CN101976152A - Method and device of handwriting recognition - Google Patents
Method and device of handwriting recognition Download PDFInfo
- Publication number
- CN101976152A CN101976152A CN 201010542942 CN201010542942A CN101976152A CN 101976152 A CN101976152 A CN 101976152A CN 201010542942 CN201010542942 CN 201010542942 CN 201010542942 A CN201010542942 A CN 201010542942A CN 101976152 A CN101976152 A CN 101976152A
- Authority
- CN
- China
- Prior art keywords
- turn time
- character
- current
- lifting
- threshold value
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Landscapes
- Character Discrimination (AREA)
Abstract
The invention discloses a method and a device of handwriting recognition. The method comprises the following steps: receiving character data signals input by handwriting; updating a pen-rising waiting time threshold according to the real-time emergent pen-rising waiting time intervals among the character data signals; judging whether the current pen-rising waiting time interval exceeds the pen-rising waiting time threshold which is updated in real time; if so, determining that the current characters are completely input; and recognizing the current characters. In the invention, the writing habit of the user can be continuously monitored and studied, the pen-rising waiting time is updated in real time and the character recognition efficiency is improved.
Description
Technical field
The present invention relates to discern processing technology field, more particularly, relate to a kind of hand-written recognition method and device.
Background technology
Through technical development for many years, handwriting input is stepped into the practical stage as a kind of important non-keyboard input method, is widely used in equipment such as mobile phone, notebook computer, iPad.When adopting the handwriting input mode to carry out the information data input, judge that accurately EOC is the key that realizes the correct input of information data.
In recent years, along with the development of recognition technology, handwriting recognition has proposed the scheme that multiple Multi strokes EOC is judged in the industry, and more common having is following several:
(1) uses two lattice/many lattice inputs.Usually, writing the zone has two or more windows of writing, and the user can replace the isolated character of input in two lattice/many lattice, realizes the typing of the continuous whole sentence of individual character.Thisly separate the end that the mode of isolated word can the single character of high efficient and reliable sex determination, thereby become a kind of main flow input mode on the current handwriting recognition application market by the interface of determining.But this input mode is limited in the specific hand-written frame and accurately locatees, and is liked the user of natural and tripping handwriting input to bring inconvenience.
(2) introduce the end that specific identifier is demarcated the monocase input, as can be by the detected gesture symbol of simple algorithm.Obviously, the increase of identifier can bring obscures word identification difficulty, and simultaneously, system needs monitoring identifier in real time, will bring certain influence to discrimination and recognition efficiency.
(3) setting is lifted a turn time.What is called is lifted pen and is waited in turn and be meant the user after having write an isolated character, and the conscious pen of lifting allows the tip of the brushstyle of a writing or painting leave touch-screen, prepares the next character of acceptance until the interface cls.The mechanism that is provided with of lifting a turn time of intelligence is improved user experience and is played crucial effects improving system performance.Judge the end of character typing too early and start recognition engine if lift a turn time, will bring the reduction of discrimination.Otherwise, finish though can better guarantee individual character reliably, need test user's patience, the waiting time that cost is long.
At present, the most frequently used being provided with lifted the individual character of a turn time and imported continuously in the recognition system, is provided with to lift a turn time integrating instrument, and this is lifted a turn time integrating instrument and is provided with a time threshold value.If currently lift a turn time and surpass this time threshold, determine then that current character is write to finish, start recognition engine current character is discerned.
Yet by discovering of inventor, the individual character of lifting a turn time that is provided with of the prior art is imported continuously and is still had following problem in the recognition system:
Different user is applicable to different lifting a turn time.More expect long lifting a turn time such as older user, and young user is suitable for of short duration lifting a turn time.As seen, prior art can't adapt to all users' writing style, is easy to generate wrong EOC judged result, thereby causes the character recognition inefficiency.
Summary of the invention
In view of this, the embodiment of the invention provides a kind of hand-written recognition method and device, and to realize the continuous monitoring study to the user writing custom, real-time update is lifted a turn time, improves character recognition efficient.
The embodiment of the invention provides a kind of hand-written recognition method, comprising:
Receive the character data signal of handwriting input;
Lift a turn time at interval according to what occur in real time between described character data signal, upgrade and lift a turn time threshold value;
Judge and current lift a turn time and whether surpass the turn time threshold value of lifting after the real-time update at interval, if determine that then the current character input finishes;
Current character is discerned.
Further, described method also comprises:
Obtain all written handwriting signals of the described character of current accumulative total;
Determine the length breadth ratio of current character according to described written handwriting signal, judge by the length breadth ratio of described current character whether described character is imported and finish.
Further, described method also comprises:
All the written handwriting data-signals of described character that obtain current accumulative total are in the distribution situation of writing four quadrants in interface;
Judge describedly write in four quadrants in interface whether the written handwriting data-signal is all arranged, if determine that then described character input finishes.
Further, described method also comprises:
When the described character input of false judgment finishes, utilize a turn time threshold value lower limit of lifting that presets that a turn time threshold value of lifting of current acquisition is upgraded.
Preferably, described according to a turn time interval of lifting that occurs in real time between described character data signal, a turn time threshold value is lifted in renewal, specifically comprises:
According to a turn time interval samples of lifting of history accumulative total turn time first average spaced apart and first variance and current appearance, according to second average and the second variance that a turn time distributes of lifting that presets after more new model obtains to upgrade;
The Gaussian distribution model that a turn time satisfies is lifted in described second average and second variance substitution, and respectively lift the accumulated probability that presets that a turn time satisfies in the turn time threshold range, lift a turn time threshold value after obtaining upgrading according to current lifting.
A kind of handwriting recognition device comprises:
Signal receiving module is used to receive the character data signal of handwriting input;
The time threshold update module is used for lifting a turn time at interval according to what occur in real time between described character data signal, upgrades and lifts a turn time threshold value;
First judge module is used to judge and current lift a turn time and whether surpass the turn time threshold value of lifting after the real-time update at interval, if determine that then the current character input finishes;
Identification module is used for current character is discerned.
Further, described device also comprises:
Written handwriting data-signal acquisition module is used to obtain all written handwriting data-signals of the described character of current accumulative total;
Second judge module is used for determining according to described written handwriting data-signal the length breadth ratio of current character, judges by the length breadth ratio of described current character whether described character is imported and finishes.
Further, described device also comprises:
Quadrant distribution acquisition module, all the written handwriting data-signals of described character that are used to obtain current accumulative total are in the distribution situation of writing four quadrants in interface;
The 3rd judge module is used for judging whether described four quadrants in interface of writing all have the written handwriting data-signal, if determine that then described character input finishes.
Further, described device also comprises:
Lift a turn time threshold value lower limit module is set, be used for when the described character input of false judgment finishes, utilizing a turn time threshold value lower limit of lifting that presets that a turn time threshold value of lifting of current acquisition is upgraded.
Preferably, described time threshold update module specifically comprises:
First calculating sub module, be used for a turn time interval samples of lifting, according to second average and the second variance that a turn time distributes of lifting that presets after more new model obtains to upgrade according to history accumulative total turn time first average spaced apart and first variance and current appearance;
Second calculating sub module, be used for the Gaussian distribution model that a turn time satisfies is lifted in described second average and second variance substitution, and respectively lift the accumulated probability that presets that a turn time satisfies in the turn time threshold range according to current lifting, lift a turn time threshold value after obtaining upgrading.
Compare with prior art, technical scheme provided by the invention is lifted a turn time at interval according to what occur in real time between described character data signal, a turn time threshold value is lifted in renewal, thereby reach monitoring of lifting a turn time and adaptive learning in the user writing process, can be provided with at different user and lift a turn time threshold value accordingly, to adapt to the writing style of relative users, avoid occurring false judgment to the character end of input, improve character recognition efficient.
Description of drawings
In order to be illustrated more clearly in the technical scheme of the embodiment of the invention, to do to introduce simply to the accompanying drawing of required use in embodiment or the description of the Prior Art below, apparently, accompanying drawing in describing below only is some embodiments of the present invention, for those of ordinary skills, under the prerequisite of not paying creative work, can also obtain other accompanying drawing according to these accompanying drawings.
A kind of hand-written recognition method schematic flow sheet that Fig. 1 provides for the embodiment of the invention;
The another kind of hand-written recognition method schematic flow sheet that Fig. 2 provides for the embodiment of the invention;
Another hand-written recognition method schematic flow sheet that Fig. 3 provides for the embodiment of the invention;
The structural representation of a kind of handwriting recognition device that Fig. 4 provides for the embodiment of the invention;
The structural representation of the another kind of handwriting recognition device that Fig. 5 provides for the embodiment of the invention;
The structural representation of another handwriting recognition device that Fig. 6 provides for the embodiment of the invention;
The structural representation of another handwriting recognition device that Fig. 7 provides for the embodiment of the invention.
Embodiment
Below in conjunction with the accompanying drawing in the embodiment of the invention, the technical scheme in the embodiment of the invention is clearly and completely described, obviously, described embodiment only is the present invention's part embodiment, rather than whole embodiment.Based on the embodiment among the present invention, those of ordinary skills belong to the scope of protection of the invention not making the every other embodiment that is obtained under the creative work prerequisite.
The people tends to continuous input more when normally writing, if it is to be identified to write word etc., write next word again, and thinking tends to be interrupted.Thereby support that the handwriting recognition of input receives increasing concern continuously.Current, the input pattern that adopt in the handwriting recognition field still is the monocase pattern of input continuously.
So-called monocase input continuously is meant that system starts the rear end recognition engine character is discerned after the judgement user finishes the monocase input, and provides recognition result.Obviously under such recognition mode, to realize the function of input continuously, need possess several requirements simultaneously.At first the individual character recognition engine efficient of its rear end is higher, can finish identification as soon as possible; There is the isolated EOC judgment criterion of a kind of efficiently and accurately and intelligence simultaneously, so that when user's monocase end of input, can in time start recognition engine and prepare to accept next character.
In to isolated EOC deterministic process, lifting a setting of turn time threshold value is a lot of handwriting input products, particularly is input as the focus of main product with individual character.At present most of in the industry handwriting input products to lift a setting of turn time threshold value be mechanical, promptly default a plurality of alternative for users in the page is set according to one of own customs selection suitable lift a turn time threshold value.In the handwriting input process, when lifting a dead time, the user surpasses the schedule time during threshold value, and system promptly assert the current character end of input and begins identification.So,, necessarily have the word of not expecting and be identified if when current single character is write half and had intermittently.Particularly when the user will lift for pursuing quick input that a turn time, threshold value was made as smaller value, as long as start writing slightly, leave handwriting pad, then will be considered to the character end of input, and when actual characters input does not finish, start recognition engine, thereby cause the interruption of normal input, character-recognition errors.Otherwise when setting lifted that a turn time, threshold value was longer, the user often needed to wait for that time enough just can enter the character recognition stage, thereby has influenced the continuous writing effect.As seen, reach a kind of balance between handwriting input efficient and the system identification efficient, need rationally to be provided with to lift a turn time threshold value, under the prerequisite that guarantees recognition efficiency, meet the user writing custom as far as possible, help the user to realize natural and tripping continuous writing.
In order to solve above-mentioned topic, correspondingly, the embodiment of the invention provides a kind of hand-written recognition method, correspondingly, also provides a kind of handwriting recognition device.
At first a kind of hand-written recognition method provided by the invention is described, with reference to shown in Figure 1, described method comprises:
The character data signal of step 101, reception handwriting input;
The technical scheme that the embodiment of the invention provides is in step 102, lift a turn time at interval according to what occur in real time between described character data signal, a turn time threshold value is lifted in renewal, thereby reach monitoring of lifting a turn time and adaptive learning in the user writing process, can be provided with at different user and lift a turn time threshold value accordingly, to adapt to the writing style of relative users, avoid occurring false judgment to the character end of input, improve character recognition efficient.
Usually, after definite current character input finishes, then can discern this character.When embodiment of the invention technical scheme is implemented, in order to guarantee the accuracy of Handwritten Digits Recognition, improve the efficient of character recognition, whether import in the process that finishes in definite current character, not only to investigate a current turn time interval of lifting, also need specifically to judge according to the written handwriting of current character.Thus, in another embodiment of the present invention, when determine current lift a turn time after surpassing real-time update at interval lift a turn time threshold value after, as shown in Figure 2, further comprising the steps of:
The embodiment of the invention is by the length breadth ratio from character, and whether geometrical property aspects such as integrality are imported to finish to it is judged.The length breadth ratio of a complete character is about about 1, and too small or excessive length and width ratio has then reflected the imperfection of character.For example: for the Chinese character of left and right sides structure, when user writing was finished left half of radicals by which characters are arranged in traditional Chinese dictionaries, the length of current character structure was big more a lot of than width, and length and width ratio is out and away greater than 1.Otherwise for the Chinese character of up-down structure, when part was finished, the width of current character structure will be far longer than length.Thereby, to a certain extent, the performance level of character as can be seen from the length and width ratio of character.In the embodiment of the invention, can introduce a reference parameter " degree of confidence ", utilize this parametric representation to judge whether to occur complete, significant character by all written handwriting signals of the described character of current accumulative total, should " degree of confidence " can represent with Y, the length that can suppose character is H, width is W, and then the degree of confidence that is obtained by the property calculation of length breadth ratio is:
According to above-mentioned definition, when the length and width ratio of character is between 0.8～1.2, can determine the integrality of character substantially to degree of confidence; And when the length and width ratio of character less than 0.5 or greater than 2 the time, think that then character do not finish; Under other conditions, calculate concrete degree of confidence according to the mode of line shape function.
In the specific implementation, the written handwriting signal shows as the tip of the brushstyle of a writing or painting in the x that writes each contact point on the interface, y axial coordinate value.At the coordinate figure of writing each contact point on the interface, obtain the written handwriting of current character by the accumulative total tip of the brushstyle of a writing or painting.
In order further to improve the accuracy that character input integrality is judged, in another embodiment of the present invention, as shown in Figure 3, after the length and width ratio of having determined current input character, can also may further comprise the steps:
When judging the integrality of input character, as to the replenishing of input character length and width ratio characteristic, can also be in conjunction with the distribution situation of written handwriting four quadrants in writing the interface of this input character.Usually, the normal character of complete input all has the distribution of pixel writing each quadrant on the interface.And, often do not have stroke to occur at some quadrant for uncompleted stroke still.Such as character " greatly ", after finishing the first stroke and second, second quadrant does not have stroke to occur to its character on the interface writing, thereby when definite this character is writing that second quadrant written handwriting do not occur on the interface, can determine that then this character do not finish.Confidence calculations based on this characteristic can be designated as:
Therefore, after definite degree of confidence Y and Y ', can obtain comprehensive degree of confidence Y1=Y*Y '.Just when character was the character of complete input, it required to satisfy simultaneously length and width ratio and quadrant integrality.For example: when character was writing that there is the clear area in certain quadrant on the interface, even the length breadth ratio of character equals 1, comprehensive degree of confidence Y1 still was 0, and still can be defined as is the character of not complete input.During concrete enforcement, can preset comprehensive confidence threshold value,, determine that then the current character input is imperfect when the comprehensive degree of confidence that obtains during less than this comprehensive confidence threshold value.
Need to prove, lift a turn time in real time and upgrade at interval and lift in the process of turn time threshold value utilizing, following situation may occur:
When determining that according to the turn time threshold value of lifting after upgrading current character is a complete character, but, the comprehensive degree of confidence that obtains according to the written handwriting of this current character is less than comprehensive confidence threshold value, when being an imperfect character, then be owing to too short lift the error in judgement that a turn time threshold value (can be made as Ta) causes.If lift a turn time threshold value Ta numerical value be provided with too small, so in user's input process in character the minibreak between the different strokes also can be identified as be that the intercharacter pen of lifting pauses, thereby cause error in judgement, this kind situation was become training usually.For fear of crossing the too short turn time threshold value of lifting that training causes, can preestablish long turn time threshold value lower limit Tb that lifts, after renewal lift the too short incompatibility practical application of a turn time threshold value time, utilize Tb to upgrade waiting time Ta.Lift a turn time threshold value lower limit Tb by setting, avoid causing too short irrational turn time threshold value of lifting because of the too short dead time of user.It is bigger to lift a surplus that turn time threshold value lower limit Tb often is provided with, the isolated character interval time that is generally normally or writes slightly slowly, for example 500ms.
In addition, lift a turn time at interval, upgrade and lift a turn time threshold value according to what occur in real time between described character data signal, specific implementation can for:
According to a turn time interval samples of lifting of history accumulative total turn time first average spaced apart and first variance and current appearance, according to second average and the second variance that a turn time distributes of lifting that presets after more new model obtains to upgrade;
The Gaussian distribution model that a turn time satisfies is lifted in described second average and second variance substitution, and respectively lift the accumulated probability that presets that a turn time satisfies in the turn time threshold range, lift a turn time threshold value after obtaining upgrading according to current lifting.
For the ease of understanding, be described in detail explanation to lifting a update scheme of turn time threshold value below by concrete example to this part technical scheme.
In the analysis to a large amount of handwriting input data, those skilled in the art can be known: between the stroke in the single character input process between the distribution of dead time and the character distribution of dead time follow Gaussian distribution:
The concrete distribution determined by these two parameters of average μ and variances sigma in the Gaussian distribution model.
Setting initial average μ is μ
0(for example 160 milliseconds), initial variance σ is σ
0(for example 200 milliseconds).Along with the lift pen of user in writing process pauses, average μ and variances sigma in the Gaussian distribution model are upgraded in the following manner:
μ
n＝(1-α)·μ
n-1+α·t
n
Wherein, t
nBe user the n time lift a turn time at interval, α and β are two definite parameter values that set in advance in the system, represent the forgetting factor of Mean Parameters and the forgetting factor of variance parameter respectively.The setting of α and β is for the influence to the model parameter change of the current up-to-date training sample of balance and historical sample.
Can see that from above-mentioned formula the value of α and β is more little, model parameter is that average and variance are more guided by historical sample, thereby keeps bigger stability.On the contrary, the value of α and β is big more, and the renewal of parameter value average and variance can absorb more new samples information, and random fluctuation is bigger.In the embodiment of the invention, α=β=0.1 is set respectively rule of thumb, when absorbing the new samples fresh information, guarantees the relative stability of model parameter.Certainly, those skilled in the art can carry out the setting of α and β according to the practical application scene when specifically implementing technical solution of the present invention, and to this, the present invention does not do concrete qualification.
In lifting the renewal process of turn time threshold value Ta, utilized the μ that upgrades just
nAnd σ
nIn order to guarantee to lift validity and the rationality of a turn time threshold value Ta on character integrity is judged, can set all and surpass 0.99, that is: less than the accumulated probability of turn time appearance of respectively lifting in the time range of Ta value
In realizing the process that current character is discerned, can be provided for starting the identification information of identification.In the embodiment of the invention, this identification information can be the Boolean type variable.The rear end recognition engine can start the identification to input character according to this identification information of front end output.
Corresponding above-mentioned hand-written recognition method embodiment, the present invention also provides a kind of handwriting recognition device, and as shown in Figure 4, described device comprises:
Signal receiving module 401 is used to receive the character data signal of handwriting input;
Time threshold update module 402 is used for lifting a turn time at interval according to what occur in real time between described character data signal, upgrades and lifts a turn time threshold value;
Handwriting recognition device provided by the invention is by being provided with the time threshold update module, lift a turn time at interval according to what occur in real time between described character data signal, a turn time threshold value is lifted in renewal, thereby reach monitoring of lifting a turn time and adaptive learning in the user writing process, can be provided with at different user and lift a turn time threshold value accordingly, to adapt to the writing style of relative users, avoid occurring false judgment to the character end of input, improve character recognition efficient.
Usually, after definite current character input finishes, then can discern this character.When embodiment of the invention technical scheme is implemented, in order to guarantee the accuracy of Handwritten Digits Recognition, improve the efficient of character recognition, whether import in the process that finishes in definite current character, not only to investigate a current turn time interval of lifting, also need specifically to judge according to the written handwriting of current character.Therefore, in another embodiment of the present invention, as shown in Figure 5, described handwriting recognition device can also comprise:
Written handwriting data-signal acquisition module 405 is used to obtain all written handwriting data-signals of the described character of current accumulative total;
Written handwriting data-signal acquisition module in the embodiment of the invention can be lifted a turn time and triggers after confirming at interval and start current at first judge module, and then, by the length breadth ratio of second judge module from character, whether geometrical property aspects such as integrality are imported to finish to it is judged.
In order further to improve the accuracy that character input integrality is judged, in another embodiment of the present invention, as shown in Figure 6, described handwriting recognition device can also comprise:
Quadrant distribution acquisition module 407, all the written handwriting data-signals of described character that are used to obtain current accumulative total are in the distribution situation of writing four quadrants in interface;
The 3rd judge module 408 is used for judging whether described four quadrants in interface of writing all have the written handwriting data-signal, if determine that then described character input finishes.
Quadrant distribution acquisition module in the embodiment of the invention can trigger after second judge module obtains the length breadth ratio of current character and start, and then, further judge the integrality of current input character by the 3rd judge module.
In another embodiment of the present invention, as shown in Figure 7, can also comprise:
Lift a turn time threshold value lower limit module 409 is set, be used for when the described character input of false judgment finishes, utilizing a turn time threshold value lower limit of lifting that presets that a turn time threshold value of lifting of current acquisition is upgraded.
When determining that according to the turn time threshold value of lifting after upgrading current character is a complete character, but, the comprehensive degree of confidence that obtains according to the written handwriting of this current character is less than comprehensive confidence threshold value, when being an imperfect character, then be owing to too short lift the error in judgement that a turn time threshold value (can be made as Ta) causes.If lift a turn time threshold value Ta numerical value be provided with too small, so in user's input process in character the minibreak between the different strokes also can be identified as be that the intercharacter pen of lifting pauses, thereby cause error in judgement, this kind situation was become training usually.For fear of crossing the too short turn time threshold value of lifting that training causes, can preestablish long turn time threshold value lower limit Tb that lifts, after renewal lift the too short incompatibility practical application of a turn time threshold value time, utilize Tb to upgrade waiting time Ta.Lift a turn time threshold value lower limit Tb by setting, avoid causing too short irrational turn time threshold value of lifting because of the too short dead time of user.
Need to prove that when specifically implementing, described time threshold update module specifically can comprise:
First calculating sub module, be used for a turn time interval samples of lifting, according to second average and the second variance that a turn time distributes of lifting that presets after more new model obtains to upgrade according to history accumulative total turn time first average spaced apart and first variance and current appearance;
Second calculating sub module, be used for the Gaussian distribution model that a turn time satisfies is lifted in described second average and second variance substitution, and respectively lift the accumulated probability that presets that a turn time satisfies in the turn time threshold range according to current lifting, lift a turn time threshold value after obtaining upgrading.
As seen, handwriting recognition device provided by the invention can be lifted a turn time at interval according to what occur in real time between described character data signal, a turn time threshold value is lifted in renewal, thereby reach monitoring of lifting a turn time and adaptive learning in the user writing process, can be provided with at different user and lift a turn time threshold value accordingly, to adapt to the writing style of relative users; Simultaneously, distribution situation in conjunction with written handwriting four quadrants in writing the interface of the length breadth ratio of input character and input character, integrality to input character further judges, further avoids occurring the false judgment to the character end of input, improves character recognition efficient.
For device embodiment, because it is substantially corresponding to method embodiment, so describe fairly simplely, relevant part gets final product referring to the part explanation of method embodiment.Device embodiment described above only is schematic, wherein said unit as the separating component explanation can or can not be physically to separate also, the parts that show as the unit can be or can not be physical locations also, promptly can be positioned at a place, perhaps also can be distributed on a plurality of network element.Can select wherein some or all of module to realize the purpose of present embodiment scheme according to the actual needs.Those of ordinary skills promptly can understand and implement under the situation of not paying creative work.
One of ordinary skill in the art will appreciate that all or part of flow process that realizes in the foregoing description method, be to instruct relevant hardware to finish by computer program, described program can be stored in the computer read/write memory medium, this program can comprise the flow process as the embodiment of above-mentioned each side method when carrying out.Wherein, described storage medium can be magnetic disc, CD, read-only storage memory body (Read-Only Memory, ROM) or at random store memory body (Random Access Memory, RAM) etc.
To the above-mentioned explanation of the disclosed embodiments, make this area professional and technical personnel can realize or use the present invention.Multiple modification to these embodiment will be conspicuous concerning those skilled in the art, and defined herein General Principle can realize under the situation of the spirit or scope that do not break away from the embodiment of the invention in other embodiments.Therefore, the embodiment of the invention will can not be restricted to these embodiment shown in this article, but will meet and principle disclosed herein and features of novelty the wideest corresponding to scope.
Claims (10)
1. a hand-written recognition method is characterized in that, comprising:
Receive the character data signal of handwriting input;
Lift a turn time at interval according to what occur in real time between described character data signal, upgrade and lift a turn time threshold value;
Judge and current lift a turn time and whether surpass the turn time threshold value of lifting after the real-time update at interval, if determine that then the current character input finishes;
Current character is discerned.
2. hand-written recognition method according to claim 1 is characterized in that, described method also comprises:
Obtain all written handwriting signals of the described character of current accumulative total;
Determine the length breadth ratio of current character according to described written handwriting signal, judge by the length breadth ratio of described current character whether described character is imported and finish.
3. hand-written recognition method according to claim 2 is characterized in that, described method also comprises:
All the written handwriting data-signals of described character that obtain current accumulative total are in the distribution situation of writing four quadrants in interface;
Judge describedly write in four quadrants in interface whether the written handwriting data-signal is all arranged, if determine that then described character input finishes.
4. hand-written recognition method according to claim 3 is characterized in that, described method also comprises:
When the described character input of false judgment finishes, utilize a turn time threshold value lower limit of lifting that presets that a turn time threshold value of lifting of current acquisition is upgraded.
5. according to each described hand-written recognition method among the claim 1-4, it is characterized in that described according to a turn time interval of lifting that occurs in real time between described character data signal, a turn time threshold value is lifted in renewal, specifically comprises:
According to a turn time interval samples of lifting of history accumulative total turn time first average spaced apart and first variance and current appearance, according to second average and the second variance that a turn time distributes of lifting that presets after more new model obtains to upgrade;
The Gaussian distribution model that a turn time satisfies is lifted in described second average and second variance substitution, and respectively lift the accumulated probability that presets that a turn time satisfies in the turn time threshold range, lift a turn time threshold value after obtaining upgrading according to current lifting.
6. a handwriting recognition device is characterized in that, comprising:
Signal receiving module is used to receive the character data signal of handwriting input;
The time threshold update module is used for lifting a turn time at interval according to what occur in real time between described character data signal, upgrades and lifts a turn time threshold value;
First judge module is used to judge and current lift a turn time and whether surpass the turn time threshold value of lifting after the real-time update at interval, if determine that then the current character input finishes;
Identification module is used for current character is discerned.
7. handwriting recognition device according to claim 6 is characterized in that, also comprises:
Written handwriting data-signal acquisition module is used to obtain all written handwriting data-signals of the described character of current accumulative total;
Second judge module is used for determining according to described written handwriting data-signal the length breadth ratio of current character, judges by the length breadth ratio of described current character whether described character is imported and finishes.
8. handwriting recognition device according to claim 7 is characterized in that, also comprises:
Quadrant distribution acquisition module, all the written handwriting data-signals of described character that are used to obtain current accumulative total are in the distribution situation of writing four quadrants in interface;
The 3rd judge module is used for judging whether described four quadrants in interface of writing all have the written handwriting data-signal, if determine that then described character input finishes.
9. handwriting recognition device according to claim 8 is characterized in that, also comprises:
Lift a turn time threshold value lower limit module is set, be used for when the described character input of false judgment finishes, utilizing a turn time threshold value lower limit of lifting that presets that a turn time threshold value of lifting of current acquisition is upgraded.
10. according to each described handwriting recognition device among the claim 6-9, it is characterized in that described time threshold update module specifically comprises:
First calculating sub module, be used for a turn time interval samples of lifting, according to second average and the second variance that a turn time distributes of lifting that presets after more new model obtains to upgrade according to history accumulative total turn time first average spaced apart and first variance and current appearance;
Second calculating sub module, be used for the Gaussian distribution model that a turn time satisfies is lifted in described second average and second variance substitution, and respectively lift the accumulated probability that presets that a turn time satisfies in the turn time threshold range according to current lifting, lift a turn time threshold value after obtaining upgrading.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201010542942 CN101976152B (en) | 2010-11-12 | 2010-11-12 | Method and device of handwriting recognition |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201010542942 CN101976152B (en) | 2010-11-12 | 2010-11-12 | Method and device of handwriting recognition |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101976152A true CN101976152A (en) | 2011-02-16 |
| CN101976152B CN101976152B (en) | 2013-10-23 |
Family
ID=43576040
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 201010542942 Active CN101976152B (en) | 2010-11-12 | 2010-11-12 | Method and device of handwriting recognition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101976152B (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102750106A (en) * | 2012-07-02 | 2012-10-24 | 安徽科大讯飞信息科技股份有限公司 | Full-screen handwriting identification input method and system |
| CN103167098A (en) * | 2011-12-15 | 2013-06-19 | 联想(北京)有限公司 | Communication method and communication device and communication terminals |
| CN104881218A (en) * | 2015-04-30 | 2015-09-02 | 努比亚技术有限公司 | Mobile terminal screen scrolling method and mobile terminal screen scrolling device |
| CN109858323A (en) * | 2018-12-07 | 2019-06-07 | 广州光大教育软件科技股份有限公司 | A kind of character hand-written recognition method and system |
| CN110795986A (en) * | 2019-07-26 | 2020-02-14 | 广东小天才科技有限公司 | Handwriting input recognition method and system |
| CN111063223A (en) * | 2020-01-07 | 2020-04-24 | 杭州大拿科技股份有限公司 | English word spelling practice method and device |
| CN111538417A (en) * | 2020-04-17 | 2020-08-14 | 京东方科技集团股份有限公司 | Control method for character input, touch pen, display device, system and medium |
| CN111722727A (en) * | 2019-03-22 | 2020-09-29 | 北京搜狗科技发展有限公司 | Model training method applied to handwriting input, handwriting input method and device |
| CN112215178A (en) * | 2020-10-19 | 2021-01-12 | 南京大学 | Chemical experiment recording system based on pen type interaction |
| CN114942714A (en) * | 2022-05-24 | 2022-08-26 | 深圳市奥振电子科技有限公司 | Intelligent auxiliary method, system and storage medium for electronic whiteboard |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105678126A (en) * | 2014-11-20 | 2016-06-15 | 阿里巴巴集团控股有限公司 | Handwriting processing method and device |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1991723A (en) * | 2005-12-28 | 2007-07-04 | 中兴通讯股份有限公司 | Method for realizing automatic hand-write input |
| CN1991699A (en) * | 2005-12-28 | 2007-07-04 | 中兴通讯股份有限公司 | Method for realizing hand-write input |
-
2010
- 2010-11-12 CN CN 201010542942 patent/CN101976152B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1991723A (en) * | 2005-12-28 | 2007-07-04 | 中兴通讯股份有限公司 | Method for realizing automatic hand-write input |
| CN1991699A (en) * | 2005-12-28 | 2007-07-04 | 中兴通讯股份有限公司 | Method for realizing hand-write input |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103167098A (en) * | 2011-12-15 | 2013-06-19 | 联想(北京)有限公司 | Communication method and communication device and communication terminals |
| CN102750106A (en) * | 2012-07-02 | 2012-10-24 | 安徽科大讯飞信息科技股份有限公司 | Full-screen handwriting identification input method and system |
| CN102750106B (en) * | 2012-07-02 | 2016-05-18 | 安徽科大讯飞信息科技股份有限公司 | Full frame handwriting recognition input method and system |
| CN104881218A (en) * | 2015-04-30 | 2015-09-02 | 努比亚技术有限公司 | Mobile terminal screen scrolling method and mobile terminal screen scrolling device |
| CN109858323A (en) * | 2018-12-07 | 2019-06-07 | 广州光大教育软件科技股份有限公司 | A kind of character hand-written recognition method and system |
| CN111722727A (en) * | 2019-03-22 | 2020-09-29 | 北京搜狗科技发展有限公司 | Model training method applied to handwriting input, handwriting input method and device |
| CN111722727B (en) * | 2019-03-22 | 2024-01-12 | 北京搜狗科技发展有限公司 | Model training method applied to handwriting input, handwriting input method and device |
| CN110795986A (en) * | 2019-07-26 | 2020-02-14 | 广东小天才科技有限公司 | Handwriting input recognition method and system |
| CN111063223A (en) * | 2020-01-07 | 2020-04-24 | 杭州大拿科技股份有限公司 | English word spelling practice method and device |
| CN111063223B (en) * | 2020-01-07 | 2022-02-08 | 杭州大拿科技股份有限公司 | English word spelling practice method and device |
| CN111538417A (en) * | 2020-04-17 | 2020-08-14 | 京东方科技集团股份有限公司 | Control method for character input, touch pen, display device, system and medium |
| CN111538417B (en) * | 2020-04-17 | 2024-06-18 | 京东方科技集团股份有限公司 | Character input control method, touch pen, display device, system and medium |
| CN112215178A (en) * | 2020-10-19 | 2021-01-12 | 南京大学 | Chemical experiment recording system based on pen type interaction |
| CN112215178B (en) * | 2020-10-19 | 2024-05-28 | 南京大学 | Chemical experiment recording system based on pen type interaction |
| CN114942714A (en) * | 2022-05-24 | 2022-08-26 | 深圳市奥振电子科技有限公司 | Intelligent auxiliary method, system and storage medium for electronic whiteboard |
| CN114942714B (en) * | 2022-05-24 | 2024-07-09 | 深圳市奥振电子科技有限公司 | Intelligent auxiliary method, system and storage medium for electronic whiteboard |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101976152B (en) | 2013-10-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101976152B (en) | Method and device of handwriting recognition | |
| US10664695B2 (en) | System and method for managing digital ink typesetting | |
| KR102383624B1 (en) | System and method for superimposed handwriting recognition technology | |
| US9384403B2 (en) | System and method for superimposed handwriting recognition technology | |
| US8879845B2 (en) | Character recognition for overlapping textual user input | |
| CN100587660C (en) | Method and device for forecasting and discriminating hand-written characters | |
| US8667414B2 (en) | Gestural input at a virtual keyboard | |
| US20190235749A1 (en) | Progress display of handwriting input | |
| CN108700996B (en) | System and method for multiple input management | |
| EP2088536A1 (en) | Text input system and method involving finger-based handwriting recognition and word prediction | |
| EP2960761A1 (en) | Method, system and device for inputting text by consecutive slide | |
| CN105488544A (en) | Method and system for tracing handwriting by tracing in red | |
| CN102184074B (en) | Keyboard, automatic hand input switching method and device and mobile terminal | |
| WO2006091156A1 (en) | Recognition graph | |
| WO2006061852A1 (en) | System, method, and apparatus for triggering recognition of a handwritten shape | |
| CN102063620A (en) | Handwriting identification method, system and terminal | |
| US20140219565A1 (en) | Multiple-input handwriting recognition system and measure thereof | |
| CN108701215A (en) | The system and method for multipair image structures for identification | |
| US11393231B2 (en) | System and method for text line extraction | |
| KR20220103102A (en) | Gesture Stroke Recognition in Touch-Based User Interface Input | |
| CN110491394A (en) | Wake up the acquisition methods and device of corpus | |
| CN109977737A (en) | A kind of character recognition Robust Method based on Recognition with Recurrent Neural Network | |
| WO2017176470A2 (en) | Faster text entry on mobile devices through user-defined stroke patterns | |
| US20230096728A1 (en) | System and method for text line and text block extraction | |
| CN1142471C (en) | Method and apparatus for operation by hand written alphabets and symbols |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CP03 | Change of name, title or address | ||
| CP03 | Change of name, title or address |
Address after: 230031 666 Wangjiang West Road, Hefei high tech Zone, Anhui Patentee after: Iflytek Co., Ltd. Address before: 230088 No. 616, Mount Huangshan Road, hi tech Development Zone, Anhui, Hefei Patentee before: Anhui USTC iFLYTEK Co., Ltd. |