CN110134257A - Electronic pen and its writing identification model method for building up based on inertial sensor - Google Patents
Electronic pen and its writing identification model method for building up based on inertial sensor Download PDFInfo
- Publication number
- CN110134257A CN110134257A CN201910383907.3A CN201910383907A CN110134257A CN 110134257 A CN110134257 A CN 110134257A CN 201910383907 A CN201910383907 A CN 201910383907A CN 110134257 A CN110134257 A CN 110134257A
- Authority
- CN
- China
- Prior art keywords
- electronic pen
- pen
- gyroscope
- identification model
- acceleration transducer
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
- G06F3/03—Arrangements for converting the position or the displacement of a member into a coded form
- G06F3/033—Pointing devices displaced or positioned by the user, e.g. mice, trackballs, pens or joysticks; Accessories therefor
- G06F3/0346—Pointing devices displaced or positioned by the user, e.g. mice, trackballs, pens or joysticks; Accessories therefor with detection of the device orientation or free movement in a 3D space, e.g. 3D mice, 6-DOF [six degrees of freedom] pointers using gyroscopes, accelerometers or tilt-sensors
Abstract
The invention discloses the electronic pen based on inertial sensor and its identification model method for building up is write, the electronic pen includes the pen core for being set to nib;It is set to inside nib, connects the spring of pen core;The point-touch switch being connected with spring;Acceleration transducer inside pen body and close to described written one end, and the gyroscope of connection acceleration transducer;Connect the power supply of gyroscope and point-touch switch;Connect the MCU of gyroscope and acceleration transducer;With the wireless transmitter module of connection MCU;Wherein, acceleration transducer and gyroscope acquire writing track data, and the MCU writes identification model by electronic pen and carries out input Text region.Compared with prior art, the present invention does not need that input can be realized by other ancillary equipments, small and exquisite convenient, can be used for the input of each class of electronic devices, and can be write in arbitrary plane.
Description
Technical field
The present invention relates to a kind of electronic pens, more particularly to the electronic pen based on inertial sensor and its writing identification model are built
Cube method.
Background technique
Existing electronic pen is predominantly based on multiple sensors (such as force snesor, inertial sensor, touch sensor)
Stylus and writing pencil based on Chinese character database, and most popular capacitor or voltage-type handwriting pad writing pencil at present.
However existing electronic pen often has the disadvantage that:
(1) integrated circuit is more, and volume is larger or overweight, inconvenient to use;
(2) it is only applicable to particular device, cannot be popularized;
(3) additional hardware equipment (such as handwriting pad) is needed to be used cooperatively, efficiency is lower.
The prior art such as " a kind of based on electromagnetic touch-control double write electronic pen "-CN201520998617.7 proposes one kind
Double to write pen, it can be completed at the same time paper record and corresponding notes electronical record data by realizing, pass through special oil
While ink realizes that papery is write, realize based on electromagnetic touch device progress electronic writing, however the reality of the technical solution
It is now that record notes information is carried out based on electromagnetic type principle, it is therefore desirable to electromagnetic touch-control screen be cooperated to be used cooperatively;A kind of " letter
Breath processing method and electronic pen "-CN201310347517.3 proposes a kind of electronic pen, passes through inertial sensor and obtains electronic pen
Tilt angle obtains user to the grip position of electronic pen, to obtain the tilt angle and handwriting thickness of electronic pen by camera
Parameter needs additional hardware facility to cooperate, and efficiency is lower.
Summary of the invention
For overcome the problems, such as in the prior art electronic pen must mating handwriting pad or conventional belt ink electronic pen change ink and
The problem of fluency difference, the present invention propose a kind of electronic pen based on inertial sensor, and more environmental protection is practical, more convenient to relax
It is suitable, it directly can replace keyboard and mouse device.
The technical scheme is that being achieved: a kind of electronic pen based on inertial sensor, including
It is set to the pen core of nib;
It is set to inside nib, connects the spring of the pen core;
The point-touch switch being connected with the spring;
Acceleration transducer inside pen body and close to described written one end, and the connection acceleration transducer
Gyroscope;
Connect the power supply of the gyroscope and the point-touch switch;
Connect the MCU of the gyroscope and acceleration transducer;With
Connect the wireless transmitter module of the MCU;
Wherein, the acceleration transducer and the gyroscope acquire writing track data, the electricity built in the MCU use
Sub- pen writes identification model and carries out Text region to collected writing track data.
Further, the acceleration transducer is 3-axis acceleration sensor, and the gyroscope is three-axis gyroscope.
Further, the wireless transmitter module is bluetooth module or WIFI module.
Further, buffer unit is equipped between the pen core and nib, the buffer unit alleviates pen tip and crosses stiff ribbon
The discomfort come, enhances the fluency of writing.
A kind of electronic pen writes the method for building up of identification model, including
Step 1: data collector's hand-held electronic pen is write in two-dimensional space (such as desktop, paper etc.), in electronic pen
Acceleration transducer and gyroscope the acquisition electronic pen writing track data set;
Step 2: collected electronic pen writing track data is pre-processed, and is rejected unrelated electronic pen and is write rail
Mark data, and normalized regular to electronic pen writing track data progress data, remove gravity and component of acceleration
Interference;
Step 3: electronic pen writing track data collection of the building based on acceleration transducer and gyroscope;
Step 4: building electronic pen writes identification model, uses the electronic pen writing track data collection training electricity
Sub- pen writes identification model.
Further, data described in step 2 are regular comprising steps of being write using interpolation method in collected electronic pen
The beginning or ending of track data carry out interpolation.
Further, it is LSTM model that electronic pen described in step 4, which writes identification model,.
The beneficial effects of the present invention are compared with prior art, the present invention is not needed by other ancillary equipments (such as electricity
Resistance screen or capacitance plate) input can be realized, it is small and exquisite convenient, it can be used for the input of each class of electronic devices, and can be in arbitrary plane
It is write.
Detailed description of the invention
Fig. 1 is a kind of structural schematic diagram of electronic pen one embodiment based on inertial sensor of the present invention;
Fig. 2 is that electronic pen writes Arabic numerals stroke writing schematic diagram in the establishment process of identification model in the present invention;
Fig. 3 is the LSTM model signal that electronic pen writes in the establishment process of identification model in one embodiment of the invention
Figure.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
Referring to Figure 1, when user's hand electronic pen is planar write, the pen tip of pen core is when writing, the pressure that is subject to
It is transferred to spring, promotes spring that elastic deformation occurs, pen core is mobile towards electronic pen inside;When camber of spring reaches a certain level
Promote pen core to trigger point-touch switch, close the switch, power supply is initially powered up, 3-axis acceleration sensor and three-axis gyroscope sensing
Device will start to acquire data.When user is when writing a number, Chinese character or letter, hand force measurement disappears, and electronic pen lifts,
Spring occurs to restore deformation in the case where non-stress, makes pen core vertical downward movement, and point-touch switch disconnects, and power supply stops logical
Electricity, 3-axis acceleration sensor and three-axis gyroscope sensor will stop data acquisition.In electronic pen electric power starting and close this
In one time electronic pen acquire data be electronic pen complete one input all data, these input can be write Ah
Arabic numbers, letter and Chinese character etc..Electronic pen is moved rail by 3-axis acceleration sensor built in electronic pen and three-axis gyroscope
Mark data collection is simultaneously transmitted to MCU, and trained electronic pen writes identification model for the motion profile of quick identification electronic pen in MCU
Data, and recognition result is transferred in equipment by the wireless transmitter module of electronic pen, such as bluetooth or wifi, to complete electricity
The entire input function of sub- pen.
Wherein, electronic pen writes the establishment process of identification model are as follows:
Step 1: data collector's hand-held electronic pen is write in two-dimensional space (such as desktop, paper etc.), in electronic pen
Acceleration transducer and gyroscope the acquisition electronic pen writing track data set;
Step 2: collected electronic pen writing track data being pre-processed, unrelated electronic pen handwriting trace is rejected
Data, and normalized regular to electronic pen writing track data progress data, removes gravity and component of acceleration is dry
It disturbs;
Step 3: electronic pen writing track data collection of the building based on acceleration transducer and gyroscope;
Step 4: building electronic pen writes identification model, uses the electronic pen writing track data collection training electronics
Pen writes identification model.
Wherein, in step 1, data acquisition has selected certain university student enrollment, age distribution 18 years old to 30 years old it
Between, for male to female ratio close to 1:1, this can be very good interference brought by the difference for excluding to be accustomed to as male, schoolgirl's behavior act.
In addition, data collector is university student from various parts of the country, this can be very good to be avoided causing behavior dynamic due to regional factor
Make habit bring interference.
Meanwhile in data acquisition, each Arabic numerals, phonetic alphabet, Chinese character and other characters can one write as,
As Arabic numerals stroke writing can as shown in Figure 2 one write as.
In step 2, due to 3-axis acceleration sensor and three-axis gyroscope manufacture craft the problem of, even causing same
Also difference, the columns of obtained initial data have been 3 column to the data length of one electronic pen acquisition, and line number is not more than n row (phase
The data length acquired under the conditions of slightly has difference, and n row here is in all data collector smart phone acquisition data
Maximum length), n and frequency acquisition are positively correlated.By by collected data visualization, it is seen that in each data
Critical data is all distributed in middle section, therefore carries out interpolation in the beginning of data or ending using interpolation method, to be counted
According to regular.
Acceleration transducer is identical as the method that gyro data requires to do normalized, this is because in acquisition number
During, each data collector speed and amplitude of hand exercise during writing number is different.In order to disappear
Except the interference of these components, it is necessary to be normalized to pre-process collected all electronic pen writing track datas using numerical value.
By formulaThe data conversion of any value range can be turned in [0,1] section
Numerical value, wherein oldValue and newValue respectively refers to the numerical value after raw value and normalization, and max and min respectively refer to sense
The a certain axis of device (x, y or z-axis) collects the maximum value and minimum value of data.
Gyroscope sensor data does not need to remove other components, and acceleration needs to remove weight component, this is because
The acceleration information acquired from smart phone consists of two parts: motional induction component of acceleration and gravitational acceleration component.Weight
Force component is considered as noise, should be removed, because of its hand exercise independent of user.It can be calculated by formula (1)~(6)
Linear acceleration out, i.e. hand exercise acceleration after removal weight component
Gravity_x=alpha*gravity+ (1-alpha) * x_values (1)
Gravity_y=alpha*gravity+ (1-alpha) * y_values (2)
Gravity_z=alpha*gravity+ (1-alpha) * z_values (3)
Linear_acceleration_x=x_values-gravity_x (4)
Linear_acceleration_y=y_values-gravity_y (5)
Linear_acceleration_z=z_values-gravity_z (6)
Wherein, alpha=0.8, gravity=9.8m2/s;* _ values is initial data, and gravity_* is gravity point
Amount, linear_acceleration_* linear acceleration,*Refer to x, y or z.
There are two text files for each input data, and one is gyro data, the other is acceleration transducer number
According to, and each text file data are that n row 3 arranges.The data of this data set are also by each each number of data collector
Write 10 times compositions.
Since Arabic numerals, phonetic alphabet, Chinese character and other characters are numerous, the data set workload of all inputs is constructed
It is very big.Therefore, can be constructed by this patent the method early period common Arabic numerals, phonetic alphabet, Chinese character and its
The data set of its character;Later period is continuously increased number of data sets evidence by crowdsourcing model;Finally, personal user uses electronic pen
When, electronic pen meeting automatic on-line training electronic pen identification model simultaneously modifies model parameter, constructs user according to user's writing style
Personal input identification model, to promote the accuracy rate of isolated user identification model.
It is typical non-stationary signal that electronic pen based on inertial sensor, which writes identification signal, belongs to the typical time
Sequence or sequence problem analysis, feature is complicated, various, and key feature is difficult to extract, and the feature only manually extracted is often difficult
Adequately to be expressed original signal.LSTM is very suitable to handle this problem, thus this patent establish it is hidden more than one
Deep learning model containing layer, multi-neuron node extracts the key feature of electronic pen acceleration, and realizes to electronic pen
The identification of input.The LSTM model of foundation is as shown in Figure 3.
It is write in identification model training process in the electronic pen, data set is divided into two parts, 70% data are as instruction
Practice collection, 30% is used as test set.The parameter that model is mainly trained is the number of plies of hidden layer and of every layer of hidden layer neural unit
Number, with the model of training set training different depth and width until finding accuracy rate best model.
The above is a preferred embodiment of the present invention, it is noted that for those skilled in the art
For, various improvements and modifications may be made without departing from the principle of the present invention, these improvements and modifications are also considered as
Protection scope of the present invention.
Claims (7)
1. a kind of electronic pen based on inertial sensor, which is characterized in that including
It is set to the pen core of nib;
It is set to inside nib, connects the spring of the pen core;
The point-touch switch being connected with the spring;
Acceleration transducer inside pen body and close to described written one end, and the gyro of the connection acceleration transducer
Instrument;
Connect the power supply of the gyroscope and the point-touch switch;
Connect the MCU of the gyroscope and acceleration transducer;With
Connect the wireless transmitter module of the MCU;
Wherein, the acceleration transducer and the gyroscope acquire writing track data, the electronic pen built in the MCU use
It writes identification model and Text region is carried out to collected writing track data.
2. as described in claim 1 based on the electronic pen of inertial sensor, which is characterized in that the acceleration transducer is three
Axle acceleration sensor, the gyroscope are three-axis gyroscope.
3. as described in claim 1 based on the electronic pen of inertial sensor, which is characterized in that the wireless transmitter module is indigo plant
Tooth module or WIFI module.
4. as described in claim 1 based on the electronic pen of inertial sensor, which is characterized in that set between the pen core and nib
There is buffer unit, the buffer unit enhances the fluency of writing for mitigating pen tip bring discomfort really up to the mark.
5. a kind of method established electronic pen and write identification model, which is characterized in that including
Step 1: data collector's hand-held electronic pen is write in two-dimensional space, the acceleration transducer built in electronic pen and
Gyroscope acquires electronic pen writing track data;
Step 2: collected electronic pen writing track data is pre-processed, and rejects unrelated electronic pen handwriting trace number
According to, and normalized regular to electronic pen writing track data progress data, removal gravitational acceleration component interference;
Step 3: electronic pen writing track data collection of the building based on acceleration transducer and gyroscope;
Step 4: building electronic pen writes identification model, uses the electronic pen writing track data collection training electronic pen
Write identification model.
6. establishing the method that electronic pen writes identification model as claimed in claim 5, which is characterized in that number described in step 2
According to regular comprising steps of carrying out interpolation in the beginning of collected electronic pen writing track data or ending using interpolation method.
7. establishing the method that electronic pen writes identification model as claimed in claim 5, which is characterized in that electricity described in step 4
It is LSTM model that sub- pen, which writes identification model,.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910383907.3A CN110134257A (en) | 2019-05-09 | 2019-05-09 | Electronic pen and its writing identification model method for building up based on inertial sensor |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910383907.3A CN110134257A (en) | 2019-05-09 | 2019-05-09 | Electronic pen and its writing identification model method for building up based on inertial sensor |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110134257A true CN110134257A (en) | 2019-08-16 |
Family
ID=67576923
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910383907.3A Pending CN110134257A (en) | 2019-05-09 | 2019-05-09 | Electronic pen and its writing identification model method for building up based on inertial sensor |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110134257A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111169201A (en) * | 2020-03-04 | 2020-05-19 | 黑龙江大学 | Calligraphy practicing monitor and monitoring method |
EP3989114A1 (en) * | 2020-10-26 | 2022-04-27 | Société BIC | System and method for recognizing online handwriting |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH1153104A (en) * | 1996-10-11 | 1999-02-26 | Ricoh Co Ltd | Pen-type input device and pattern recognition method for the pen-type input device |
CN101872259A (en) * | 2010-06-03 | 2010-10-27 | 中国人民解放军第二炮兵工程学院 | Natural interaction pen capable of automatically inputting written contents and handwriting detection method |
CN109263362A (en) * | 2018-10-29 | 2019-01-25 | 广东小天才科技有限公司 | A kind of smart pen and its control method |
CN210072562U (en) * | 2019-05-09 | 2020-02-14 | 兰州工业学院 | Electronic pen based on inertial sensor |
-
2019
- 2019-05-09 CN CN201910383907.3A patent/CN110134257A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH1153104A (en) * | 1996-10-11 | 1999-02-26 | Ricoh Co Ltd | Pen-type input device and pattern recognition method for the pen-type input device |
CN101872259A (en) * | 2010-06-03 | 2010-10-27 | 中国人民解放军第二炮兵工程学院 | Natural interaction pen capable of automatically inputting written contents and handwriting detection method |
CN109263362A (en) * | 2018-10-29 | 2019-01-25 | 广东小天才科技有限公司 | A kind of smart pen and its control method |
CN210072562U (en) * | 2019-05-09 | 2020-02-14 | 兰州工业学院 | Electronic pen based on inertial sensor |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111169201A (en) * | 2020-03-04 | 2020-05-19 | 黑龙江大学 | Calligraphy practicing monitor and monitoring method |
CN111169201B (en) * | 2020-03-04 | 2024-03-26 | 黑龙江大学 | Handwriting practicing monitor and monitoring method |
EP3989114A1 (en) * | 2020-10-26 | 2022-04-27 | Société BIC | System and method for recognizing online handwriting |
EP3989108A1 (en) * | 2020-10-26 | 2022-04-27 | Invoxia | System and method for recognizing online handwriting |
WO2022090096A1 (en) * | 2020-10-26 | 2022-05-05 | SOCIéTé BIC | System for recognizing online handwriting |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Hsu et al. | An inertial pen with dynamic time warping recognizer for handwriting and gesture recognition | |
Amma et al. | Airwriting: a wearable handwriting recognition system | |
Amma et al. | Airwriting recognition using wearable motion sensors | |
TW403877B (en) | Handwriting input apparatus using more than one sensing technique | |
CN103257711B (en) | space gesture input method | |
CN101882000B (en) | Gesture identification method based on acceleration sensor | |
CN107678550A (en) | A kind of sign language gesture recognition system based on data glove | |
Bui et al. | Recognizing postures in Vietnamese sign language with MEMS accelerometers | |
KR20050085897A (en) | Method and system for three-dimentional handwriting recognition | |
CN104385808A (en) | Intelligent original handwriting recording pen based on inertia measurement | |
Lee et al. | On-line handwritten character recognition with 3D accelerometer | |
Jeen-Shing et al. | Online handwriting recognition using an accelerometer-based pen device | |
CN103995600B (en) | A kind of braille Chinese character converter and its method | |
Xu et al. | Air-writing characters modelling and recognition on modified CHMM | |
CN110134257A (en) | Electronic pen and its writing identification model method for building up based on inertial sensor | |
Luo et al. | Wearable air-writing recognition system employing dynamic time warping | |
CN109002803A (en) | A kind of pen holding posture detection and Chinese characters' order of strokes recognition method based on smartwatch | |
CN107204141A (en) | Wearable Braille reading equipment | |
CN202649946U (en) | Multifunctional aerial writing drawing pen | |
Li et al. | Hand gesture recognition and real-time game control based on a wearable band with 6-axis sensors | |
Bi et al. | SmartGe: identifying pen-holding gesture with smartwatch | |
KR20150063699A (en) | Smart Pen | |
US20210345913A1 (en) | System and Method for Detecting Handwriting Problems | |
CN210072562U (en) | Electronic pen based on inertial sensor | |
CN103295570A (en) | Glove type sound production system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |