CN109663323A - A kind of motion state monitoring method and wearable device based on wearable device - Google Patents
A kind of motion state monitoring method and wearable device based on wearable device Download PDFInfo
- Publication number
- CN109663323A CN109663323A CN201811565345.6A CN201811565345A CN109663323A CN 109663323 A CN109663323 A CN 109663323A CN 201811565345 A CN201811565345 A CN 201811565345A CN 109663323 A CN109663323 A CN 109663323A
- Authority
- CN
- China
- Prior art keywords
- motion
- motion state
- sensing data
- pattern
- peak 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.)
- Pending
Links
Classifications
-
- A—HUMAN NECESSITIES
- A63—SPORTS; GAMES; AMUSEMENTS
- A63B—APPARATUS FOR PHYSICAL TRAINING, GYMNASTICS, SWIMMING, CLIMBING, OR FENCING; BALL GAMES; TRAINING EQUIPMENT
- A63B71/00—Games or sports accessories not covered in groups A63B1/00 - A63B69/00
- A63B71/06—Indicating or scoring devices for games or players, or for other sports activities
- A63B71/0619—Displays, user interfaces and indicating devices, specially adapted for sport equipment, e.g. display mounted on treadmills
-
- A—HUMAN NECESSITIES
- A63—SPORTS; GAMES; AMUSEMENTS
- A63B—APPARATUS FOR PHYSICAL TRAINING, GYMNASTICS, SWIMMING, CLIMBING, OR FENCING; BALL GAMES; TRAINING EQUIPMENT
- A63B2220/00—Measuring of physical parameters relating to sporting activity
- A63B2220/40—Acceleration
-
- A—HUMAN NECESSITIES
- A63—SPORTS; GAMES; AMUSEMENTS
- A63B—APPARATUS FOR PHYSICAL TRAINING, GYMNASTICS, SWIMMING, CLIMBING, OR FENCING; BALL GAMES; TRAINING EQUIPMENT
- A63B2220/00—Measuring of physical parameters relating to sporting activity
- A63B2220/80—Special sensors, transducers or devices therefor
- A63B2220/803—Motion sensors
-
- A—HUMAN NECESSITIES
- A63—SPORTS; GAMES; AMUSEMENTS
- A63B—APPARATUS FOR PHYSICAL TRAINING, GYMNASTICS, SWIMMING, CLIMBING, OR FENCING; BALL GAMES; TRAINING EQUIPMENT
- A63B2225/00—Miscellaneous features of sport apparatus, devices or equipment
- A63B2225/60—Apparatus used in water
-
- A—HUMAN NECESSITIES
- A63—SPORTS; GAMES; AMUSEMENTS
- A63B—APPARATUS FOR PHYSICAL TRAINING, GYMNASTICS, SWIMMING, CLIMBING, OR FENCING; BALL GAMES; TRAINING EQUIPMENT
- A63B2244/00—Sports without balls
- A63B2244/20—Swimming
Abstract
The embodiment of the invention discloses a kind of motion state monitoring method and wearable device based on wearable device, it is input to by the characteristic information for the acceleration information window that will acquire in motion pattern classification model trained in advance to obtain motion pattern classification, and motion state parameters are determined according to the motion pattern classification and motion-sensing data, thus, the generalization ability and robustness of motion state monitoring method can be improved, and improve the accuracy rate of motion state monitoring.
Description
Technical field
The present invention relates to technical field of data processing, more particularly, to a kind of motion state based on wearable device
Monitoring method and wearable device.
Background technique
In the conventional technology, to the monitoring of the motion state of sporter (for example, swimming state) mainly pass through vision into
Row, is analyzed and is identified according to video data later, this scheme cannot monitor motion state in real time and count identification
As a result.Currently, generally use intelligent wearable device (for example, Intelligent bracelet, smart motion wrist-watch etc.) to monitor motion state,
But it is how accurate and monitor and export motion state parameters in real time, it is still current urgent problem to be solved.
Summary of the invention
In view of this, the embodiment of the present invention provides a kind of motion state monitoring method based on wearable device and wearable
Equipment to improve the generalization ability and robustness of motion state monitoring method, and improves the accuracy rate of motion state monitoring.
In a first aspect, the embodiment of the present invention provides a kind of motion state monitoring method based on wearable device, the side
Method includes:
Acceleration information window is obtained, the acceleration information window has scheduled time span；
Obtain the characteristic information of the acceleration information window；
The characteristic information is input in motion pattern classification model trained in advance to obtain motion pattern classification knot
Fruit, the motion pattern classification result include non-moving pattern and at least one motor pattern；
Motion state parameters are determined according to the motion pattern classification result and motion-sensing data；And
Export the motor pattern and the motion state parameters in current kinetic period.
Further, the characteristic information for obtaining the acceleration information window includes:
The motion-sensing data of the acceleration information window are obtained by inertial measuring unit；
According to the characteristic information of acceleration information window described in the motion-sensing data acquisition, the characteristic information includes
Related coefficient on three axis between the minimum value, maximum value of acceleration, average value, standard deviation and two axis.
Further, the motion state parameters include that the swimming in current kinetic period is struck number；
It is described to determine that motion state parameters include: according to the motion pattern classification result and motion-sensing data
The peak value number for obtaining the motion-sensing data is struck according to the swimming that the peak value number obtains current period
Number.
Further, the peak value number for obtaining the motion-sensing data includes:
It is greater than first threshold in response to the peak value, and the time interval of the peak value and previous peak value is greater than the second threshold
Value, count is incremented for peak value number.
Further, the motion state parameters include the number of touch turn；
It is described to determine that motion state parameters include: according to the motion pattern classification result and motion-sensing data
Obtain the number of touch turn.
Further, the number for obtaining touch turn includes:
In response to the motion pattern classification result be non-moving pattern and the non-moving pattern duration is less than predetermined connect
The continuous time is greater than steering angle threshold value in the current steering angle and the swimming in current kinetic period strikes number in pre-determined number
When in range, the number of touch turn adds 1；
Wherein, the steering angle is calculated by the motion-sensing data and is obtained, and the pre-determined number range is according to movement
Period dynamic updates.
Further, the output motion pattern classification result includes:
The motion pattern classification of multiple acceleration information windows of current period is recorded as a result, and to the movement mould
The corresponding motor pattern of formula classification results counts respectively；
Output counts highest motor pattern.
Further, the method also includes:
It is non-moving pattern within predetermined continuous time in response to the motion pattern classification result, determines movement knot
Beam.
Further, the type of sports is swimming exercise, and the motor pattern includes breaststroke, butterfly stroke, backstroke and freedom
Swimming.
Second aspect, the embodiment of the present invention provide a kind of wearable device, and inertia survey is provided in the wearable device
Device, processor and memory are measured, the memory is for storing one or more computer program instructions, wherein described one
Item or a plurality of computer program instructions are executed by the processor to realize method as described above.
The characteristic information that the technical solution of the embodiment of the present invention passes through the acceleration information window that will acquire is input in advance
To obtain motion pattern classification in trained motion pattern classification model, and according to the motion pattern classification and motion-sensing number
According to motion state parameters are determined, thus, it is possible to improve the generalization ability and robustness of motion state monitoring method, and movement is improved
The accuracy rate of status monitoring.
Detailed description of the invention
By referring to the drawings to the description of the embodiment of the present invention, the above and other purposes of the present invention, feature and
Advantage will be apparent from, in the accompanying drawings:
Fig. 1 is the flow chart of the motion state monitoring method of the embodiment of the present invention；
Fig. 2 is the schematic diagram of the peak value of the motion-sensing data of the embodiment of the present invention；
Fig. 3 is the data flow figure of the method for monitoring operation states of the embodiment of the present invention；
Fig. 4 is the schematic diagram of the wearable device of the embodiment of the present invention.
Specific embodiment
Below based on embodiment, present invention is described, but the present invention is not restricted to these embodiments.Under
Text is detailed to describe some specific detail sections in datail description of the invention.Do not have for a person skilled in the art
The present invention can also be understood completely in the description of these detail sections.In order to avoid obscuring essence of the invention, well known method, mistake
There is no narrations in detail for journey, process, element and circuit.
In addition, it should be understood by one skilled in the art that provided herein attached drawing be provided to explanation purpose, and
What attached drawing was not necessarily drawn to scale.
Unless the context clearly requires otherwise, "include", "comprise" otherwise throughout the specification and claims etc. are similar
Word should be construed as the meaning for including rather than exclusive or exhaustive meaning；That is, be " including but not limited to " contains
Justice.
In the description of the present invention, it is to be understood that, term " first ", " second " etc. are used for description purposes only, without
It can be interpreted as indication or suggestion relative importance.In addition, in the description of the present invention, unless otherwise indicated, the meaning of " multiple "
It is two or more.
Below for the embodiment of the present invention description by taking swimming exercise as an example, it should be appreciated that the embodiment of the present invention can also answer
For other type of sports.
Fig. 1 is the flow chart of the motion state monitoring method of the embodiment of the present invention.As shown in Figure 1, the movement of the present embodiment
State monitoring method the following steps are included:
Step S100 obtains acceleration information window.Wherein, acceleration information window has scheduled time span.It is excellent
Selection of land, the time span of acceleration information window are 2 seconds.Wherein, the time span of acceleration information window can be according to data
The sample rate of sampling is arranged.In an optional implementation manner, adjacent acceleration information window has scheduled overlapping time,
So that the corresponding event to be identified of motion-sensing data of the acceleration information window of sampling has continuity, thus, it is possible to
Improve the accuracy rate of moving state identification.For example, the time span of an acceleration information window is 2 seconds, adjacent two add
Speed data window has 1 second overlapping time.
Step S200 obtains the characteristic information of acceleration information window.In an optional implementation manner, by can
Built-in IMU (Inertial measurement unit, inertial measuring unit) obtains acceleration information window in wearable device
Motion-sensing data, and according to the characteristic information of motion-sensing data acquisition acceleration information window.Wherein, motion-sensing data
Including the acceleration in three axis directions ((namely x-axis, y-axis, z-axis)) and angular speed when movement etc..Acceleration information window
Characteristic information includes but is not limited to the correlation on three axis between the minimum value, maximum value of acceleration, average value, standard deviation and two axis
Coefficient.Specifically, on wearable device be configured with inertial measuring unit, inertial measuring unit include 3-axis acceleration sensor,
Gyroscope and/or magnetometer.Wherein, 3-axis acceleration sensor obtains user's three axis directions in motion state for acquiring
Acceleration, gyroscope, which is used to acquire, obtains angular speed of the user in motion state, and magnetometer exists for acquiring acquisition user
Magnetic induction intensity when motion state.
The characteristic information of acceleration information window is input in motion pattern classification model trained in advance by step S300
To obtain the motion pattern classification result of current acceleration data window.Wherein, motion pattern classification result includes non-athletic mould
Formula and at least one motor pattern.Such as swimming exercise, motor pattern may include breaststroke, butterfly stroke, backstroke and freestyle swimming etc..
In an optional implementation manner, classification of motions model is trained by deep learning algorithm in advance.For example, SVM
(Support vector machine, support vector machines) and ANN (Artificial neural network, artificial neural network
Network).
SVM is a kind of a kind of supervised learning model in classification and analysis data in regression analysis.SVM is non-by one
Linear Mapping is mapped to sample space in one higher-dimension or even infinite dimensional feature space, so that in original sample space
The problem of middle Nonlinear separability, is converted into the problem of linear separability in feature space.Meanwhile SVM is based on structural risk minimization
Theory component optimal hyperlane in feature space, so that learning model obtains global optimization, and in entire sample space
The probability that is desired for meet certain upper limit.SVM can still have preferable general in the case where amount of training data is less as a result,
Change ability.
ANN is the mathematical model or computation model of a kind of structure and function of mimic biology neural network, such as convolution mind
Through network, Recognition with Recurrent Neural Network etc..ANN has the following characteristics that (1) Nonlinear Mapping ability, and ANN can be approached with arbitrary accuracy to be appointed
What non-linear continuous function.(2) Serial Distribution Processing mode, information is to be distributed storage and parallel processing in ANN, therefore
ANN has stronger fault-tolerance and faster data processing speed.(3) self study and adaptive ability, ANN are being trained
When, the knowledge of regularity can be extracted from input, the data exported, remembered in the weight of network, therefore ANN has preferably
Generalization ability.(4) data fusion ability, ANN can handle quantitative information and qualitative information simultaneously, therefore ANN has preferably
Data fusion ability.(5) ANN is multi-variable system, and the number for outputting and inputting variable can be arbitrary.
Step S400 determines motion state parameters according to motion pattern classification result and motion-sensing data.Below with trip
It describes to determine that motion state parameters determine motion state according to motion pattern classification result and motion-sensing data for swimming.Its
In, the motor pattern of swimming exercise may include breaststroke, butterfly stroke, backstroke and freestyle swimming etc..
Motion state parameters may include the number, swimming rate of striking of swimming, the number etc. turned round of swimming.In the present embodiment
In, according to motion pattern classification result and motion-sensing data determine a period of motion swimming strike number be specially respond
It is in swimming state in user, namely the motion pattern classification result of motion pattern classification result output trained in advance is one
Motor pattern obtains the peak value number of motion-sensing data, is struck according to the swimming that peak value number obtains the current kinetic period secondary
Number.Wherein, one time in swimming exercise of the period of motion, namely swim over to from swimming pool the process of swimming pool another side on one side and be known as one
The period of motion.
In an optional implementation manner, it is greater than first threshold, and the peak value in response to the peak value of motion-sensing data
It is greater than second threshold with the time interval of previous peak value, count is incremented for peak value number.Wherein, first threshold is according to actually striking
The motion-sensing parameter of movement is configured.Second threshold is configured according to the practical swimming rate of user.Specifically, to used
Property the motion-sensing data that obtain of measuring device pre-processed (such as being smoothed by filter), it is pre- to obtain
Acceleration on treated three axis.The 3-axis acceleration vector at each moment is obtained according to the acceleration on pretreated three axis
(namely the vector being made of acceleration value of the movement on three axis of synchronization), and obtain the 3-axis acceleration at each moment to
The mould of amount, to obtain the 3-axis acceleration vector field homoemorphism in current kinetic period according to the 3-axis acceleration vector field homoemorphism at each moment
Curve.Wherein, the peak value in the curve of the 3-axis acceleration vector field homoemorphism in current kinetic period is motion-sensing data
Peak value.It should be understood that the method for the peak value of above-mentioned acquisition motion-sensing data is only exemplary, other motion-sensing data
The acquisition methods (such as passing through the peak value in 3-axis acceleration and acquisition curve) of peak value can be applied to the present embodiment
In.
Fig. 2 is the peak value schematic diagram of the motion-sensing data of the embodiment of the present invention.As shown in Fig. 2, curve movement 2 be into
The curve of the 3-axis acceleration vector field homoemorphism of the certain time period obtained when row swimming exercise.Movement caused by athletic performance 21
It is not arm stroke that the peak value of sensing data, which is less than first threshold th1 namely athletic performance 21, and swimming number of striking is not counted upwards
Number.The peak value of motion-sensing data caused by athletic performance 22 is greater than first threshold th1, also, athletic performance 22 is corresponding
Time interval t1 between peak value peak value corresponding with athletic performance 21 is greater than second threshold th2, therefore athletic performance 22 is to draw
Hydrodynamic(al) is made, and swimming number of striking adds 1.The peak value of motion-sensing data caused by athletic performance 23 is greater than first threshold th1, and
And the time interval t2 between the corresponding peak value of athletic performance 23 peak value corresponding with athletic performance 22 is greater than second threshold th2,
Therefore athletic performance 23 is arm stroke, and swimming number of striking adds 1.The peak value of motion-sensing data caused by athletic performance 24
Greater than first threshold th1, still, the time interval between the corresponding peak value of athletic performance 24 peak value corresponding with athletic performance 23
T3 is less than second threshold th2, therefore athletic performance 24 is not arm stroke, and swimming number of striking does not count up.Preferably, it rings
First threshold should be greater than in the peak value of motion-sensing data, and the time interval of the peak value and the previous peak value being counted is greater than
Second threshold, count is incremented for peak value number.For example, the peak value of motion-sensing data caused by athletic performance 25 is greater than the first threshold
Value th1, also, the time interval t4 between the corresponding peak value of athletic performance 25 peak value corresponding with athletic performance 23 is greater than second
Threshold value th2, therefore athletic performance 25 is arm stroke, swimming number of striking add 1.
In the present embodiment, determine that motion state parameters are specially according to motion pattern classification result and motion-sensing data
Obtain the number of touch turn.It is readily appreciated that, during exercise, generally when moving to athletic ground edge, there are touch turns.
In the present embodiment, motion-sensing data can also include steering angle, and in an optional implementation manner, steering angle can lead to
It crosses 3-axis acceleration and angular speed is estimated.The evaluation method of the steering angle of user during exercise in the present embodiment can join
According to paper: Sebastian O.H., Madgwick, Andrew J.L. etc.；Estimation of IMU and MARG
orientation using a gradient descent algorithm[C]//IEEE International
Conference on Rehabilitation Robotics.IEEE, 2011.In another optional implementation, turn to
Angle can also be calculated by 3-axis acceleration and magnetic induction intensity.Magnetic induction intensity of the user in motion state can use magnetic
The acquisition of power meter obtains.As a result, when steering angle when obtaining user movement, magnetometer can replace gyroscope.It should be understood that other
The method for obtaining steering angle can be applied in the present embodiment.
It in an optional implementation manner, is non-moving pattern and non-moving pattern in response to motion pattern classification result
Duration is less than predetermined continuous time, is greater than steering angle threshold value in current steering angle and the swimming in current kinetic period is struck
When number is within the scope of pre-determined number, the number of touch turn adds 1.Wherein, the pre-determined number range of a period of motion according to
Period of motion dynamic updates, and predetermined continuous time is configured according to average time needed for turning round movement etc., steering angle threshold value
Steering angle when being turned round according to this movement is configured, and the swimming of pre-determined number range namely a period of motion are struck number
Range.
Wherein, motion pattern classification result is non-moving pattern, namely the time range in current acceleration data window
Interior, user does not carry out any motor pattern, still, if user exports non-moving pattern, user in Long time scale
Movement may be stopped.Therefore, judge user whether generate turn round movement when, need to judge the non-moving pattern duration
Length.
May occur the case where steering angle obtained is greater than steering angle threshold value once in a while during the motion, it is therefore desirable to adopt
Swimming with user in a period of motion strike number participate in determine user touch turn.That is, in user's current kinetic
When the swimming in period strikes number within the scope of pre-determined number, user is likely to reach swimming pool edge so that touch turn occurs.Together
When, since different classes of (such as child, adult) of user is different, same user is different in the physical strength of different time, therefore pre-
Determining numbers range is dynamic change, to further increase the accuracy of motion state monitoring.
Step S500 exports the motor pattern and motion state parameters of current period.That is, output current kinetic week
The swimming in the motor pattern of phase and each period is struck the number etc. of number, movement velocity and touch turn.A kind of optional
Implementation in, by record current period multiple acceleration information windows motion pattern classification as a result, and to movement
The corresponding motor pattern of pattern classification result counts respectively, and then output counts highest motor pattern.That is, each fortune
In the dynamic period, motor pattern of the highest motor pattern of motor pattern frequency that user uses as current period, to avoid
Error caused by erroneous judgement further improves the accuracy of moving state identification.Wherein, movement velocity can be drawn according to swimming
The run duration in waterside number and current kinetic period calculates, when can also according to the movement in move distance and current kinetic period
Between calculate.
In an optional implementation manner, the motion state monitoring method of the present embodiment further include: in response to moving mould
Formula classification results are non-moving pattern within predetermined continuous time, determine that movement terminates.
The characteristic information that the technical solution of the embodiment of the present invention passes through the acceleration information window that will acquire is input in advance
To obtain motion pattern classification in trained motion pattern classification model, and according to the motion pattern classification and motion-sensing number
According to motion state parameters are determined, thus, it is possible to improve the generalization ability and robustness of motion state monitoring method, and movement is improved
The accuracy rate of status monitoring.
Fig. 3 is the data flow figure of the method for monitoring operation states of the embodiment of the present invention.As shown in figure 3, pretreatment and spy
The training data data_tra of 31 pairs of extraction unit inputs of sign carries out pretreatment and feature information extraction, and characteristic information is exported
It is trained into motion pattern classification training unit 32.Wherein, training data data_tra may include the various marks of acquisition
The data of quasi-moving mode.Pretreatment and feature extraction unit 31 be configured as by pretreatment to training data do it is smooth, return
The processing such as one change, and extract the characteristic information of pretreated training data.Motion pattern classification training unit 32 using SVM or
ANN is trained characteristic information to obtain motion pattern classification model 33.
Trained motion pattern classification model 33 can be embedded in wearable device (such as sports watch, motion bracelet
Deng) in.When the motion state to user is monitored, by motion-sensing data acquisition unit 34 to the motion-sensing of user
Data data_exp is acquired.In an optional implementation manner, pass through IMU built-in in wearable device
(Inertial measurement unit, inertial measuring unit) obtains the predetermined acceleration data window of user to acquire
Motion-sensing data data_exp.The motion-sensing data data_exp of user includes 3-axis acceleration, angular speed etc..Pretreatment
Unit 35 carries out the processing such as smooth, normalization to motion-sensing data data_exp, and pretreated motion-sensing data are defeated
Out to feature extraction unit 36.Feature extraction unit 36 extracts the characteristic information of pretreated motion-sensing data.Feature letter
Breath includes the related coefficient etc. on three axis between the minimum value, maximum value of acceleration, average value, standard deviation and two axis.Move mould
Formula disaggregated model 33 obtains the motion pattern classification result of current acceleration data window according to characteristic information.Wherein, mould is moved
Formula classification results include non-moving pattern and at least one motor pattern.Such as swimming exercise, motor pattern may include breaststroke,
Butterfly stroke, backstroke and freestyle swimming etc..
Motion state monitoring unit 37 is configured as the motion pattern classification knot exported according to motion pattern classification model 33
The pretreated motion-sensing data that fruit and pretreatment unit 35 export determine motion state parameters, and export each period
Motor pattern and motion state parameters.Wherein, motion state parameters may include swimming strike number, movement velocity, turn round it is dynamic
The number etc. of work.In an optional implementation manner, pass through the movement of multiple acceleration information windows of record current period
Pattern classification as a result, and the corresponding motor pattern of motion pattern classification result is counted respectively, then output counts highest fortune
The motor pattern in dynamic model formula namely current kinetic period.
Wherein, motion state monitoring unit 37 includes touch turn monitoring subelement 371 and number monitoring subelement of striking
372。
Touch turn monitoring subelement 371 is configured as obtaining time of user's touch turn within the scope of the secondary run duration
Number.It in an optional implementation manner, is that non-moving pattern and non-moving pattern continue in response to motion pattern classification result
Time is less than predetermined continuous time, is greater than steering angle threshold value and the swimming in current kinetic period in current steering angle and strikes number
When within the scope of pre-determined number, the number of touch turn adds 1.Wherein, the pre-determined number range of a period of motion is according to movement
Period dynamic updates, and predetermined continuous time is configured according to movement required average time etc. is turned round, steering angle threshold value according to
This steering angle of movement when turning round is configured, and the swimming of pre-determined number range namely a period of motion are struck number.?
In swimming exercise, generally when swimming over to swimming pool edge, there are touch turns, and therefore, often turning round primary is to complete a swimming.
Thus, it is possible to take over number of the family in swimming exercise for use by the frequency table of touch turn.
In the present embodiment, motion-sensing data can also include steering angle.Wherein it is possible to pass through 3-axis acceleration and angle
Speed calculates steering angle, can also calculate steering angle by 3-axis acceleration and magnetic induction intensity.It is readily appreciated that, moves mould
Formula classification results are non-moving pattern, namely in the time range of current acceleration data window, user does not carry out any
Motor pattern, still, if user exports non-moving pattern in Long time scale, user may stopped movement.Therefore,
Judge user whether generate turn round movement when, need to judge the length of non-moving pattern duration.During the motion may be used
The case where steering angle obtained is greater than steering angle threshold value can occur once in a while, it is therefore desirable to using user in period of motion
Swimming strike number participate in determine user touch turn.That is, the swimming in user's current kinetic period strikes number pre-
When determining in numbers range, user is likely to reach athletic ground edge so that touch turn occurs.Simultaneously as the inhomogeneity of user
(such as child, adult) be not different, and same user is different in the physical strength of different time, therefore pre-determined number range is dynamic change
, to further increase the accuracy of motion state monitoring.
Number of striking monitoring subelement 372, which is configured as obtaining the swimming in current kinetic period, strikes number.One kind can
In the implementation of choosing, number of striking monitoring subelement 372 is kept in motion in response to user, namely movement trained in advance
The motion pattern classification result of pattern classification result output is a motor pattern, obtains the peak value number of motion-sensing data,
It is struck number according to the swimming that peak value number obtains current period.Specifically, it is greater than the in response to the peak value of motion-sensing data
One threshold value, and the time interval of the peak value and previous peak value is greater than second threshold, count is incremented for peak value number.Wherein, the first threshold
The motion-sensing parameter that value is acted according to actual motion is configured.Second threshold is set according to the actual motion speed of user
It sets.Specifically, the 3-axis acceleration vector at each moment is obtained (namely by same a period of time according to the acceleration on pretreated three axis
The vector of acceleration value composition of the movement at quarter on three axis), and the 3-axis acceleration vector field homoemorphism at each moment is obtained, thus root
The curve of the 3-axis acceleration vector field homoemorphism in current kinetic period is obtained according to the 3-axis acceleration vector field homoemorphism at each moment.Wherein,
Peak value in the curve of the 3-axis acceleration vector field homoemorphism in current kinetic period is the peak value of motion-sensing data.It should be understood that
The method of the peak value of above-mentioned acquisition motion-sensing data is only exemplary, the acquisition side of the peak value of other motion-sensing data
Method (such as passing through the peak value in 3-axis acceleration and acquisition curve) can be applied in the present embodiment.
The characteristic information that the technical solution of the embodiment of the present invention passes through the acceleration information window that will acquire is input in advance
To obtain motion pattern classification in trained motion pattern classification model, and according to the motion pattern classification and motion-sensing number
According to motion state parameters are determined, thus, it is possible to improve the generalization ability and robustness of motion state monitoring method, and movement is improved
The accuracy rate of status monitoring.
Fig. 4 is the schematic diagram of the wearable device of the embodiment of the present invention.As shown in figure 4, the wearable device of the present embodiment
Including inertial measuring unit 41, memory 42, communication component 43 and at least one processor 44.Inertial measuring unit 41 is deposited
Reservoir 42, communication component 43 and at least one processor 44 are connected by bus or other modes, in Fig. 4 by taking bus as an example.
Wherein, inertial measuring unit 41 is configured as the motion-sensing data of acquisition user.Wherein, inertial measuring unit includes that three axis add
Velocity sensor, gyroscope and/or magnetometer, motion-sensing data include 3-axis acceleration, angular speed and/or magnetic induction intensity
Deng.Communication component 43 sends and receivees data under the control of processor 44.Memory 42 is stored with can be by least one processor
44 instructions executed, instruction are executed the motion state monitoring method to realize the embodiment of the present invention by least one processor 44.
Memory 42 is used as a kind of non-volatile computer readable storage medium storing program for executing, can be used for storing non-volatile software journey
Sequence, non-volatile computer executable program and module.Processor 44 is stored in non-volatile in memory 42 by operation
Property software program, instruction and module, thereby executing the various function application and data processing of equipment, i.e. realization motion state
The method of monitoring.
Memory 42 may include storing program area and storage data area, wherein storing program area can storage program area,
Application program required at least one function；It storage data area can the Save option list etc..In addition, memory 42 may include
High-speed random access memory can also include nonvolatile memory, for example, at least disk memory, a flash memories
Part or other non-volatile solid state memory parts.In some embodiments, it includes relative to processor 42 that memory 42 is optional
Remotely located memory, these remote memories can pass through network connection to external equipment.One or more unit is deposited
Storage, when being executed by one or more processor 44, executes the fortune in above-mentioned any means embodiment in memory 42
Dynamic state monitoring method.
Method provided by the embodiment of the present invention can be performed in the said goods, has the corresponding functional unit of execution method and has
Beneficial effect, the not technical detail of detailed description in the present embodiment, reference can be made to method provided by embodiment of the present invention.
The above description is only a preferred embodiment of the present invention, is not intended to restrict the invention, for those skilled in the art
For, the invention can have various changes and changes.All any modifications made within the spirit and principles of the present invention are equal
Replacement, improvement etc., should all be included in the protection scope of the present invention.
Claims (10)
1. a kind of motion state monitoring method based on wearable device, which is characterized in that the described method includes:
Acceleration information window is obtained, the acceleration information window has scheduled time span；
Obtain the characteristic information of the acceleration information window；
The characteristic information is input in motion pattern classification model trained in advance to obtain motion pattern classification as a result, institute
Stating motion pattern classification result includes non-moving pattern and at least one motor pattern；
Motion state parameters are determined according to the motion pattern classification result and motion-sensing data；And
Export the motor pattern and the motion state parameters in current kinetic period.
2. motion state monitoring method according to claim 1, which is characterized in that described to obtain the acceleration information window
Mouthful characteristic information include:
The motion-sensing data of the acceleration information window are obtained by inertial measuring unit；
According to the characteristic information of acceleration information window described in the motion-sensing data acquisition, the characteristic information includes three axis
Related coefficient between the minimum value of upper acceleration, maximum value, average value, standard deviation and two axis.
3. motion state monitoring method according to claim 1, which is characterized in that the motion state parameters include current
The swimming of the period of motion is struck number；
It is described to determine that motion state parameters include: according to the motion pattern classification result and motion-sensing data
The peak value number for obtaining the motion-sensing data is struck secondary according to the swimming that the peak value number obtains current period
Number.
4. motion state monitoring method according to claim 3, which is characterized in that described to obtain the motion-sensing data
Peak value number include:
It is greater than first threshold in response to the peak value, and the time interval of the peak value and previous peak value is greater than second threshold,
Count is incremented for peak value number.
5. motion state monitoring method according to claim 1, which is characterized in that the motion state parameters include turning round
The number of movement；
It is described to determine that motion state parameters include: according to the motion pattern classification result and motion-sensing data
Obtain the number of touch turn.
6. motion state monitoring method according to claim 5, which is characterized in that the number packet for obtaining touch turn
It includes:
In response to the motion pattern classification result be non-moving pattern and the non-moving pattern duration is less than predetermined consecutive hours
Between, it is greater than steering angle threshold value in the current steering angle and the swimming in current kinetic period strikes number in pre-determined number range
When interior, the number of touch turn adds 1；
Wherein, the steering angle is calculated by the motion-sensing data and is obtained, and the pre-determined number range is according to the period of motion
Dynamic updates.
7. motion state monitoring method according to claim 1, which is characterized in that the output motion pattern classification
Result includes:
The motion pattern classification of multiple acceleration information windows of current period is recorded as a result, and to the motor pattern point
The corresponding motor pattern of class result counts respectively；
Output counts highest motor pattern.
8. motion state monitoring method according to claim 1, which is characterized in that the method also includes:
It is non-moving pattern within predetermined continuous time in response to the motion pattern classification result, determines that movement terminates.
9. motion state monitoring method according to claim 1 to 8, which is characterized in that the type of sports is
Swimming exercise, the motor pattern include breaststroke, butterfly stroke, backstroke and freestyle swimming.
10. a kind of wearable device, which is characterized in that be provided in the wearable device inertial measuring unit, processor and
Memory, the memory is for storing one or more computer program instructions, wherein the one or more computer journey
Sequence instruction is executed by the processor to realize method as claimed in any one of claims 1-9 wherein.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811565345.6A CN109663323A (en) | 2018-12-20 | 2018-12-20 | A kind of motion state monitoring method and wearable device based on wearable device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811565345.6A CN109663323A (en) | 2018-12-20 | 2018-12-20 | A kind of motion state monitoring method and wearable device based on wearable device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109663323A true CN109663323A (en) | 2019-04-23 |
Family
ID=66144929
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811565345.6A Pending CN109663323A (en) | 2018-12-20 | 2018-12-20 | A kind of motion state monitoring method and wearable device based on wearable device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109663323A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110108278A (en) * | 2019-05-22 | 2019-08-09 | 北京卡路里信息技术有限公司 | It is landed determining method and device based on the foot of six axle sensors |
| CN110584675A (en) * | 2019-09-30 | 2019-12-20 | 北京卡路里信息技术有限公司 | Information triggering method and device and wearable device |
| CN114533047A (en) * | 2022-02-23 | 2022-05-27 | 首都体育学院 | Motion pattern recognition algorithm based on wearable equipment |
| WO2023178594A1 (en) * | 2022-03-24 | 2023-09-28 | 广东高驰运动科技股份有限公司 | Action counting method and apparatus, device, and storage medium |
| WO2024000211A1 (en) * | 2022-06-29 | 2024-01-04 | 广东高驰运动科技股份有限公司 | Outdoor swimming trajectory determination method and system, and device and storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080018532A1 (en) * | 2002-11-01 | 2008-01-24 | Sportzco Pty Ltd | Monitoring sports and swimming |
| CN105080111A (en) * | 2014-05-14 | 2015-11-25 | 阿迪达斯股份公司 | Sport ball motion monitoring methods and systems |
| CN106237604A (en) * | 2016-08-31 | 2016-12-21 | 歌尔股份有限公司 | Wearable device and the method utilizing its monitoring kinestate |
| CN106914011A (en) * | 2017-05-03 | 2017-07-04 | 盐城工学院 | Movement locus tape deck and locomotion evaluation system |
| CN107469326A (en) * | 2017-07-04 | 2017-12-15 | 广东乐心医疗电子股份有限公司 | Swimming monitoring method and device for wearable equipment and wearable equipment |
-
2018
- 2018-12-20 CN CN201811565345.6A patent/CN109663323A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080018532A1 (en) * | 2002-11-01 | 2008-01-24 | Sportzco Pty Ltd | Monitoring sports and swimming |
| CN105080111A (en) * | 2014-05-14 | 2015-11-25 | 阿迪达斯股份公司 | Sport ball motion monitoring methods and systems |
| CN106237604A (en) * | 2016-08-31 | 2016-12-21 | 歌尔股份有限公司 | Wearable device and the method utilizing its monitoring kinestate |
| CN106914011A (en) * | 2017-05-03 | 2017-07-04 | 盐城工学院 | Movement locus tape deck and locomotion evaluation system |
| CN107469326A (en) * | 2017-07-04 | 2017-12-15 | 广东乐心医疗电子股份有限公司 | Swimming monitoring method and device for wearable equipment and wearable equipment |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110108278A (en) * | 2019-05-22 | 2019-08-09 | 北京卡路里信息技术有限公司 | It is landed determining method and device based on the foot of six axle sensors |
| CN110108278B (en) * | 2019-05-22 | 2021-07-09 | 北京卡路里信息技术有限公司 | Foot landing determination method and device based on six-axis sensor |
| CN110584675A (en) * | 2019-09-30 | 2019-12-20 | 北京卡路里信息技术有限公司 | Information triggering method and device and wearable device |
| CN110584675B (en) * | 2019-09-30 | 2022-05-03 | 北京卡路里信息技术有限公司 | Information triggering method and device and wearable device |
| CN114533047A (en) * | 2022-02-23 | 2022-05-27 | 首都体育学院 | Motion pattern recognition algorithm based on wearable equipment |
| WO2023178594A1 (en) * | 2022-03-24 | 2023-09-28 | 广东高驰运动科技股份有限公司 | Action counting method and apparatus, device, and storage medium |
| WO2024000211A1 (en) * | 2022-06-29 | 2024-01-04 | 广东高驰运动科技股份有限公司 | Outdoor swimming trajectory determination method and system, and device and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109663323A (en) | A kind of motion state monitoring method and wearable device based on wearable device | |
| Hernández et al. | Human activity recognition on smartphones using a bidirectional LSTM network | |
| US20180249908A1 (en) | Multi-state performance monitoring system | |
| Nakai et al. | Prediction of basketball free throw shooting by openpose | |
| CN109740742A (en) | A kind of method for tracking target based on LSTM neural network | |
| US10486049B2 (en) | Device and method for monitoring swimming performance | |
| CN108681774A (en) | Based on the human body target tracking method for generating confrontation network negative sample enhancing | |
| CN102640085A (en) | System and method for recognizing gestures | |
| CN110263870A (en) | Moving state identification method, apparatus, intelligent terminal and storage medium | |
| GB2529082A (en) | Device and method for monitoring swimming performance | |
| Rao et al. | Ctin: Robust contextual transformer network for inertial navigation | |
| CN108827290A (en) | A kind of human motion state inverting device and method | |
| CN113743273A (en) | Real-time rope skipping counting method, device and equipment based on video image target detection | |
| Li et al. | Video-based table tennis tracking and trajectory prediction using convolutional neural networks | |
| Bezobrazov et al. | Artificial intelligence for sport activitity recognition | |
| CN110132276A (en) | A kind of adaptive step estimation method based on pedestrian movement's state | |
| KR102363879B1 (en) | Method for predicting clinical functional assessment scale using feature values derived by upper limb movement of patients | |
| CN106073793A (en) | Attitude Tracking based on micro-inertia sensor and recognition methods | |
| Chen | An LSTM recurrent network for step counting | |
| Brzostowski et al. | Data fusion in ubiquitous sports training: methodology and application | |
| Viyanon et al. | Swingpong: analysis and suggestion based on motion data from mobile sensors for table tennis strokes using decision tree | |
| AU2019379337B2 (en) | System and method for sweet spot detection | |
| Naik et al. | LSTM-BEND: predicting the trajectories of basketball | |
| CN108051001A (en) | A kind of robot movement control method, system and inertia sensing control device | |
| CN107329563A (en) | A kind of recognition methods of type of action, device and equipment |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190423 |