical by the step 32, executing a step 34 of completing prediction, and performing completing prediction by using a deep neural network model to obtain a completing character set; and then selecting candidate words from the completion word set and displaying the candidate words.
In the practical application process, if the character which the user wants to input is 'win', the stroke data input by the user is 'death', and the complementary character set predicted by the deep neural network model is 'death', 'top', 'win' and 'blind' in sequence; and obtaining the associated character sets of 'death', 'busy', 'forgetting', 'blind', 'watching', 'winning', 'dell' and 'gad' in turn through the mapping of the radical mapping character table, thus when the associated character set and the complementary character set are merged, because the 'death', 'winning' and 'blind' are all contained, the candidate character is determined to necessarily contain 'death', 'winning' and 'blind', and the candidate character is displayed, at the moment, the user can select 'win' according to the displayed candidate character without stroke input, at the moment, the character needing the 17 strokes input by the user can only be selected through the input 3 strokes, thus, the efficiency of handwriting input of the user can be effectively improved
Device embodiment
Referring to fig. 4, a block diagram of a recognition apparatus for handwriting input according to an embodiment of the present invention is shown, which may specifically include:
a stroke data obtaining module 401, configured to obtain stroke data input by a user through handwriting in real time;
a stroke recognition module 402, configured to recognize the stroke data obtained in real time by using a radical model, so as to obtain a recognition result;
a display module 403, configured to, if the recognition result represents that the stroke data corresponds to a target radical, obtain an associated word set corresponding to the target radical according to a pre-established radical mapping word table; performing completion prediction on the stroke data according to a deep neural network model to obtain a completion word set corresponding to the stroke data; and determining candidate characters to be displayed according to the associated character set and the completion character set, and displaying the candidate characters.
In an alternative embodiment, the identification means further comprises:
the system comprises a radical mapping table creating module, a radical mapping table creating module and a radical mapping table creating module, wherein the radical mapping table creating module is used for acquiring a high-frequency word table and a non-high-frequency word table according to historical input data, the frequency of each word in the written high-frequency word table is not less than a first preset frequency, and the frequency of each word in the non-high-frequency word table is not less than the first preset frequency; and creating the radical mapping word table according to the high-frequency word table, the non-high-frequency word table and the radical.
In an optional implementation manner, the displaying module 403 is configured to, if the recognition result represents that the stroke data corresponds to a non-radical, perform completion prediction on the stroke data according to a deep neural network model to obtain a completion word set corresponding to the stroke data; and determining candidate words to be displayed according to the completion word set, and displaying the candidate words.
In an alternative embodiment, the deep neural network model is created based on DropStroke technology.
In an alternative embodiment, the set of training samples for the deep neural network model includes a complete set of words and an incomplete set of words, wherein for each word in the complete set of words, the strokes making up the word are complete strokes, and for each word in the incomplete set of words, the strokes making up the word lack at least one stroke compared to the complete strokes of the word.
In an alternative embodiment, the display module 403 further includes:
an associated word set obtaining unit, configured to obtain, from the radical mapping word table, a mapping word set in which a radical is the target radical; and acquiring the associated word set from the mapping word set according to a second preset frequency, wherein the frequency of the associated word set is greater than the second preset frequency.
For the device embodiment, since it is basically similar to the method embodiment, the description is simple, and for the relevant points, refer to the partial description of the method embodiment.
The embodiments of the present invention are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other.
With regard to the apparatus in the above-described embodiment, the specific manner in which each module performs the operation has been described in detail in the embodiment related to the method, and will not be elaborated here.
Fig. 5 is a block diagram illustrating a recognition apparatus for handwriting input as a device according to an exemplary embodiment. For example, the apparatus 900 may be a mobile phone, a computer, a digital broadcast terminal, a messaging device, a game console, a tablet device, a medical device, an exercise device, a personal digital assistant, and the like.
Referring to fig. 5, apparatus 900 may include one or more of the following components: processing component 902, memory 904, power component 906, multimedia component 908, audio component 910, input/output (I/O) interface 912, sensor component 914, and communication component 916.
The processing component 902 generally controls overall operation of the device 900, such as operations associated with display, incoming calls, data communications, camera operations, and recording operations. Processing element 902 may include one or more processors 920 to execute instructions to perform all or a portion of the steps of the methods described above. Further, processing component 902 can include one or more modules that facilitate interaction between processing component 902 and other components. For example, the processing component 902 can include a multimedia module to facilitate interaction between the multimedia component 908 and the processing component 902.
The memory 904 is configured to store various types of data to support operation at the device 900. Examples of such data include instructions for any application or method operating on device 900, contact data, phonebook data, messages, pictures, videos, and so forth. The memory 904 may be implemented by any type or combination of volatile or non-volatile memory devices such as Static Random Access Memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic or optical disks.
The power supply component 906 provides power to the various components of the device 900. The power components 906 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power for the device 900.
The multimedia component 908 comprises a screen providing an output interface between the device 900 and a user. In some embodiments, the screen may include a Liquid Crystal Display (LCD) and a Touch Panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive an input signal from a user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor may not only sense the boundary of a touch or slide motion action, but also detect the duration and pressure associated with the touch or slide operation. In some embodiments, the multimedia component 908 includes a front facing camera and/or a rear facing camera. The front-facing camera and/or the rear-facing camera may receive external multimedia data when the device 900 is in an operating mode, such as a shooting mode or a video mode. Each front camera and rear camera may be a fixed optical lens system or have a focal length and optical zoom capability.
The audio component 910 is configured to output and/or input audio signals. For example, audio component 910 includes a Microphone (MIC) configured to receive external audio signals when apparatus 900 is in an operating mode, such as a call mode, a recording mode, and a voice recognition mode. The received audio signals may further be stored in the memory 904 or transmitted via the communication component 916. In some embodiments, audio component 910 also includes a speaker for outputting audio signals.
I/O interface 912 provides an interface between processing component 902 and peripheral interface modules, which may be keyboards, click wheels, buttons, etc. These buttons may include, but are not limited to: a home button, a volume button, a start button, and a lock button.
The sensor component 914 includes one or more sensors for providing status assessment of various aspects of the apparatus 900. For example, the sensor assembly 914 may detect an open/closed state of the device 900, the relative positioning of the components, such as a display and keypad of the apparatus 900, the sensor assembly 914 may also detect a change in the position of the apparatus 900 or a component of the apparatus 900, the presence or absence of user contact with the apparatus 900, orientation or acceleration/deceleration of the apparatus 900, and a change in the temperature of the apparatus 900. The sensor assembly 914 may include a proximity sensor configured to detect the presence of a nearby object in the absence of any physical contact. The sensor assembly 914 may also include a light sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, the sensor assembly 914 may also include an acceleration sensor, a gyroscope sensor, a magnetic sensor, a pressure sensor, or a temperature sensor.
The communication component 916 is configured to facilitate communications between the apparatus 900 and other devices in a wired or wireless manner. The apparatus 900 may access a wireless network based on a communication standard, such as WiFi, 2G or 3G, or a combination thereof. In an exemplary embodiment, the communication component 916 receives a broadcast signal or broadcast associated information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communications component 916 further includes a Near Field Communication (NFC) module to facilitate short-range communications. For example, the NFC module may be implemented based on Radio Frequency Identification (RFID) technology, infrared data association (IrDA) technology, Ultra Wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.
In an exemplary embodiment, the apparatus 900 may be implemented by one or more Application Specific Integrated Circuits (ASICs), Digital Signal Processors (DSPs), Digital Signal Processing Devices (DSPDs), Programmable Logic Devices (PLDs), Field Programmable Gate Arrays (FPGAs), controllers, micro-controllers, microprocessors or other electronic components for performing the above-described methods.
In an exemplary embodiment, a non-transitory computer readable storage medium comprising instructions, such as the memory 904 comprising instructions, executable by the processor 920 of the apparatus 900 to perform the above-described method is also provided. For example, the non-transitory computer readable storage medium may be a ROM, a Random Access Memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, and the like.
Fig. 6 is a block diagram of a server in some embodiments of the invention. The server 1900 may vary widely by configuration or performance and may include one or more Central Processing Units (CPUs) 1922 (e.g., one or more processors) and memory 1932, one or more storage media 1930 (e.g., one or more mass storage devices) storing applications 1942 or data 1944. Memory 1932 and storage medium 1930 can be, among other things, transient or persistent storage. The program stored in the storage medium 1930 may include one or more modules (not shown), each of which may include a series of instructions operating on a server. Still further, a central processor 1922 may be provided in communication with the storage medium 1930 to execute a series of instruction operations in the storage medium 1930 on the server 1900.
The server 1900 may also include one or more power supplies 1926, one or more wired or wireless network interfaces 1950, one or more input-output interfaces 1958, one or more keyboards 1956, and/or one or more operating systems 1941, such as Windows Server, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM, etc.
A non-transitory computer-readable storage medium in which instructions, when executed by a processor of an apparatus (device or server), enable the apparatus to perform a method of recognition of handwriting input, the method comprising: acquiring stroke data input by a user through handwriting in real time; identifying the stroke data acquired in real time by using a radical model to obtain an identification result; if the recognition result represents that the stroke data corresponds to the target radical, acquiring an associated word set corresponding to the target radical according to a pre-established radical mapping word table; performing completion prediction on the stroke data according to a deep neural network model to obtain a completion word set corresponding to the stroke data; and determining candidate characters to be displayed according to the associated character set and the completion character set, and displaying the candidate characters.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all such alterations and modifications as fall within the scope of the invention.
It will be apparent to those skilled in the art that various changes and modifications may be made in the present invention without departing from the spirit and scope of the invention. Thus, if such modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include such modifications and variations.
Claims (10)
1. A method for recognizing a handwriting input, comprising:
acquiring stroke data input by a user through handwriting in real time;
identifying the stroke data acquired in real time by using a radical model to obtain an identification result;
if the recognition result represents that the stroke data corresponds to the target radical, acquiring an associated word set corresponding to the target radical according to a pre-established radical mapping word table; performing completion prediction on the stroke data according to a deep neural network model to obtain a completion word set corresponding to the stroke data; and determining candidate characters to be displayed according to the associated character set and the completion character set, and displaying the candidate characters.
2. The method of claim 1, wherein the step of creating the table of radical maps comprises:
acquiring a high-frequency word list and a non-high-frequency word list according to historical input data, wherein the frequency of each word in the written high-frequency word list is not less than a first preset frequency, and the frequency of each word in the non-high-frequency word list is not less than the first preset frequency;
and creating the radical mapping word table according to the high-frequency word table, the non-high-frequency word table and the radical.
3. The method as claimed in claim 2, wherein after said recognizing the stroke data acquired in real time using the radical model to obtain a recognition result, the method further comprises:
if the recognition result represents that the stroke data corresponds to the non-radical, performing completion prediction on the stroke data according to a deep neural network model to obtain a completion word set corresponding to the stroke data; and determining candidate words to be displayed according to the completion word set, and displaying the candidate words.
4. The method of claim 3, wherein the deep neural network model is created based on DropStroke technology.
5. The method of claim 4, wherein the set of training samples for the deep neural network model includes a complete set of words and an incomplete set of words, wherein for each word in the complete set of words, the strokes making up the word are complete strokes, and wherein for each word in the incomplete set of words, the strokes making up the word lack at least one stroke as compared to the complete strokes of the word.
6. The method as claimed in claim 1, wherein said obtaining a set of related words corresponding to the target radical from a pre-established radical mapping word table comprises:
acquiring a mapping word set of which the radical is the target radical from the radical mapping word table;
and acquiring the associated word set from the mapping word set according to a second preset frequency, wherein the frequency of the associated word set is greater than the second preset frequency.
7. An apparatus for recognizing a handwriting input, comprising:
the stroke data acquisition module is used for acquiring stroke data input by a user through handwriting in real time;
the stroke identification module is used for identifying the stroke data acquired in real time by using the component model to obtain an identification result;
the display module is used for acquiring an associated character set corresponding to the target radical according to a pre-established radical mapping character table if the identification result represents that the stroke data corresponds to the target radical; performing completion prediction on the stroke data according to a deep neural network model to obtain a completion word set corresponding to the stroke data; and determining candidate characters to be displayed according to the associated character set and the completion character set, and displaying the candidate characters.
8. The apparatus of claim 7, further comprising:
the system comprises a radical mapping table creating module, a radical mapping table creating module and a radical mapping table creating module, wherein the radical mapping table creating module is used for acquiring a high-frequency word table and a non-high-frequency word table according to historical input data, the frequency of each word in the written high-frequency word table is not less than a first preset frequency, and the frequency of each word in the non-high-frequency word table is not less than the first preset frequency; and creating the radical mapping word table according to the high-frequency word table, the non-high-frequency word table and the radical.
9. An apparatus for data processing, comprising a memory, and one or more programs, wherein the one or more programs are stored in the memory, and wherein the one or more programs being configured to be executed by the one or more processors comprises the method steps of any of claims 1-6.
10. A machine-readable medium having stored thereon instructions which, when executed by one or more processors, cause an apparatus to perform a method of recognition of a handwritten input as recited in one or more of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010942894.1A CN114237484A (en) | 2020-09-09 | 2020-09-09 | Handwriting input recognition method and device, electronic equipment and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010942894.1A CN114237484A (en) | 2020-09-09 | 2020-09-09 | Handwriting input recognition method and device, electronic equipment and medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114237484A true CN114237484A (en) | 2022-03-25 |
Family
ID=80742761
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010942894.1A Pending CN114237484A (en) | 2020-09-09 | 2020-09-09 | Handwriting input recognition method and device, electronic equipment and medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114237484A (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101276249A (en) * | 2007-03-30 | 2008-10-01 | 北京三星通信技术研究有限公司 | Method and device for forecasting and discriminating hand-written characters |
| CN102360265A (en) * | 2011-09-29 | 2012-02-22 | 中兴通讯股份有限公司 | Method and device for determining words to be selected in hand writing input |
| CN110968246A (en) * | 2018-09-28 | 2020-04-07 | 北京搜狗科技发展有限公司 | Intelligent Chinese handwriting input recognition method and device |
-
2020
- 2020-09-09 CN CN202010942894.1A patent/CN114237484A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101276249A (en) * | 2007-03-30 | 2008-10-01 | 北京三星通信技术研究有限公司 | Method and device for forecasting and discriminating hand-written characters |
| CN102360265A (en) * | 2011-09-29 | 2012-02-22 | 中兴通讯股份有限公司 | Method and device for determining words to be selected in hand writing input |
| CN110968246A (en) * | 2018-09-28 | 2020-04-07 | 北京搜狗科技发展有限公司 | Intelligent Chinese handwriting input recognition method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3173948A1 (en) | Method and apparatus for recommendation of reference documents | |
| CN108227950B (en) | Input method and device | |
| CN109144285B (en) | Input method and device | |
| CN107291772B (en) | Search access method and device and electronic equipment | |
| CN109918565B (en) | Processing method and device for search data and electronic equipment | |
| CN116166843B (en) | Text video cross-modal retrieval method and device based on fine granularity perception | |
| CN112784142A (en) | Information recommendation method and device | |
| CN107132927B (en) | Input character recognition method and device for recognizing input characters | |
| CN112307281A (en) | Entity recommendation method and device | |
| CN111739535A (en) | Voice recognition method and device and electronic equipment | |
| CN109388249B (en) | Input information processing method and device, terminal and readable storage medium | |
| CN109918624B (en) | Method and device for calculating similarity of webpage texts | |
| CN110968246A (en) | Intelligent Chinese handwriting input recognition method and device | |
| CN111831132A (en) | Information recommendation method and device and electronic equipment | |
| CN107301188B (en) | Method for acquiring user interest and electronic equipment | |
| CN115994266A (en) | Resource recommendation method, device, electronic equipment and storage medium | |
| CN111382598A (en) | Identification method and device and electronic equipment | |
| CN114237484A (en) | Handwriting input recognition method and device, electronic equipment and medium | |
| CN111382367B (en) | Search result ordering method and device | |
| US11403534B2 (en) | Acquiring entity webpage link based on topological relationship graph | |
| CN107977089B (en) | Input method and device and input device | |
| CN112052395A (en) | Data processing method and device | |
| CN109032374B (en) | Candidate display method, device, medium and equipment for input method | |
| CN112214114A (en) | Input method and device and electronic equipment | |
| CN110554780A (en) | sliding input method and device |
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 |