CN107180423B - A kind of ball service training mate method based on motion profile - Google Patents
A kind of ball service training mate method based on motion profile Download PDFInfo
- Publication number
- CN107180423B CN107180423B CN201710225245.8A CN201710225245A CN107180423B CN 107180423 B CN107180423 B CN 107180423B CN 201710225245 A CN201710225245 A CN 201710225245A CN 107180423 B CN107180423 B CN 107180423B
- Authority
- CN
- China
- Prior art keywords
- service
- motion profile
- point
- ball
- training
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/223—Analysis of motion using block-matching
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10048—Infrared image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30221—Sports video; Sports image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30241—Trajectory
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Quality & Reliability (AREA)
- Multimedia (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The ball service training mate method based on motion profile that the invention discloses a kind of, comprising the following steps: step 1: the motion profile of racket when obtaining multiple ball services；Step 2: extracting feature vector by multiple motion profiles that step 1 acquires；Step 3: the feature vector obtained for step 2, training obtain population-supporting vector machine model；Step 4: the racket motion profile for treating evaluation is searched, and service track is extracted in motion profile according to service track characteristic, trained population-supporting vector machine model in input step 3 carries out the evaluation of service success or not；The method of the present invention realizes technology action analysis when more objective player service, helps coach and sportsman to find nonstandard movement or malfunction, improves athletic training efficiency, Motion Technology is improved, to achieve the purpose that supplemental training.
Description
Technical field
The present invention relates to three-dimensional motion process field, in particular to a kind of ball service training mate side based on motion profile
Method.
Background technique
In tennis, service is one of basic technology of most critical.Its technical movements can be complete by sportsman itself
Implement, is a kind of important Goal-method in tennis without being fought with opponent.In tennis service, racket is most
Best placement, motion profile, height and speed are very important.
Currently, Olivier Girard et al. has studied knee fortune by the stretching angle of knee during limitation service
The dynamic influence to tennis flat service.Ikram Hussain et al. is in order to improve the serve speed of tennis, to body various pieces
Deep kinematics analysis is carried out.In numerous extraneous factors, Pedro C.Mendes etc. thinks wind to service success rate shadow
Sound is bigger, and has carried out detailed demonstration to different wind speed, wind direction.
The above-mentioned research to tennis service process is either from human body attitude or from external environment, though energy is right
Directive function is played in the raising of tennis service technology, but research model complexity is not easy to realize.In view of the above-mentioned problems, in order to improve training
Effect also has researcher to propose some service training devices, such as the patent document of Publication No. CN203469419U discloses
Shuttlecock is served a ball training system, is related to a kind of shuttlecock training system.It is to adapt to badminton service robot and move property, accuracy
Intelligent training demand.Array indicator light is distributed in the side in shuttlecock court, and the control signal of array indicator light is defeated
The control signal output for entering end and control circuit connects；The infrared signal output end and control circuit of infrared signal receiving circuit
Infrared signal input terminal connection；Infrared signal receiving circuit is used to receive the infrared signal of infrared signal transmission circuit transmitting；
The push button signalling output end of array key and the push button signalling input terminal of infrared signal transmission circuit connect；Video camera is for adopting
Collect the image of service placement；The image signal output end of video camera and the picture signal input terminal of control circuit connect；Control electricity
The control signal output on road and the control signal input of video camera connect.The invention is suitable for shuttlecock service training process
In.
But above-mentioned serving trainer structure exists the problems such as structure is complicated, inconvenient for use and test inaccuracy, therefore,
For this problem, it proposes a kind of ball service training mate method based on motion profile, realizes objective, the effective finger of ball training
It leads very necessary.
Summary of the invention
The ball service training mate method based on motion profile that the present invention provides a kind of, by ball certain point or ball
Judgement of the research realization of vertex movements track to flat service success or not is clapped, and then provides technological guidance, can be improved ball
Service training effectiveness assists ball service training.
A kind of ball service training mate method based on motion profile, comprising the following steps:
Step 1: the motion profile of racket when obtaining multiple ball services；
Step 2: extracting feature vector by multiple motion profiles that step 1 acquires；
Step 3: the feature vector obtained for step 2, training obtain population-supporting vector machine model (abbreviation PSO-
SVM model)；
Step 4: the racket motion profile for treating evaluation is searched, and is extracted in motion profile according to service track characteristic
Set out sphere path curve, trained population-supporting vector machine model in input step 3, carries out the evaluation of service success or not.
For the ease of realizing, simplify operating procedure, it is preferred that in step 1, the tool of the motion profile of racket when obtaining service
Steps are as follows for body:
Mark point is placed on racket by 1-1；In order to improve accuracy, it is preferred that mark point is placed in the bulb top of racket
Far from bat larynx side.Mark point preferably refers to signal point herein, and signal point is as processing target easily by infrared light phase machine examination
It measures.Signal point is placed in racket top.
1-2 acquires mark point motion profile, the i.e. motion profile of racket, and only square perpendicular to the ground in retaining space coordinate system
To y-axis motion profile it is stand-by, altogether acquire n group, wherein service successful trail n1Group, service failure track n2Group, wherein n1+n2
=n；System is captured using OptiTrack three-dimensional infrared moving, cooperates 12 Prime13 (abbreviation p13) infrared light high speed cameras,
Acquire marker point (mark point) motion profile.
With OptiTrack acquire delivery of service, every group of about 800 frames, the X that makes discovery from observation, Z-direction data because not
The service skill that uses with people, motor habit is variant that trajectory shape is caused to vary with each individual, and regularity is weaker, but Y-axis (Vertical Square
To coordinate components) trail change rule is identical.Therefore Y-axis data are chosen as final analysis object.
1-3 intercepts the motion profile of service process in the motion profile that step 1-2 is obtained；
Here the service track intercepted needs accurately, for training PSO-SVM model so when combining shooting service video
Actual conditions, the artificial motion profile for intercepting service process.
N group that step 1-3 is extracted service motion profile is divided into two groups by 1-4: training group and test group.Wherein, it instructs
Practicing the service tracking quantity that group is got is nx=0.75n=0.75n1+0.75n2, test group service tracking quantity is nc=0.25n
=0.25n1+0.25n2.The selection of being adapted to property of data.
Preferably, in step 2, the step of motion profile acquired by step 1 extracts feature vector, is as follows:
The extreme point of the 2-1 record service track y, and following value: first minimum point to first is recorded according to extreme point
The time t of a maximum point experience0, the vertical range d that passes by0；Second minimum point to second maximum point undergo when
Between t2, the vertical range d that passes by2；The time t that second maximum point is undergone to third minimum point3, pass by it is vertical away from
From d3；The time t that first minimum point is undergone to third minimum point；The y of maximum of points in one service period is sat
Mark d；
There is the racket vertex y-axis motion profile shape of the reason of carrying out step 2-1 operation predominantly label rule can follow:
It is all the preparation stage of service before T0, before service starts, whole body holding is loosened, and attention is concentrated, and leans to one side to stand
In teeing ground, left scapuloanterior (LScA) inclines, and against left side net post direction, towards right front, bipod is separated standing, at splayed, makes
Center of gravity is located in front foot.
T0-T1 is to draw the bat stage: first maximum point that (T0) is clapped to batting moment forecourt since delivery of service
(T1), centre of body weight is suitably downward at this time, and the automatic bottom right of hand of holding bat is swung toward back upper place, and racket generally guides to shoulder, and elbow hangs down
Directly, face is parallel to mesh belt, perpendicular to the ground.Draw bat stage elbow joint natural torsion, sagging, center of gravity is moved on the foot at rear,
Work is ready to have an effect.
Before wave boost phase (T1-T2): first maximum point (T1) of service track is to next minimum before batting
It is worth point (T2).After ball is spilled over, racket, which continues up, to be put, and trunk is swung forward, makes body before batting in " bow " type.
Batting stage (T2-T3): when ball drops to hitting point, batting of swinging the bat rapidly upward keeps arm and body abundant
Stretching, extension.It swings the bat square stance, holds wrist strap of clapping hands and move the hip beats that forearm has a medial rotation, this is the key operations having an effect,
The This move strength such as the center of gravity of early period can be moved forward, kick one's legs, turn, swing the bat all are discharged on ball.
With the stage of waving (T3-T4): occurring after batting (T3), hold bat and continue to stretch to direction of stroke, utilize inertia arm
It slightly guides, racket is finally placed in left side of body.Smoothness unfold with waving the damage that can be reduced Ji Qiu to joint and muscle.
From the foregoing, it can be seen that some extreme point in the starting of each stage and end of a period homologous thread of process of serving a ball, because
The amount of this said extracted characterizes the distance of the vertical direction of different service stage durations and operation.
2-2 calculates following feature: x according to the data of step 2-1 record0=t0/t；x1=t1/t；x2=t2/t；x3=t3/
t；x4=d0/(d×x0)；x5=d1/(d×x1)；x6=d2/(d×x2)；x7=d3/(d×x3)；x8=d2/t2；x9=d/t；
2-3 further extracts feature according to the step 2-2 characteristic provided: calculating x in step 2-20To x3This group of number
According to variance δ1, x4To x7The variance δ of this group of data2, enable x10=δ1, x11=δ2。
2-4 constitutes 11 dimensional feature vectors: X={ x according to the feature that step 2-2,2-3 is extracted0,x1,···x11}；
The meaning for each feature that above-mentioned steps 2-2 to step 2-4 is extracted is as follows: x0To x3And feature x10Each rank of characterization service
The section time distributes whether reasonable, x4To x7And x11Then indicate whether movement velocity distribution is reasonable；x8For embody batting speed,
Explosive force；x9For embodying toss height and the matching relationship of hitting time of swinging the bat, be implied by height, ball hands-off speed, ball from
The information such as hand height.
2-5 extracts training group n respectivelyxThe feature vector, X of a sample indicates are as follows:Wherein Xi∈R11, be i-th group of sample extraction the feature containing 11 features to
Amount, Yi∈ { -1,1 }, i=1,2 ..., nx, as vector XiBe successfully serve a ball trajectory extraction feature when, Yi takes 1, otherwise takes -1.
Preferably, in step 3, for step 2 obtain feature vector training population-supporting vector machine model the step of
Are as follows:
3-1 selects Radial basis kernel function as population-supporting vector machine model core, Radial basis kernel function such as following formula institute
Show:
X in above formulajFor sample data to be processed, xcFor kernel function center, σ is the width parameter of function, controls function
Radial effect range；
The reason of choosing Radial basis kernel function are as follows: Radial basis kernel function parameter in need of consideration is few, the complexity of model selection
It spends just smaller.
The step 2-5 feature vector extracted is divided into 4 groups by 3-2, carries out cross validation to population-supporting vector machine model
Parameter σ and punishment parameter C, wherein 0.1≤C≤100,0.01≤σ≤1000；
3-3 passes through PSO algorithm optimization punishment parameter C and kernel functional parameter σ value.
Preferably, 4-1 obtains the track data of the signal point y placed at racket vertex to be evaluated, when encountering track
On minimum point when, (entire service process is generally interior at 2-3 second just to be completed, therefore is comprehensively considered 4 seconds after sequential hunting
Afterwards, 4 second durations of the selection greater than 3 seconds find service track, avoid the truncation of complete service track, or the hair of front and back two occur
The overlapping of sphere path curve) in curve extreme point；
4-2 is maximum point, minimum point, again maximum point, minimum when the sequence that the extreme point of subsequent search occurs
Point, and meet the latter maximum point value greater than previous maximum point, first minimum point that subsequent searches are encountered takes
When value is greater than first minimum point that entire search process is encountered, it is believed that find a service track, and the track that will serve a ball is defeated
Enter the judgement that population-supporting vector machine model in step 3 carries out service success or not；
If 4-3 service failure, by the standard service track in the feature vector and model of this service trajectory extraction
The feature vector of extraction is compared, to judge there are gaps with standard operation in which stage delivery of service, to carry out
Special training is corrected.
The present invention is on the basis of analyzing the parameter selection method of traditional support vector machine, by particle swarm algorithm to branch
It holds vector machine and carries out parameter preferentially, propose the ball service model of PSO_SVM, realize the movement according to ball racket apex
The method that track determines service success or not, to realize the guidance to ball flat service process.
Beneficial effects of the present invention:
The method of the present invention realizes technology action analysis when more objective player service, helps coach and fortune
It mobilizes and finds nonstandard movement or malfunction, improve athletic training efficiency, improve Motion Technology, to reach auxiliary instruction
Experienced purpose.
Detailed description of the invention
Fig. 1 is signal point structure schematic diagram in the method for the present invention.
Fig. 2 is the structural schematic diagram that placed the racket of signal point in the method for the present invention.
Fig. 3 is world coordinate system and camera placement position schematic diagram.
Fig. 4 is the movement locus schematic diagram of racket mark point y-axis.
Fig. 5 is the wire frame flow chart of the ball service training mate method of the invention based on motion profile.
Specific embodiment
As shown in Fig. 1~5, the ball service training mate method based on motion profile of the present embodiment, by taking tennis service as an example,
As shown in Figure 5, comprising the following steps:
Step 1: the motion profile on racket vertex when system obtains tennis service is captured by three-dimensional infrared moving；
Step 2: feature vector is extracted by the motion profile that step 1 acquires；
Step 3: feature vector training population-supporting vector machine model (the abbreviation PSO-SVM mould obtained for step 2
Type)；
Step 4: the racket vertex movements track for treating evaluation is searched, according to service track characteristic in motion profile
Service track is extracted, trained PSO-SVM model is inputted, the evaluation of service success or not is carried out, tennis service is provided
Technical instruction.
Specifically, capturing the movement rail on racket vertex when system obtains tennis service by three-dimensional infrared moving in step 1
Specific step is as follows for mark:
1-1 is as illustrated in fig. 1 and 2, and marker point 1 is placed in the bulb top of tennis racket far from bat that side of larynx；
1-2 cooperates 12 Prime13 (referred to as shown in figure 3, using OptiTrack three-dimensional infrared moving capture system
P13, the triangle in figure) infrared light high speed camera, acquire marker point motion profile, i.e. the movement rail on tennis bulb top
Mark, and only the y-axis motion profile in direction is stand-by perpendicular to the ground in retaining space coordinate system, 60 groups is acquired altogether, wherein serving a ball successfully
46 groups of track, 14 groups of track of service failure.
1-3 artificially intercepts the motion profile of service process in motion profile.
With OptiTrack acquire delivery of service, every group of about 800 frames, the X that makes discovery from observation, Z-direction data because not
The service skill that uses with people, motor habit is variant that trajectory shape is caused to vary with each individual, and regularity is weaker, but Y-axis (Vertical Square
To coordinate components) trail change rule is identical.Therefore Y-axis data are chosen as final analysis object.It is with first group of data
Example (totally 868 frame), artificial to choose 0-632 frame Y-axis data, as shown in figure 4, being analyzed, tennis service process be can be divided into: quasi-
It is standby, draw bat, preceding wave acceleration, batting and with waving five stages.Before respectively corresponding the T0 in Fig. 4, T0-T1, T1-T2, T2-T3,
The T3-T4 stage.
It is all the preparation stage of service before T0, before service starts, whole body holding is loosened, and attention is concentrated, and leans to one side to stand
In teeing ground, left scapuloanterior (LScA) inclines, and against left side net post direction, towards right front, bipod is separated standing, at splayed, makes
Center of gravity is located in front foot.
T0-T1 is to draw the bat stage: first maximum point that (T0) is clapped to batting moment forecourt since delivery of service
(T1), centre of body weight is suitably downward at this time, and the automatic bottom right of hand of holding bat is swung toward back upper place, and racket generally guides to shoulder, and elbow hangs down
Directly, face is parallel to mesh belt, perpendicular to the ground.Draw bat stage elbow joint natural torsion, sagging, center of gravity is moved on the foot at rear,
Work is ready to have an effect.
Before wave boost phase (T1-T2): first maximum point (T1) of service track is to next minimum before batting
It is worth point (T2).After ball is spilled over, racket, which continues up, to be put, and trunk is swung forward, makes body before batting in " bow " type.
Batting stage (T2-T3): when ball drops to hitting point, batting of swinging the bat rapidly upward keeps arm and body abundant
Stretching, extension.It swings the bat square stance, holds wrist strap of clapping hands and move the hip beats that forearm has a medial rotation, this is the key operations having an effect,
The This move strength such as the center of gravity of early period can be moved forward, kick one's legs, turn, swing the bat all are discharged on ball.
With the stage of waving (T3-T4): occurring after batting (T3), hold bat and continue to stretch to direction of stroke, utilize inertia arm
It slightly guides, racket is finally placed in left side of body.Smoothness unfold with waving the damage that can be reduced Ji Qiu to joint and muscle.
From the foregoing, it can be seen that some extreme point in the starting of each stage and end of a period homologous thread of process of serving a ball, because
The amount of this said extracted characterizes the distance of the vertical direction of different service stage durations and operation.
60 groups of service motion profiles that 1-4 extracts step 1-3 are divided into two groups: training group, test group.Training component
The service tracking quantity obtained is nx=0.75 × 60=0.75 × 46+0.75 × 14=45, test group serve a ball tracking quantity as nc=
0.25 × 60=0.25 × 46+0.25 × 14=15.
In step 2, the step of motion profile acquired by step 1 extracts feature vector, is as follows:
The extreme point of the 2-1 record service track y, and following value: first minimum point to first is recorded according to extreme point
The time t of a maximum point experience0, the vertical range d that passes by0；Second minimum point to second maximum point undergo when
Between t2, the vertical range d that passes by2；The time t that second maximum point is undergone to third minimum point3, pass by it is vertical away from
From d3；The time t that first minimum point is undergone to third minimum point；The y of maximum of points in one service period is sat
Mark d.
2-2 calculates following feature: x according to the data of step 2-1 record0=t0/t；x1=t1/t；x2=t2/t；x3=t3/
t；x4=d0/(d×x0)；x5=d1/(d×x1)；x6=d2/(d×x2)；x7=d3/(d×x3)；x8=d2/t2；x9=d/t.
2-3 further extracts feature according to the step 2-2 characteristic provided: calculating x in step 2-20To x3This group of number
According to variance δ1, x4To x7The variance δ of this group of data2, enable x10=δ1, x11=δ2。
2-4 constitutes 11 dimensional feature vectors: X={ x according to the feature that step 2-2,2-3 is extracted0,x1,···x11}。
2-5 extracts training group n respectivelyxThe feature vector, X of a sample indicates are as follows:Wherein Xi∈R11, be i-th group of sample extraction the feature containing 11 features to
Amount, Yi∈ { -1,1 }, i=1,2 ..., nx, as vector XiBe successfully serve a ball trajectory extraction feature when, Yi takes 1, otherwise takes -1.
In step 3, population-supporting vector machine model (abbreviation PSO-SVM is trained for the feature vector that step 2 obtains
Model) the step of are as follows:
3-1 selects core of the Radial basis kernel function as SVM, and Radial basis kernel function is shown below:
X in above formulajFor sample data to be processed, xcFor kernel function center, σ is the width parameter of function, controls function
Radial effect range.
The step 2-5 feature vector extracted is divided into 4 groups by 3-2, carries out cross validation parameter σ and punishment parameter C to SVM,
Wherein 0.1≤C≤100,0.01≤σ≤1000.
3-3 passes through PSO algorithm optimization punishment parameter C and kernel functional parameter σ value.
MATLAB R2012b is used herein, is Intel Xeon (R) E5-2620 v3 CPU in experimental situation
On the machine of 2.40GHz, 32.0GB memory, PSO_SVM tennis service model is realized.The training process of 45 groups of feature vectors
Are as follows: the feature vector of extraction is read in memory, when SVM cross validation parameter, is optimized by PSO algorithm, is come
Determine optimal penalty parameter c and kernel functional parameter g.The default local search ability of PSO algorithm is c1=1.5, default is global
Search capability is c2=1.7, maximum evolution number maxgen=100, maximum population number sizepop=20, rate more new formula
Coefficient of elasticity before middle speed is ω v=1, and the maximum value and minimum value of SVM parameter c is respectively 100 and 0.1, SVM parameter σ
Maximum value and minimum value be respectively 1000 and 0.01.PSO algorithm returns to the value of optimal parameter c and σ, is respectively as follows: bestc=
0.1, bestg=738.8443.We first train PSO_SVM network with 45 groups of service data, then use 15 groups of data as
Test set examines this model, tests training effect.
Program is run altogether 10 times, generates 15 error amounts every time, is then taken to error amount runing time absolutely average
Value, the results are shown in Table 1.
1 PSO_SVM simulated effect of table
Claims (5)
1. a kind of ball service training mate method based on motion profile, which comprises the following steps:
Step 1: the motion profile of racket when obtaining multiple ball services；
Step 2: extracting feature vector by multiple motion profiles that step 1 acquires；
Step 3: the feature vector obtained for step 2, training obtain population-supporting vector machine model；
Step 4: the racket motion profile for treating evaluation is searched, and is extracted and is set out in motion profile according to service track characteristic
Sphere path curve, trained population-supporting vector machine model in input step 3 carry out the evaluation of service success or not；
In step 2, the step of motion profile acquired by step 1 extracts feature vector, is as follows:
The extreme point of the 2-1 record service track y, and following value: first minimum point to first pole is recorded according to extreme point
The time t of big value point experience0, the vertical range d that passes by0；The time that second minimum point is undergone to second maximum point
t2, the vertical range d that passes by2；The time t that second maximum point is undergone to third minimum point3, the vertical range passed by
d3；The time t that first minimum point is undergone to third minimum point；The y-coordinate of maximum of points in one service period
d；
2-2 calculates following feature: x according to the data of step 2-1 record0=t0/t；x1=t1/t；x2=t2/t；x3=t3/t；x4
=d0/(d×x0)；x5=d1/(d×x1)；x6=d2/(d×x2)；x7=d3/(d×x3)；x8=d2/t2；x9=d/t；
2-3 further extracts feature according to the step 2-2 characteristic provided: calculating x in step 2-20To x3This group of data
Variance δ1, x4To x7The variance δ of this group of data2, enable x10=δ1, x11=δ2；
2-4 constitutes 11 dimensional feature vectors: X={ x according to the feature that step 2-2,2-3 is extracted0,x1,…x11}；
2-5 extracts training group n respectivelyxThe feature vector, X of a sample indicates are as follows:
Wherein Xi∈R11, it is the feature vector containing 11 features of i-th group of sample extraction, Yi∈ { -1,1 }, i=1,2 ..., nx, when
Vector XiBe successfully serve a ball trajectory extraction feature when, Yi takes 1, otherwise takes -1.
2. the ball service training mate method based on motion profile as described in claim 1, which is characterized in that in step 1, obtain
Specific step is as follows for the motion profile of racket when service:
Mark point is placed on racket by 1-1；
1-2 acquires mark point motion profile, i.e. the motion profile of racket, and only direction perpendicular to the ground in retaining space coordinate system
Y-axis motion profile is stand-by, acquires n group altogether, wherein service successful trail n1Group, service failure track n2Group, wherein n1+n2=n；
1-3 intercepts the motion profile of service process in the motion profile that step 1-2 is obtained；
N group that step 1-3 is extracted service motion profile is divided into two groups by 1-4: training group and test group.
3. the ball service training mate method based on motion profile as claimed in claim 2, which is characterized in that mark point is placed in ball
The bulb top of bat is far from bat larynx side.
4. the ball service training mate method based on motion profile as described in claim 1, which is characterized in that in step 3, for
The step of feature vector training population-supporting vector machine model that step 2 obtains are as follows:
3-1 selects Radial basis kernel function as population-supporting vector machine model core, and Radial basis kernel function is shown below:
X in above formulajFor sample data to be processed, xcFor kernel function center, σ is the width parameter of function, controls the radial direction of function
Sphere of action；
The step 2-5 feature vector extracted is divided into 4 groups by 3-2, carries out cross validation parameter to population-supporting vector machine model
σ and punishment parameter C, wherein 0.1≤C≤100,0.01≤σ≤1000；
3-3 passes through PSO algorithm optimization punishment parameter C and kernel functional parameter σ value.
5. the ball service training mate method based on motion profile as described in claim 1, which is characterized in that in step 4 into
The detailed process of the evaluation of row service success or not are as follows:
4-1 obtains the track data of the signal point y placed at racket vertex to be evaluated, when encountering the minimum point on track
When, the extreme point of the curve after sequential hunting in 4 seconds；
4-2 is maximum point, minimum point, again maximum point, minimum point when the sequence that the extreme point of subsequent search occurs, and
Meet the latter maximum point value greater than previous maximum point, first minimum point value that subsequent searches are encountered is greater than
When first minimum point that entire search process is encountered, it is believed that find a service track, and the track input step 3 that will serve a ball
In population-supporting vector machine model carry out service success or not judgement；
If 4-3 service failure, by the standard service trajectory extraction in the feature vector and model of this service trajectory extraction
Feature vector be compared, to judge there are gaps with standard operation in which stage delivery of service, to be directed to
Property training, corrected.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710225245.8A CN107180423B (en) | 2017-04-07 | 2017-04-07 | A kind of ball service training mate method based on motion profile |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710225245.8A CN107180423B (en) | 2017-04-07 | 2017-04-07 | A kind of ball service training mate method based on motion profile |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107180423A CN107180423A (en) | 2017-09-19 |
| CN107180423B true CN107180423B (en) | 2019-08-20 |
Family
ID=59830495
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710225245.8A Active CN107180423B (en) | 2017-04-07 | 2017-04-07 | A kind of ball service training mate method based on motion profile |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107180423B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112621773A (en) * | 2020-12-07 | 2021-04-09 | 陈贺龄 | Partner training robot control method, system, device and storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101833232A (en) * | 2010-04-20 | 2010-09-15 | 浙江大学 | Visual support and match analysis system for ping-pong match and method for running same |
| US8175326B2 (en) * | 2008-02-29 | 2012-05-08 | Fred Siegel | Automated scoring system for athletics |
| CN104107042A (en) * | 2014-07-10 | 2014-10-22 | 杭州电子科技大学 | Electromyographic signal gait recognition method based on particle swarm optimization and support vector machine |
| CN104268511A (en) * | 2014-09-17 | 2015-01-07 | 河海大学常州校区 | Tennis pattern recognition system and method based on three-axis acceleration sensor |
| CN104606891A (en) * | 2015-02-11 | 2015-05-13 | 浙江海洋学院 | Network remote interaction tennis system |
-
2017
- 2017-04-07 CN CN201710225245.8A patent/CN107180423B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8175326B2 (en) * | 2008-02-29 | 2012-05-08 | Fred Siegel | Automated scoring system for athletics |
| CN101833232A (en) * | 2010-04-20 | 2010-09-15 | 浙江大学 | Visual support and match analysis system for ping-pong match and method for running same |
| CN104107042A (en) * | 2014-07-10 | 2014-10-22 | 杭州电子科技大学 | Electromyographic signal gait recognition method based on particle swarm optimization and support vector machine |
| CN104268511A (en) * | 2014-09-17 | 2015-01-07 | 河海大学常州校区 | Tennis pattern recognition system and method based on three-axis acceleration sensor |
| CN104606891A (en) * | 2015-02-11 | 2015-05-13 | 浙江海洋学院 | Network remote interaction tennis system |
Non-Patent Citations (2)
| Title |
|---|
| 一种新的羽毛球挥拍轨迹识别方法;崔婀娜 等;《东北大学学报（自然科学版）》;20170131;第38卷(第1期);全文 * |
| 基于视频图像处理技术的网球最佳击球点预测;周志湖 等;《工业控制计算机》;20170228;第30卷(第2期);论文第0、2-3节 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107180423A (en) | 2017-09-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20200098113A1 (en) | Stereoscopic image capture with performance outcome prediction in sporting environments | |
| US8636605B2 (en) | IR system for kinematic analysis | |
| Ahmadi et al. | Towards a wearable device for skill assessment and skill acquisition of a tennis player during the first serve | |
| CN110929596A (en) | Shooting training system and method based on smart phone and artificial intelligence | |
| CN105080111A (en) | Sport ball motion monitoring methods and systems | |
| CN106823345A (en) | Sports monitoring method and system | |
| KR100907704B1 (en) | Golfer's posture correction system using artificial caddy and golfer's posture correction method using it | |
| US20170272703A1 (en) | Athletic performance data acquisition systems, apparatus, and methods | |
| Li et al. | Baseball swing pose estimation using openpose | |
| CN111514568A (en) | Application of control device of table tennis serving machine in comprehensive athletic training of athletes | |
| CN103940422A (en) | System for acquiring composite posture parameters of shooting and archery | |
| CN107180423B (en) | A kind of ball service training mate method based on motion profile | |
| Paradisis et al. | Kinematic analysis of golf putting for expert and novice golfers | |
| CN114049590A (en) | Video-based ski-jump analysis method | |
| Taylor et al. | Intraindividual movement variability within the 5 m water polo shot | |
| CN110013658A (en) | A kind of dedicated tennis of tennis training pushes service control system and method | |
| Pavlov et al. | Comparison of spike techniques in college beach and classical volleyball | |
| Utomo et al. | Analysis of Biomechanics Slap Hit and Push in The Field Hockey | |
| Kim et al. | Motion analysis of squash backhand drop shot–A kinematic analysis study | |
| CN115414646B (en) | Basketball auxiliary training method and device | |
| CN113975772B (en) | Swing efficiency-based swing posture correcting method for players | |
| WO2014123419A1 (en) | Motion tracking method and device | |
| Chen et al. | Application of Artificial Intelligence Vision Technology in Chinese Fencing Teaching | |
| US11707663B1 (en) | System for tracking, locating and predicting the position of a ball in a game of baseball or similar | |
| EP4325443A1 (en) | Data processing apparatus and method |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20201014 Address after: 226200 No. 46 Industrial Park, Nanyang Town, Nantong, Jiangsu, Qidong Patentee after: Qidong plant medical equipment factory Address before: 310018, No. 928, No. 2, Poplar Street, Hangzhou economic and Technological Development Zone, Zhejiang Patentee before: Zhejiang University of Technology |
|
| TR01 | Transfer of patent right |