CN113095171A

CN113095171A - Method and device for recognizing written characters, electronic equipment and storage medium

Info

Publication number: CN113095171A
Application number: CN202110333948.9A
Authority: CN
Inventors: 陈科鑫; 曾理; 张晓帆
Original assignee: Guangdong Oppo Mobile Telecommunications Corp Ltd
Current assignee: Guangdong Oppo Mobile Telecommunications Corp Ltd
Priority date: 2021-03-29
Filing date: 2021-03-29
Publication date: 2021-07-09

Abstract

A method and a device for recognizing written characters, electronic equipment and a storage medium are provided, wherein the method comprises the following steps: responding to a writing input operation, and acquiring at least one writing track corresponding to the writing input operation; stroke recognition is carried out on each writing track respectively to obtain stroke recognition information and continuous stroke information of each writing track; performing continuous correction on corresponding stroke identification information according to the continuous stroke information of each writing track to obtain target stroke information corresponding to each writing track; and performing character recognition according to the target stroke information corresponding to each writing track to obtain a character recognition result matched with the at least one writing track. By implementing the embodiment of the application, the accuracy of the recognition of the written characters can be improved.

Description

Method and device for recognizing written characters, electronic equipment and storage medium

Technical Field

The present application relates to the field of information recognition technologies, and in particular, to a method and an apparatus for recognizing written characters, an electronic device, and a storage medium.

Background

Currently, various types of character recognition technologies have been widely used with the development of computer technology. However, in practice, it is found that when characters are written on a non-printed body (such as characters written by bare hands, characters written by a touch pen, characters written by air, etc.), the written characters are easily affected by writers, writing forms, writing media, etc., so that various uncontrollable problems such as non-normalcy, continuous writing, shaking, etc. often exist, and the accuracy of character recognition is greatly reduced.

Disclosure of Invention

The embodiment of the application discloses a method and a device for recognizing written characters, electronic equipment and a storage medium, which can improve the accuracy of written character recognition.

A first aspect of an embodiment of the present application discloses a method for recognizing a written character, including:

responding to a writing input operation, and acquiring at least one writing track corresponding to the writing input operation;

stroke recognition is carried out on each writing track respectively to obtain stroke recognition information and continuous stroke information of each writing track;

performing continuous correction on corresponding stroke identification information according to continuous stroke information of each writing track to obtain target stroke information corresponding to each writing track;

and performing character recognition according to the target stroke information corresponding to each writing track to obtain a character recognition result matched with the at least one writing track.

A second aspect of the embodiments of the present application discloses a written character recognition apparatus, including:

the input acquisition unit is used for responding to writing input operation and acquiring at least one writing track corresponding to the writing input operation;

the stroke recognition unit is used for respectively carrying out stroke recognition on each writing track to obtain stroke recognition information and continuous stroke information of each writing track;

the continuous stroke correction unit is used for carrying out continuous stroke correction on corresponding stroke identification information according to the continuous stroke information of each writing track to obtain target stroke information corresponding to each writing track;

and the character recognition unit is used for performing character recognition according to the target stroke information corresponding to each writing track to obtain a character recognition result matched with the at least one writing track.

A third aspect of the embodiments of the present application discloses an electronic device, which includes a memory and a processor, where the memory stores a computer program, and when the computer program is executed by the processor, the processor is enabled to implement all or part of the steps in any one of the written character recognition methods disclosed in the first aspect of the embodiments of the present application.

A fourth aspect of the embodiments of the present application discloses a computer-readable storage medium, on which a computer program is stored, wherein the computer program, when executed by a processor, implements all or part of the steps of any one of the written character recognition methods disclosed in the first aspect of the embodiments of the present application.

Compared with the related art, the embodiment of the application has the following beneficial effects:

in the embodiment of the application, the electronic device can respond to the detected writing input operation to obtain at least one writing track corresponding to the writing input operation, and perform stroke recognition on each writing track respectively to obtain stroke recognition information and continuous stroke information of each writing track; the electronic equipment can perform continuous correction on corresponding stroke identification information according to continuous stroke information of each writing track to obtain target stroke information corresponding to each writing track; on the basis, the electronic equipment can perform character recognition according to the target stroke information corresponding to each writing track to obtain a character recognition result matched with the at least one writing track. Therefore, by implementing the embodiment of the application, the stroke information of the whole written character can be recognized before being recognized, so that the stroke information which is easy to ignore in the related technology can be fully utilized, the continuous strokes which often appear in the written character are corrected, all the strokes contained in the continuous strokes are accurately recognized, and the accuracy of the recognition of the written character can be effectively improved.

Drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art that other drawings can be obtained according to the drawings without creative efforts.

Fig. 1 is a schematic view of an application scenario of a method for recognizing written characters disclosed in an embodiment of the present application;

FIG. 2 is a schematic diagram of a written character disclosed in an embodiment of the present application;

FIG. 3 is a timing diagram illustrating a method for identifying written characters according to an embodiment of the present disclosure;

FIG. 4 is a flow chart illustrating a method for recognizing written characters according to an embodiment of the present disclosure;

FIG. 5 is a flow chart illustrating another method for recognizing written characters disclosed in the embodiments of the present application;

FIG. 6 is a diagram of a combination of strokes corresponding to a writing trajectory according to an embodiment of the present disclosure;

fig. 7 is a schematic diagram of a stroke interval corresponding to a writing track disclosed in an embodiment of the present application;

FIG. 8 is a flow chart illustrating another method for recognizing written characters according to the embodiment of the present application;

FIG. 9 is a block diagram of a written character recognition device according to an embodiment of the present disclosure;

fig. 10 is a schematic block diagram of an electronic device according to an embodiment of the present disclosure.

Detailed Description

The technical solutions in the embodiments of the present application will be described clearly and completely with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.

It should be noted that the terms "comprises" and "comprising," and any variations thereof, in the embodiments of the present application, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.

The following detailed description is made with reference to the accompanying drawings.

Referring to fig. 1, fig. 1 is a schematic view of an application scenario of a method for recognizing written characters according to an embodiment of the present application, including a user 10 and an electronic device 20. Wherein, the user 10 can perform writing input operation through the electronic device 20, and the electronic device 20 recognizes the written character input by the user 10. In some embodiments, the electronic device 20 may include a touch screen 21 (shown in fig. 1), so that the user 10 may perform a writing input operation by a freehand touch through the touch screen 21. It is understood that in some embodiments, the user 10 may use an auxiliary tool such as a stylus or a touch stick instead of a freehand touch to implement a similar writing input operation on the touch screen 21. It is noted that in some embodiments, the touch screen 21 on the electronic device 20 is not required. For example, the electronic device 20 may include a high-altitude writing track obtaining device (e.g., a camera, a distance sensor, etc.), so that the user 10 may complete the writing input operation for the electronic device 20 in a high-altitude writing manner without contacting the electronic device 20, and obtain a corresponding writing track by the high-altitude writing track obtaining device, so that the electronic device 20 may recognize the written character input by the user 10.

For example, the electronic device 20 may include various devices or systems with handwriting input function, such as a mobile phone, a smart wearable device, an in-vehicle terminal, a tablet Computer, a PC (Personal Computer), a PDA (Personal Digital Assistant), and the like, which are not limited in this embodiment. The handwriting input function can include various writing inputs in a broad sense, such as touch writing, high-altitude writing and the like, and is not only directed at freehand writing. It should be noted that the electronic device 20 shown in fig. 1 is a smartphone, which is merely an example and does not constitute a limitation on the device type of the electronic device 20 in the embodiment of the present application. In some embodiments, the electronic device 20 may also include a writing track obtaining device and a control device, an output device, and the like, where the writing track obtaining device may include a mouse (which may obtain a writing track according to a moving track), a laser pen (which may obtain a writing track according to a track of a laser light speed), and the like, which are not limited herein.

In the embodiment of the present application, when the electronic device 20 detects a writing input operation, at least one writing trace corresponding to the writing input operation may be obtained in response to the writing input operation, where based on differences in writing habits of the user 10 (such as whether to be used in connection with a pen, omitting a stroke, etc.), a detection response policy of the electronic device 20, a self-stroke structure of a written character, etc., the user 10 may perform one or more writing input operations while writing one character, so that the electronic device 20 may correspondingly obtain one or more writing traces corresponding to the writing input operation. As shown in fig. 2, when the user 10 writes a "permanent" word, the user may include a first writing input (shown by a solid line), a second writing input (shown by a dotted line), and a third writing input (shown by a dotted line), where the third writing input is a continuous writing including a plurality of actual strokes, and the electronic device 20 may accordingly obtain three writing traces. Then, the electronic device 20 may perform stroke recognition on each writing trace, and obtain stroke recognition information of each writing trace and corresponding stroke connection information. Then, the electronic device 20 can perform continuous correction on the corresponding stroke recognition information according to the continuous stroke information of each writing track to obtain the target stroke information corresponding to each writing track, so that the stroke information of the written character can be fully utilized to correct the continuous stroke which may occur. On this basis, the electronic device 20 can perform character recognition according to the target stroke information corresponding to each writing track, so as to combine the stroke information and the whole character information of the written character to obtain a character recognition result matched with the at least one writing track, thereby effectively improving the accuracy of the written character recognition.

To better describe an application scenario of the embodiment of the present application, fig. 3 discloses a timing chart of a method for recognizing a written character according to the embodiment of the present application. As shown in fig. 3, the electronic device 20 may include a control device 201, a writing trace obtaining device 202, a storage device 203, and a recognition device 204, and the method may include:

302. the control device detects a writing input operation.

Specifically, the control device 201 may detect a writing input start instruction initiated by the user, and a writing input pause instruction, a writing input stop instruction, and the like, to trigger the writing trajectory acquisition device 202 to acquire the written character input by the user according to the writing input operation of the user.

304. The writing trace obtaining device responds to the writing input operation and obtains at least one writing trace corresponding to the writing input operation.

For example, when writing a character, the user may perform one or more writing input operations, so that the writing trace obtaining device 202 may obtain one or more writing traces corresponding to the writing input operations accordingly.

306. The storage device stores the at least one writing track.

In the embodiment of the present application, the storage device 203 may store any valid data generated during the written character recognition process to ensure the reliability of the written character recognition.

308. The recognition device respectively carries out stroke recognition on each writing track to obtain stroke recognition information and continuous stroke information of each writing track.

In the embodiment of the application, the stroke recognition is performed on the writing track through the recognition device 204, and the stroke information of the written character can be fully utilized, so that continuous strokes frequently occurring in the written character can be corrected in the subsequent steps, and the accuracy of the written character recognition is improved.

310. The storage device stores the stroke recognition information.

312. And the control device performs continuous correction on the corresponding stroke identification information according to the continuous stroke information of each writing track to obtain target stroke information corresponding to each writing track.

Specifically, the control device 201 can perform pen breaking on a writing track with continuous strokes, so as to correct corresponding stroke recognition information and ensure the accuracy and reliability of the recognition of the written characters.

314. The storage device stores the target stroke information.

316. And the recognition device performs character recognition according to the target stroke information corresponding to each writing track to obtain a character recognition result matched with the at least one writing track.

In the embodiment of the present application, the recognition device 204 performs stroke recognition to obtain stroke recognition information, synthesizes the stroke recognition information after continuous stroke correction into a character, and performs character recognition, so that the stroke information and the whole character information of the written character are utilized in a writing character recognition process, and the accuracy of writing character recognition can be effectively improved.

318. The storage device stores the character recognition result.

320. And the control device carries out interactive display according to the character recognition result.

Illustratively, the control device 201 may communicate with a display device built in or externally connected to the electronic device 20 to display a character recognition result of the writing input of the user, so as to provide a real-time character recognition feedback for the user, improve interactivity in the process of recognizing the writing character by using the electronic device 20, and simultaneously help the user to adjust or correct the writing input operation in time, thereby further improving reliability of the writing character recognition.

By implementing the method described in the above embodiment, the electronic device 20 can make full use of the stroke information that is easy to ignore in the related art, and correct the continuous strokes that often appear in the written character, so as to accurately recognize each stroke included in the continuous strokes, and further, can effectively improve the accuracy of the written character recognition.

Referring to fig. 4, fig. 4 is a flow chart illustrating a method for recognizing written characters according to an embodiment of the present application. As shown in fig. 4, the written character recognition method may include the steps of:

402. and responding to the writing input operation, and acquiring at least one writing track corresponding to the writing input operation.

In the embodiment of the application, when the electronic device detects a writing input operation, a built-in writing track acquiring device may be triggered, and the writing track acquiring device responds to the writing input operation to acquire at least one writing track corresponding to the writing input operation. The writing input operation is generally initiated by a user actively, the electronic device can detect a writing input starting instruction initiated by the user, and at least one writing track corresponding to the corresponding writing input operation is acquired through the writing track acquisition device under the triggering of the writing input starting instruction. It can be understood that the writing trajectory acquiring device may distinguish different writing trajectories according to a writing input pause instruction, a writing input stop instruction, and the like initiated by a user, or may distinguish writing trajectories input successively according to a certain writing time interval (e.g., 0.1 second, 0.5 second, and the like). In some embodiments, the electronic device may also continuously monitor the writing input operations directed thereto, so that when a writing input operation is detected, the writing trajectory acquiring device may be triggered in time in response to the writing input operation, and when a writing input stopping feature is detected (e.g., the acquired writing trajectories overlap, exceed a certain number of thresholds, etc.), the electronic device may stop to acquire at least one writing trajectory corresponding to the corresponding writing input operation.

It can be understood that, since the user may perform one or more writing input operations while writing a character, the writing trace obtaining device of the electronic device may also obtain one or more writing traces corresponding to the writing input operations accordingly.

Specifically, the writing trajectory acquisition device may be matched with a manner in which a user performs a writing input operation. In an embodiment, if the writing input operation is touch writing input, the corresponding writing trajectory acquiring device may include a touch screen of the electronic device. Illustratively, the electronic device may acquire a next touch trajectory or trajectories through the touch screen when detecting a writing input start instruction (e.g., clicking a writing input start button, starting a writing input interface, etc.) triggered by the user through the touch screen. When the electronic device detects a writing input stopping instruction triggered by a user through the touch screen, the one or more touch tracks can be used as writing tracks corresponding to writing input operation.

In another embodiment, if the writing input operation is an up-to-the-air writing input operation, the corresponding writing trajectory acquiring device may include a camera, a distance sensor, and the like. For example, when the electronic device detects a writing input start gesture of a user, a writing trajectory acquisition device such as a camera may acquire a next volley gesture and identify a gesture trajectory corresponding to the volley gesture. When the electronic device detects a writing input stopping gesture of the user, or any volitional gesture of the user is not detected within a certain time, the gesture track can be used as a writing track corresponding to the writing input operation.

404. And respectively carrying out stroke recognition on each writing track to obtain stroke recognition information and continuous stroke information of each writing track.

Illustratively, when stroke recognition is performed on a target writing track and stroke recognition information corresponding to a plurality of strokes can be obtained, the possibility of continuous writing of the target writing track can be shown, so that the electronic device can simultaneously obtain corresponding continuous writing information according to the stroke recognition information of each writing track, and further can respectively determine continuous strokes possibly corresponding to each writing track. The stroke identification information may include stroke identification information of each stroke included in the writing track (i.e., identification corresponding to one or more strokes included in the writing track), a stroke interval position (i.e., a position of a broken stroke between each stroke), a stroke combination sequence (i.e., a writing sequence of each stroke in sequence), and the like; the stroke connection information at least includes a stroke connection probability (i.e. a probability that the writing track includes the plurality of strokes), a stroke connection confidence level, a corresponding confidence interval, and the like of the plurality of strokes possibly corresponding to the writing track.

To implement the stroke recognition, in some embodiments, the electronic device may use Re-Identification (ReID) to pass each writing track through a pre-trained matching recognition model to obtain one or more strokes matched with the writing track. Illustratively, the one or more strokes may be distinguished by stroke identification information, or may be distinguished by stroke encoding. In other embodiments, the electronic device may also employ other recognition models, such as a pre-trained unsupervised neural network model.

406. And performing continuous correction on the corresponding stroke identification information according to the continuous stroke information of each writing track to obtain target stroke information corresponding to each writing track.

For example, the stroke recognition information of each writing track may be respectively related to one or more strokes included in each writing track. For any target writing track in each writing track, the electronic device may first obtain stroke identification information and continuous stroke information of the target writing track, and then perform continuous stroke correction on the stroke identification information corresponding to the target writing track according to the continuous stroke information, so as to obtain one or more strokes with the highest possibility included in the target writing track as target stroke information thereof. The target stroke information may include stroke identification information corresponding to the one or more strokes, so that in subsequent steps, a character to be recognized may be synthesized according to the stroke identification information corresponding to each writing track, and character recognition may be further performed to obtain a final written character recognition result.

In one embodiment, when the target writing track only contains one stroke, for example, the stroke identification information of the target writing track only contains one stroke identification information, and the stroke connection information of the target writing track indicates that the target writing track has a low possibility of containing a plurality of strokes (e.g., the stroke connection probability is lower than the probability threshold), the electronic device may directly determine that the target stroke information corresponding to the target writing track corresponds to the stroke, and obtain the corresponding target stroke information.

In another embodiment, when the target writing track may include one stroke or may include a plurality of strokes, for example, the stroke identification information thereof includes a single stroke identification information and a combination of strokes composed of a plurality of stroke identification information, and the running stroke information thereof indicates that the target writing track has a low possibility of including a plurality of strokes (e.g., the running stroke probability is lower than the probability threshold), the electronic device may not perform any modification on the stroke identification information, thereby determining that the target stroke information corresponding to the target writing track corresponds to the one stroke, and obtaining the corresponding target stroke information.

In yet another embodiment, when the stroke connection information of the target writing trajectory indicates that the target writing trajectory has a high possibility of containing a plurality of strokes (e.g., the stroke connection probability is higher than a probability threshold), the electronic device may perform stroke connection correction on the stroke recognition information, so as to determine that the target stroke information corresponding to the target writing trajectory corresponds to the plurality of strokes, and obtain corresponding target stroke information.

408. And performing character recognition according to the target stroke information corresponding to each writing track to obtain a character recognition result matched with the at least one writing track.

In the embodiment of the application, the electronic device may recombine the target stroke information corresponding to each writing track into a complete character to be recognized, and then may perform character recognition on the character to be recognized, so as to obtain a character recognition result matched with each writing track after combination.

For example, in an embodiment, when performing the above character recognition, the electronic device may input a character to be recognized, which is obtained by recombining the target stroke information corresponding to each writing trajectory, into a character recognition model, perform steps of feature extraction, recognition, judgment and the like on the character to be recognized through the character recognition model, and finally output a corresponding character recognition result. In another embodiment, the electronic device may also directly input the target stroke information corresponding to each writing trajectory into the character recognition model, so as to perform subsequent steps of stroke combination, character recognition and the like in the character recognition model to obtain a corresponding character recognition result.

Therefore, the method for recognizing the written characters described in the above embodiment can fully utilize the stroke information of the written characters in the process of recognizing the written characters, recognize and correct the continuous strokes often appearing in the written characters, so as to accurately recognize each stroke contained in the continuous strokes, and further can recognize the characters by combining the stroke information of the written characters and the whole character information, thereby effectively improving the accuracy of recognizing the written characters.

Referring to fig. 5, fig. 5 is a flow chart illustrating another method for recognizing written characters according to the embodiment of the present application. As shown in fig. 5, the written character recognition method may include the steps of:

502. and responding to the writing input operation, and acquiring at least one writing track corresponding to the writing input operation.

Step 502 is similar to step 402, and is not described herein again.

504. And respectively carrying out stroke recognition on each writing track to obtain stroke recognition information and continuous stroke information of each writing track, wherein the stroke recognition information can comprise stroke identification information, and the stroke identification information is used for identifying each stroke contained in the writing track.

Illustratively, different types of strokes may correspond to different stroke identification information, respectively, and for example, stroke types such as horizontal, vertical, left-falling, right-falling, lifting, point, turning, hooking, etc., which are divided according to the basic form of the stroke, may be represented by different stroke identification information, respectively. In one embodiment, the stroke identification Information may be represented by codes such as, but not limited to, numeric codes, character codes, ASCII codes (American Standard Code for Information exchange), ASCII extended codes, and the like. The electronic equipment can determine the stroke types corresponding to all strokes contained in the writing track by identifying the writing track obtained by the writing track obtaining device, and further can determine the stroke identification information corresponding to all strokes contained in the writing track according to the stroke types.

Furthermore, after determining the stroke identification information of each writing track, the electronic device may also sort the obtained stroke identification information. Illustratively, the electronic device may perform sorting according to the timestamps corresponding to the writing tracks, so that stroke identification information sorted according to writing time sequence may be obtained; further illustratively, the electronic device may also sort according to the writing positions of the respective writing traces, so that stroke identification information sorted according to the spatial positions of the strokes in the character may be obtained. By sequencing the stroke identification information, the orderliness of the corresponding stroke identification information in the storage process can be improved, so that efficient and reliable access can be realized in the subsequent calling process, and the efficiency and reliability of character recognition are improved.

506. When the first writing track is determined to be the continuous writing track containing multiple strokes according to the continuous writing information of the first writing track, determining multiple stroke identification information corresponding to the first writing track as target stroke information corresponding to the first writing track, wherein the first writing track is any one of at least one writing track corresponding to the writing input operation.

Illustratively, the stroke connection information may include a stroke connection probability, a stroke connection confidence level, a corresponding confidence interval, and the like, for characterizing the possibility that the first writing track is a stroke connection track including a plurality of strokes. The continuous stroke probability may indicate a possibility that the first writing track includes a certain multi-stroke combination, and the continuous stroke confidence and the corresponding confidence interval may indicate a possibility that the first writing track is statistically a continuous stroke track.

In this embodiment, each writing track obtained by the electronic device may include one or more strokes, and the electronic device may accurately determine whether the first writing track is a continuous writing track including multiple strokes by determining the continuous writing information of the first writing track (e.g., determining whether the continuous writing probability satisfies a probability condition, determining whether a continuous writing confidence and a corresponding confidence interval satisfy a confidence condition, etc.). In some embodiments, if the first writing track includes a plurality of strokes, the plurality of strokes may form a corresponding combination of strokes. On this basis, when the electronic device determines that the first writing track is a continuous writing track containing multiple strokes according to continuous writing information of the first writing track, multiple stroke identification information (namely, the stroke combination) corresponding to the first writing track can be determined as target stroke information corresponding to the first writing track, so that accurate recognition of the strokes contained in the continuous writing track is achieved, namely correction of the continuous writing track is achieved, and further character recognition can be performed in subsequent steps according to the target stroke information of the writing tracks.

It will be appreciated that in stroke recognition of a first writing trace, there is a possibility of recognizing the first writing trace as different stroke combinations, i.e. the first writing trace may correspond to one or more stroke combinations, each of which may contain a plurality of stroke identification information. Illustratively, referring to fig. 2 and fig. 6, the third stroke writing input of the kanji character "permanent" is a continuous stroke, which may correspond to A, B, C stroke combinations, each stroke combination may include a plurality of stroke identification information of different numbers and types, for example, combination a may include stroke identification information corresponding to 3 strokes of turning, falling and falling, combination B may include stroke identification information corresponding to 4 strokes of pointing, falling and falling, and combination C may include stroke identification information corresponding to 3 strokes of pointing, falling and falling. On the basis, the electronic device may determine one of the stroke combinations as a target stroke combination of the first writing track, and further may determine whether the first writing track is a continuous stroke track including a plurality of strokes according to continuous stroke information (e.g., continuous stroke probability) corresponding to the target stroke combination.

As an optional implementation manner, when the electronic device determines that the first writing track is a continuous writing track including multiple strokes according to continuous writing information of the first writing track, if multiple stroke identification information corresponding to the first writing track includes invalid identification information meeting an invalid condition, the electronic device may further delete the invalid identification information, and determine the remaining stroke identification information as target stroke information corresponding to the first writing track. Illustratively, referring to FIG. 7, a stroke combination A corresponding to the first writing trajectory may be stored based on the stroke interval location corresponding to each stroke identification information it contains, e.g., the stroke combination A may originally include a stroke 701 ("fold") determined from the interval location points 1-2-3, a stroke 702 ("left-falling" or "lift") determined from the interval location points 3-4-5, a stroke 703 ("left-falling") determined from the interval location points 5-6, and a stroke 704 ("right-falling") determined from the interval location points 6-7. On this basis, the electronic device may respectively determine each stroke according to the invalidity condition to determine whether the corresponding stroke identification information is the invalid identification information satisfying the invalidity condition (e.g., whether the invalidity probability is higher than a certain probability threshold, whether the positions of the stroke sections overlap, whether the stroke identification information has an invalid feature, etc.), and then may delete the invalid identification information (e.g., the stroke 702) without corresponding to the actual stroke, and determine the remaining stroke identification information (e.g., the stroke 701, the stroke 703, and the stroke 704) as the target stroke information corresponding to the first writing trajectory.

508. The stroke recognition information further includes a stroke interval position corresponding to each stroke identification information, and a character to be recognized corresponding to the at least one writing track is synthesized based on each stroke identification information included in the target stroke information corresponding to each writing track and the stroke interval position corresponding to each stroke identification information.

In the embodiment of the application, the electronic device may synthesize strokes included in each writing track into corresponding characters to be recognized again according to the stroke identification information and the stroke interval position corresponding to each writing track. Illustratively, as shown in fig. 2 and fig. 7, for the third-stroke writing input of the kanji character "permanent", the mutual position relationship and the structural relationship between the strokes of the stroke track can be determined according to the stroke identification information and the stroke interval positions corresponding to the effective strokes (including the stroke 701 determined according to the interval position points 1-2-3, the stroke 703 determined according to the interval position points 5-6, and the stroke 704 determined according to the interval position points 6-7) included in the third-stroke writing input. Furthermore, according to the stroke combination contained in each writing track, the character to be recognized formed by each writing track can be accurately restored, and further the character to be recognized obtained through synthesis can be subjected to character recognition in the subsequent steps, so that the stroke information and the whole character information of the written character are fully utilized, and the accuracy of the written character recognition is improved.

510. And performing character recognition on the character to be recognized to obtain a character recognition result matched with the at least one writing track.

Step 510 is similar to step 408 described above, and is not described herein again.

Therefore, by implementing the method for recognizing the written characters described in the embodiment, the stroke identification information contained in the target writing track can be corrected according to the continuous stroke information, so that the accuracy of recognizing each stroke contained in the continuous stroke is improved; in addition, the target stroke information actually corresponding to each writing track is determined through threshold judgment and other modes so as to realize the continuous stroke correction process, the realization is simple, the calculated amount is small, and the efficiency of writing character recognition is favorably improved.

Referring to fig. 8, fig. 8 is a flowchart illustrating another method for recognizing written characters according to an embodiment of the present application. As shown in fig. 8, the written character recognition method may include the steps of:

802. and responding to the writing input operation, and acquiring at least one writing track corresponding to the writing input operation.

Step 802 is similar to step 402, and is not described herein again.

As an optional implementation manner, after acquiring at least one writing track corresponding to the writing input operation, the electronic device may further adjust the writing track according to device posture information of the electronic device. Specifically, the electronic device may first obtain device posture information (such as an inclination angle, a turning angle, and the like) corresponding to the writing input operation, and then perform posture adjustment on at least one writing trajectory corresponding to the writing input operation according to the device posture information (such as correspondingly adjusting the inclination angle, the turning angle, and the like of the writing trajectory). On the basis, the electronic equipment can respectively perform stroke recognition on each writing track subjected to posture adjustment to obtain stroke recognition information of each writing track, so that the reliability of character recognition performed in the electronic equipment in unstable states such as inclination, overturning and movement is improved.

804. And respectively carrying out stroke recognition on each writing track to obtain a plurality of stroke combinations corresponding to each writing track and continuous stroke information corresponding to each stroke combination, wherein the continuous stroke information comprises continuous stroke probability.

Step 804 is similar to step 504 described above. The continuous stroke probability may indicate a possibility that the first writing trajectory corresponds to one of the stroke combinations, and the first writing trajectory may be any one of at least one writing trajectory corresponding to the writing input operation.

806. And according to the continuous stroke probability corresponding to each stroke combination, selecting a target stroke combination from a plurality of stroke combinations corresponding to the first writing track, determining the target stroke combination as stroke identification information of the first writing track, and determining the continuous stroke probability corresponding to the target stroke combination as the continuous stroke probability of the first writing track.

In an embodiment of the present application, when there are multiple stroke combinations in the first writing trace, each stroke combination may correspond to a different stroke connection probability. Illustratively, the electronic device may perform analysis and judgment according to the stroke connection probability corresponding to each stroke combination, select a target stroke combination with the stroke connection probability satisfying a certain probability condition from the plurality of stroke combinations, and then may accurately determine the stroke contained in the first writing track based on the target stroke combination, and the stroke identification information, the stroke interval position, the stroke connection probability, and the like corresponding to the stroke combination, so as to improve the accuracy of subsequent character recognition by using the stroke information of the written character.

In an embodiment, if there are multiple stroke combinations in the first writing track, when determining the target stroke combination from the multiple stroke combinations, the electronic device may select a stroke combination with the highest continuous stroke probability from the multiple stroke combinations as the target stroke combination according to the continuous stroke probability corresponding to each stroke combination. For example, the electronic device may sequentially compare the stroke connection probabilities corresponding to each stroke combination, and keep the stroke combination with the higher stroke connection probability obtained by each comparison for the next comparison until all the stroke combinations are compared, obtain the stroke combination with the highest stroke connection probability, and determine the stroke combination as the target stroke combination.

For example, as shown in fig. 6, the first writing trajectory may include a third writing input of the kanji character "permanent," which may correspond to three stroke combinations, where the stroke connection probability of combination a may be a%, the stroke connection probability of combination B may be B%, and the stroke connection probability of combination C may be C%. According to the respective continuous stroke probabilities of the three continuous stroke combinations, a group of stroke combinations with the highest continuous stroke probability can be determined as the target stroke combination corresponding to the writing track.

In some embodiments, the electronic device may further determine one or more alternative stroke combinations in addition to the target stroke combination determined from the plurality of stroke combinations corresponding to the first writing trajectory, and may further obtain a corresponding alternative character recognition result according to the alternative stroke combinations in a subsequent character recognition process, so as to serve as an alternative option of the character recognition result, thereby improving flexibility of written character recognition. For example, the electronic device may determine a second probability threshold (e.g., 60%, 80%, 95%, etc.), and select a stroke combination with a stroke probability exceeding the second probability threshold from other stroke combinations except the target stroke combination as an alternative stroke combination according to the stroke connection probability corresponding to each stroke combination, and then determine the stroke connection probability corresponding to a plurality of stroke identification information included in the alternative stroke combination as the alternative stroke connection probability corresponding to the first writing trajectory, so as to be used in the subsequent steps of stroke connection correction and character recognition.

808. And if the stroke connection probability of the first writing track is higher than a first probability threshold, determining that the first writing track is a stroke connection track of a plurality of strokes, and determining a plurality of stroke identification information in a target stroke combination corresponding to the first writing track as target stroke information corresponding to the first writing track.

In this embodiment, the electronic device may determine a first probability threshold (e.g., 60%, 80%, 95%, etc.), compare the stroke connection probability of the first writing trajectory determined in step 806 with the first probability threshold (i.e., the stroke connection probability corresponding to the plurality of stroke identification information included in the target stroke combination of the first writing trajectory), and if the comparison result shows that the stroke connection probability exceeds the first probability threshold, determine that the first writing trajectory corresponds to the stroke connection, which requires further stroke connection correction in subsequent steps.

For example, continuing to refer to fig. 6, if the combination a corresponding to the first writing track is the target stroke combination, and when the continuous stroke probability a% of the combination a is determined to be higher than the first probability threshold, the electronic device may modify the stroke identification information of the first writing track into three stroke identification information (turning, left falling, right falling) included in the combination a. Alternatively, if the stroke connection probability B% of the combination B is also higher than the first probability threshold, and B% is greater than a%, the electronic device may preferentially modify the stroke identification information of the target writing trajectory into the three stroke identification information (point, left-falling, right-falling) included in the combination B.

810. The stroke recognition information further includes a stroke interval position corresponding to each stroke identification information, and a character to be recognized corresponding to the at least one writing track is synthesized based on each stroke identification information included in the target stroke information corresponding to each writing track and the stroke interval position corresponding to each stroke identification information.

812. And performing character recognition on the character to be recognized to obtain a character recognition result matched with the at least one writing track.

Step 810 and step 812 are similar to step 508 and step 510, and are not described again here.

Therefore, by implementing the method for recognizing the written characters described in the embodiment, the stroke identification information contained in the target writing track can be corrected according to the continuous stroke information, so that the accuracy of recognizing each stroke contained in the continuous stroke is improved; in addition, the target stroke information actually corresponding to each writing track is determined through threshold judgment and other modes so as to realize the continuous stroke correction process, the realization is simple, the calculation amount is small, and the efficiency of writing character recognition is favorably improved; in addition, by adjusting the posture of the writing track, the reliability of character recognition of the electronic equipment in unstable states such as inclination, overturning and movement can be improved, and the accuracy of character recognition of the electronic equipment is further improved.

Referring to fig. 9, fig. 9 is a schematic block diagram of a written character recognition apparatus according to an embodiment of the present disclosure. As shown in fig. 9, the written character recognition apparatus may include an input acquisition unit 901, a stroke recognition unit 902, a continuous stroke correction unit 903, and a character recognition unit 904, in which:

an input obtaining unit 901, configured to respond to a writing input operation and obtain at least one writing trajectory corresponding to the writing input operation;

a stroke recognition unit 902, configured to perform stroke recognition on each writing trace, to obtain stroke recognition information and stroke connection information of each writing trace;

a continuous stroke correction unit 903, configured to perform continuous stroke correction on the corresponding stroke identification information according to the continuous stroke information of each writing trajectory to obtain target stroke information corresponding to each writing trajectory;

and a character recognition unit 904, configured to perform character recognition according to the target stroke information corresponding to each writing track, and obtain a character recognition result matching with the at least one writing track.

It can be seen that, by adopting the written character recognition device described in the above embodiment, stroke information of written characters can be fully utilized in the process of recognizing the written characters, and continuous strokes often appearing in the written characters are recognized and corrected, so that all strokes contained in the continuous strokes can be accurately recognized, and further, character recognition can be performed by combining the stroke information of the written characters and the whole character information, and the accuracy of written character recognition is effectively improved.

In one embodiment, the stroke recognition information may include stroke identification information for identifying each stroke contained in the writing trace. The continuous stroke correcting unit 903 may be specifically configured to, when it is determined that the first writing trajectory is a continuous stroke trajectory including multiple strokes according to continuous stroke information of the first writing trajectory, determine multiple stroke identification information corresponding to the first writing trajectory as target stroke information corresponding to the first writing trajectory, where the first writing trajectory is any one of at least one writing trajectory corresponding to the writing input operation.

Further, the continuous stroke information may include a continuous stroke probability, and the continuous stroke correcting unit 903 may be specifically configured to compare the continuous stroke probability of the first writing track with a first probability threshold, determine that the first writing track is a continuous stroke track including multiple strokes if the continuous stroke probability is higher than the first probability threshold, and determine that multiple stroke identification information corresponding to the first writing track is target stroke information corresponding to the first writing track.

It can be seen that, with the written character recognition device described in the above embodiment, the target stroke information actually corresponding to each writing trajectory is determined by means of threshold determination and the like, so as to implement the continuous stroke correction process.

In an embodiment, the continuous stroke correcting unit 903 may be further configured to, when it is determined that the first writing track is a continuous stroke track including multiple strokes according to continuous stroke information of the first writing track, delete invalid identification information if the multiple stroke identification information corresponding to the first writing track includes invalid identification information meeting an invalid condition, and determine the remaining stroke identification information as target stroke information corresponding to the first writing track.

In an embodiment, the written character recognition apparatus may further include a combination selection unit, not shown, configured to, before the continuous stroke correction unit 903 compares the continuous stroke probability of the first writing trajectory with the first probability threshold, select, when there are multiple stroke combinations in the first writing trajectory, a target stroke combination from the multiple stroke combinations according to the continuous stroke probability corresponding to each stroke combination, and determine, as the continuous stroke probability of the first writing trajectory, the continuous stroke probability corresponding to multiple stroke identification information included in the target stroke combination.

Optionally, if there are multiple stroke combinations in the first writing track, the combination selecting unit may specifically select, according to the stroke connection probability corresponding to each stroke combination, a stroke combination with the highest stroke connection probability from the multiple stroke combinations as the target stroke combination.

In an embodiment, the combination selecting unit may be further configured to, after the target stroke combination is selected, select, according to the stroke connection probability corresponding to each stroke combination, a stroke combination with a stroke connection probability exceeding a second probability threshold from other stroke combinations except the target stroke combination as an alternative stroke combination, and determine, as the alternative stroke connection probability corresponding to the first writing trajectory, the stroke connection probability corresponding to a plurality of stroke identification information included in the alternative stroke combination.

Therefore, by adopting the written character recognition device described in the above embodiment, the corresponding alternative character recognition result can be obtained according to the alternative stroke combination in the subsequent character recognition process, so as to serve as the alternative option of the character recognition result, and improve the flexibility of written character recognition.

In one embodiment, the written character recognition apparatus may further include a posture acquiring unit and a posture adjusting unit, not shown, wherein:

a posture acquiring unit, configured to acquire device posture information corresponding to a writing input operation after the input acquiring unit 901 acquires at least one writing trajectory corresponding to the writing input operation;

the gesture adjusting unit is used for carrying out gesture adjustment on at least one writing track corresponding to the writing input operation according to the equipment gesture information;

the stroke recognition unit 902 may be specifically configured to perform stroke recognition on each writing track subjected to the posture adjustment, so as to obtain stroke recognition information of each writing track.

Therefore, the written character recognition device described in the above embodiment can correct the stroke identification information included in the target writing track according to the continuous stroke information, and improves the accuracy of recognizing each stroke included in the continuous stroke; in addition, the target stroke information actually corresponding to each writing track is determined through threshold judgment and other modes so as to realize the continuous stroke correction process, the realization is simple, the calculation amount is small, and the efficiency of writing character recognition is favorably improved; in addition, by adjusting the posture of the writing track, the reliability of character recognition of the electronic equipment in unstable states such as inclination, overturning and movement can be improved, and the accuracy of character recognition of the electronic equipment is further improved.

Referring to fig. 10, fig. 10 is a schematic view of a module of an electronic device according to an embodiment of the disclosure. As shown in fig. 10, the electronic device may include:

a memory 1001 in which executable program code is stored;

a processor 1002 coupled to a memory 1001;

the processor 1002 calls the executable program code stored in the memory 1001, and may execute all or part of the steps in any one of the written character recognition methods described in the above embodiments.

In addition, the present application further discloses a computer-readable storage medium storing a computer program for electronic data exchange, wherein the computer program enables a computer to execute all or part of the steps of any one of the written character recognition methods described in the above embodiments.

In addition, the embodiments of the present application further disclose a computer program product, which, when running on a computer, enables the computer to execute all or part of the steps of any one of the written character recognition methods described in the above embodiments.

It will be understood by those skilled in the art that all or part of the steps in the methods of the embodiments described above may be implemented by hardware instructions of a program, and the program may be stored in a computer-readable storage medium, where the storage medium includes Read-Only Memory (ROM), Random Access Memory (RAM), Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), One-time Programmable Read-Only Memory (OTPROM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Compact Disc Read-Only Memory (CD-ROM), or other Memory, such as a magnetic disk, or a combination thereof, A tape memory, or any other medium readable by a computer that can be used to carry or store data.

The method and the device for recognizing the written character, the electronic device and the storage medium disclosed in the embodiments of the present application are introduced in detail, and a specific example is applied in the present application to explain the principle and the implementation of the present application, and the description of the above embodiments is only used to help understanding the method and the core idea of the present application; meanwhile, for a person skilled in the art, according to the idea of the present application, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present application.

Claims

1. A method for written character recognition, comprising:

2. The method according to claim 1, wherein the stroke recognition information includes stroke identification information for identifying each stroke contained in a writing trace;

the continuous stroke correction of the corresponding stroke identification information according to the continuous stroke information of each writing track to obtain the target stroke information corresponding to each writing track comprises the following steps:

when the first writing track is determined to be the continuous writing track containing multiple strokes according to continuous writing information of the first writing track, determining multiple stroke identification information corresponding to the first writing track as target stroke information corresponding to the first writing track, wherein the first writing track is any one of at least one writing track corresponding to the writing input operation.

3. The method of claim 2, wherein the stroke connection information includes a stroke connection probability, and determining, when the first writing track is determined to be a stroke connection track containing a plurality of strokes according to the stroke connection information of the first writing track, a plurality of stroke identification information corresponding to the first writing track as the target stroke information corresponding to the first writing track comprises:

and if the continuous stroke probability is higher than a first probability threshold, determining that the first writing track is a continuous stroke track containing a plurality of strokes, and determining a plurality of stroke identification information corresponding to the first writing track as target stroke information corresponding to the first writing track.

4. The method of claim 2, wherein when determining that the first writing track is a continuous stroke track containing a plurality of strokes according to continuous stroke information of the first writing track, determining a plurality of stroke identification information corresponding to the first writing track as target stroke information corresponding to the first writing track comprises:

when the first writing track is determined to be the continuous writing track containing multiple strokes according to the continuous writing information of the first writing track, if the multiple stroke identification information corresponding to the first writing track contains invalid identification information meeting an invalid condition, deleting the invalid identification information, and determining the remaining stroke identification information as target stroke information corresponding to the first writing track.

5. The method of claim 1, wherein the performing stroke recognition on each writing track respectively to obtain stroke recognition information and stroke connection information of each writing track comprises:

stroke recognition is carried out on each writing track respectively to obtain a plurality of stroke combinations corresponding to the writing tracks and continuous stroke information corresponding to the stroke combinations, wherein the continuous stroke information comprises continuous stroke probability;

selecting a target stroke combination from the plurality of stroke combinations according to the stroke connection probability corresponding to each stroke combination;

and determining the target stroke combination as stroke identification information of the writing track, and determining the continuous stroke probability corresponding to the target stroke combination as the continuous stroke probability of the writing track.

6. The method of claim 5, wherein selecting a target stroke combination from the plurality of stroke combinations according to the stroke connection probability corresponding to each stroke combination comprises:

and selecting the stroke combination with the highest continuous stroke probability from the plurality of stroke combinations as a target stroke combination according to the continuous stroke probability corresponding to each stroke combination.

7. The method of claim 5, wherein after determining the stroke tie probability corresponding to the target stroke combination as the stroke tie probability of the writing trace, the method further comprises:

selecting stroke combinations with the continuous stroke probability exceeding a second probability threshold value from other stroke combinations except the target stroke combination as alternative stroke combinations according to the continuous stroke probability corresponding to each stroke combination;

and adding the alternative stroke combination into stroke identification information of the writing track, and determining the continuous stroke probability corresponding to the alternative stroke combination as the alternative continuous stroke probability of the writing track.

8. The method as claimed in any one of claims 2 to 4, wherein the stroke recognition information further includes a stroke interval position corresponding to each stroke identification information, and the performing character recognition according to the target stroke information corresponding to each writing track to obtain a character recognition result matching with the at least one writing track includes:

synthesizing characters to be recognized corresponding to the at least one writing track based on each stroke identification information included in the target stroke information corresponding to each writing track and the stroke interval position corresponding to each stroke identification information;

and performing character recognition on the character to be recognized to obtain a character recognition result matched with the at least one writing track.

9. The method according to any one of claims 1 to 7, wherein after the obtaining of the at least one writing trace corresponding to the writing input operation in response to the writing input operation, the method further comprises:

acquiring equipment posture information corresponding to the writing input operation;

performing posture adaptation adjustment on at least one writing track corresponding to the writing input operation according to the equipment posture information;

the step of respectively performing stroke recognition on each writing track to obtain stroke recognition information of each writing track comprises the following steps:

and respectively carrying out stroke recognition on each writing track subjected to posture adaptation adjustment to obtain stroke recognition information of each writing track.

10. A written character recognition apparatus, characterized in that the written character recognition apparatus comprises:

11. An electronic device comprising a memory and a processor, the memory having stored thereon a computer program that, when executed by the processor, causes the processor to implement the method of any of claims 1 to 9.

12. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the method according to any one of claims 1 to 9.