CN115497626A

CN115497626A - Body health assessment method based on joint point identification

Info

Publication number: CN115497626A
Application number: CN202211268388.4A
Authority: CN
Inventors: 刘铮
Original assignee: Beijing Xinqing Tech Co ltd
Current assignee: Beijing Xinqing Tech Co ltd
Priority date: 2022-10-17
Filing date: 2022-10-17
Publication date: 2022-12-20

Abstract

The invention discloses a body health assessment method based on joint point identification, belonging to the technical field of human body function assessment and test, and the method comprises the following steps: acquiring a human body image of a tester completing a preset test action, which is acquired by a depth camera; extracting space coordinate data of the body joint mark points of the tester from the human body image; calculating joint angle data of the body of the tester according to the space coordinate data, wherein the joint angle data comprise an angle, an angular velocity and an angular acceleration; and according to the space coordinate data and the joint angle data, respectively evaluating the joint activity, the action completion quality and the body consistency level of the completed action, and calculating to obtain a joint activity score, an action stability score and a body symmetry score. The method has comprehensive evaluation and high evaluation accuracy.

Description

Body health assessment method based on joint point identification

Technical Field

The invention relates to the technical field of human body function evaluation and test, in particular to a body health evaluation method based on joint point identification.

Background

With the advancement of science and technology, people gradually develop from the age with treatment as the main thing to the age with prevention as the main thing, and the concepts of physical fitness and health management are more and more appeared in the current life, so that it is very important to perform reasonable human body function evaluation tests on people and intervene training to perform process management.

Chinese patent application CN112102947A discloses an apparatus and a method for body posture assessment, the apparatus comprising: a data acquisition module that acquires an image including a body to be evaluated and detects the body from the image; the prediction module predicts the positions of key points of the body according to a preset key point detection model; the calculation module is used for calculating health indexes corresponding to the positions of the key points according to a preset health standard database; and the evaluation module compares the calculated health index with a reference index in the health standard database and evaluates the body based on the comparison result. In an embodiment, the apparatus may further include a training module to train a data set based on the labeled positions of the plurality of body samples corresponding to the human skeletal positions obtained with the touch to obtain a preset keypoint detection model. Through the device for evaluating the body postures, the standard of the normal value of each body posture of the children and the teenagers can be established, and a feasible basis is provided for the health detection of the children and the teenagers.

The patent application is designed based on a 2D modeling model, is mostly suitable for evaluation tests of children and teenagers about body postures, does not contain all age stage standards in a health standard database, is relatively thin in evaluation, only evaluates parts related to the body postures, and does not relate to other index evaluations related to body fitness.

Chinese patent CN108597578B discloses a human body movement evaluation method based on a two-dimensional skeleton sequence, which includes: acquiring an RGB image containing a human body through a camera, and extracting a two-dimensional human body skeleton, wherein the human body skeleton is provided with a plurality of joint points; correcting the initial pose of the human skeleton; extracting four types of characteristics of joint point coordinates, key part angles, joint point speeds and fluctuation errors of the joint point speeds to form a characteristic matrix, and constructing a corresponding frame difference degree calculation model; establishing a matrix coordinate system of the test skeleton sequence and the standard skeleton sequence, searching the front and rear sections of the two sections of sequences to obtain the starting and ending points of a matching path, and carrying out global bidirectional parallel search by taking the starting and ending points as reference points until the optimal path matching of the test skeleton sequence and the standard skeleton sequence is completed to obtain the cumulant difference value of the two sections of sequences so as to realize human motion evaluation. The invention adopts the low-cost RGB camera to accurately evaluate and trend analyze the human motion, and has better application value in the aspects of motion training and guidance such as sports, martial arts and dancing.

Although the two-dimensional human skeleton is extracted by using the RGB image of the camera, the complexity of wearable equipment is avoided, a stereoscopic human model cannot be established due to 2D projection, so that no data comparison is performed in the depth direction (Z axis) of a space in the calculation, and the reliability and stability of the evaluation are doubtful.

Disclosure of Invention

In view of this, the embodiment of the present invention provides a method for evaluating physical health based on joint point identification, which is comprehensive in evaluation and high in accuracy.

In order to solve the above technical problems, embodiments of the present invention provide the following technical solutions:

a physical fitness assessment method based on joint point identification, comprising:

acquiring a human body image which is acquired by a depth camera and used for a tester to finish a preset test action;

extracting space coordinate data of the body joint mark points of the tester from the human body image;

calculating joint angle data of the body of the tester according to the space coordinate data, wherein the joint angle data comprise angles, angular velocities and angular accelerations;

and according to the space coordinate data and the joint angle data, respectively evaluating the joint activity, the action completion quality and the body consistency level of the completed action, and calculating to obtain a joint activity score, an action stability score and a body symmetry score.

Further, the body joint mark points of the tester comprise at least 30 joint mark points which are respectively: head, neck, upper spine, middle spine, bottom spine, left/right shoulders, left/right elbows, left/right wrists, left/right hands, left/right hips, left/right knees, left/right ankles, left/right feet, left/right fingertips, left/right thumbs, nose, left/right eyes, left/right ears.

Further, the joint angle data includes data of at least 19 joint angles, the joint angles including: the cervical region, mid-spine, left/right upper spine, mid-spine, left/right bottom spine, left/right shoulder, left/right elbow, left/right wrist, left/right hip, left/right knee, left/right ankle.

Further, the estimating joint activity, motion completion quality and body consistency level of completed motion according to the spatial coordinate data and the joint angle data, and calculating to obtain a joint activity score, a motion stability score and a body symmetry score, includes:

for the joint motion degree, projecting the angle of the joint to three planes, namely a coronal plane, a sagittal plane and a horizontal plane to obtain a projection angle variation range;

and comparing the projection angle change range with the normal movement range of the joint, and grading by adopting a percentile method to obtain the joint mobility score.

for the motion completion quality, establishing a Bayesian network structure chart by using the space coordinate data, the angular velocity and the angular acceleration of the joint;

setting a critical point for each index evaluation in the Bayesian network structure chart, and defining the critical point as 60 points according to 100 points;

establishing an initial condition probability table according to the test data;

and (4) calculating the test data by adopting a PCMHS algorithm to obtain an action stability score.

Further, the step of respectively evaluating the joint activity, the motion completion quality and the body consistency level of the completed motion according to the space coordinate data and the joint angle data, and calculating to obtain a joint activity score, a motion stability score and a body symmetry score includes:

establishing a Bayesian network structure chart by utilizing the spatial coordinate data, the angle of the joint, the angular velocity and the angular acceleration for the body consistency level of finishing the action;

establishing an initial condition probability table according to the test data;

and (4) calculating the test data by adopting a PCMHS algorithm to obtain a body symmetry score.

Further, the estimating joint activity, motion completion quality and body consistency level of completed motion according to the spatial coordinate data and the joint angle data, and calculating to obtain a joint activity score, a motion stability score and a body symmetry score, and then:

and evaluating the sports injury risk according to the joint activity score, the action stability score and the body symmetry score, and calculating to obtain the sports injury risk score.

Further, the evaluating the risk of sports injury according to the joint activity score, the action stability score and the body symmetry score, and calculating to obtain a sports injury risk score includes:

and taking the joint activity degree score, the action stability score and the body symmetry score as evaluation indexes, determining the weight coefficient of each evaluation index by an AHP-CRITIC mixed weighting method, carrying out comprehensive scoring, and calculating to obtain a sports injury risk score.

Further, the evaluating the risk of sports injury according to the joint activity score, the motion stability score and the body symmetry score, and calculating to obtain a sports injury risk score, then includes:

and evaluating the body health according to the joint activity score, the action stability score, the body symmetry score and the sports injury risk score, and calculating to obtain a body health evaluation score.

Further, the evaluating the physical health according to the joint activity score, the motion stability score, the body symmetry score and the sports injury risk score, and calculating the physical health evaluation score includes:

and determining the weight coefficient of each evaluation index by using the joint activity score, the action stability score, the body symmetry score and the sports injury risk score as evaluation indexes, fusing according to the minimum discrimination information principle to obtain comprehensive weight, and calculating to obtain a body health evaluation score.

The invention has the following beneficial effects:

according to the body health assessment method based on joint point identification, firstly, a human body image of a tester completing a preset test action and acquired by a depth camera is acquired, then, space coordinate data of a mark point of a joint of the tester body is extracted from the human body image, joint angle data of the tester body are calculated according to the space coordinate data, the joint angle data comprise an angle, an angular velocity and an angular acceleration, and finally, joint activity, action completion quality and body consistency level of the completed action are evaluated respectively according to the space coordinate data and the joint angle data, and joint activity score, action stability score and body symmetry score are calculated. Therefore, the joint movement degree (movement degree), the movement finishing quality (stability) and the body consistency level (symmetry) for finishing the movement are evaluated, and the evaluation is comprehensive; in addition, the 3D human body image is used, a three-dimensional human body model is established, data comparison is carried out in the depth direction (Z axis) of the space, and the evaluation accuracy is high.

Drawings

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

FIG. 1 is a schematic flow chart of a method for evaluating health based on joint recognition according to the present invention;

FIG. 2 is a schematic view of the joint marking points of the subject;

FIG. 3 is a schematic view of the three projection angles of the joint space angle in the coronal plane, the sagittal plane, and the horizontal plane of the present invention;

FIG. 4 is a schematic diagram of a method for assessing physical fitness based on joint recognition according to the present invention;

FIG. 5 is a diagram of a Bayesian network architecture for the quality of action completion and the level of physical consistency established in the present invention;

FIG. 6 is a directed acyclic graph of the present invention with consistent levels of motion completion quality and body consistency;

FIG. 7 is a diagram showing a final interface of the evaluation results according to the embodiment of the method for evaluating physical fitness based on joint recognition of the present invention;

fig. 8 is a second display diagram of an interface of the final evaluation result in an embodiment of the joint point identification-based physical health evaluation method of the present invention.

Detailed Description

Embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

It should be understood that the described embodiments are only some embodiments of the invention, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

The embodiment of the invention provides a body health assessment method based on joint point identification, as shown in fig. 1, comprising:

step 101: acquiring a human body image of a tester completing a preset test action, which is acquired by a depth camera;

in this step, the test action can be flexibly selected according to the needs, and may include one or more of wall angel, over-top deep squat, single foot balance, trunk rotation and side bow step, for example. The tester is instructed to complete the testing action, and simultaneously the human body image is collected through a camera, wherein the camera is preferably a depth camera with a 3D function.

Step 102: extracting space coordinate data of the body joint mark points of the tester from the human body image;

in the step, a visual recognition system can be used for extracting the space coordinate data of the mark points of the body joints of the testee from the human body image.

As an alternative embodiment, the body joint marker points of the tester may include at least 30 joint marker points, which are respectively: head, neck, upper spine, middle spine, bottom spine, left/right shoulders, left/right elbows, left/right wrists, left/right hands, left/right hips, left/right knees, left/right ankles, left/right feet, left/right fingertips, left/right thumbs, nose, left/right eyes, left/right ears.

The extracted spatial coordinate data may be, for example, as follows:

ι _i ＝(x _i ，y _i ，z _i ) The coordinates of the ith joint marker point are expressed, and the set of tester body joint markers points is expressed as R (x, y, z). 30 human body mark points in each time sequence are acquired in the test process.

The subject's body joint marker points and numbers may be as shown in table 1 and fig. 2.

TABLE 1 Joint markers and numbering

Numbering	Mark point	Number of	Mark point
				0	Bottom of spinal column	15	Left foot
1	Middle part of the spine	16	Right hip
				2	Neck	17	Right knee
3	Head with a rotatable shaft	18	Right ankle
				4	Left shoulder	19	Right foot
5	Left elbow	20	Upper part of the spinal column
				6	Left wrist	21	Left hand tip
7	Left hand	22	Left thumb
				8	Right shoulder	23	Tip of right hand
9	Right elbow	24	Right thumb
				10	Right wrist	25	Nose
11	Right hand	26	Left eye
				12	Left hip	27	Left ear
13	Left knee	28	Right eye
				14	Left ankle	29	Right ear

The joint mark points defined in the invention are usually defined corresponding to joints or body surface bony markers with certain degree of freedom on human bodies, the current state of the human body is estimated by calculating the relative positions of the joint points of the human body in a three-dimensional space, and due to the particularity of the structure of the human body, connecting lines among the mark points are only connected by the structure of the human body.

Step 103: calculating joint angle data of the body of the tester according to the space coordinate data, wherein the joint angle data comprise angles, angular velocities and angular accelerations;

as an alternative embodiment, the joint angle data may include data of at least 19 joint angles, the joint angles including: neck, mid-spine, left/right upper spine, mid-spine, left/right bottom spine, left/right shoulder, left/right elbow, left/right wrist, left/right hip, left/right knee, left/right ankle.

Definition of joint angle of tester body

Because the fixed points of the muscle of the human body are different when the muscle works, the muscle is divided into a near fixation point, a far fixation point, an upper fixation point, a lower fixation point and a non-fixation point. Muscle start and stop points are specifically defined as: the point of attachment of the muscle near the torso or near the median plane of the body is called the origin, the point of attachment away from the torso or from the median plane is called the apex, and the muscle origin and apex are constant. When muscles contract, if the starting point is relatively fixed, the starting point is called near fixation, and if the stopping point is relatively fixed, the stopping point is called far fixation; the trunk muscles (e.g., rectus abdominis) are parallel to the long axis of the body, and when these muscles contract, the upper end is called upper fixation if the attachment point is relatively fixed, the lower end is called lower fixation if the attachment point is relatively fixed, and neither the upper end nor the lower end is fixed, the contraction is called no fixation.

In the present invention, different fixation means are preferably defined by different joint angle start-stop points. The joint angles are defined in the order of the muscle start points in the movements of upper-fixation, lower-fixation and no-fixation (such as sit-up, roll-belly and two-head start) of the trunk muscles, and the joint angles and their start points and intermediate points are shown in table 2:

TABLE 2 Joint Angle definitions

Serial number	Name(s)	Starting point	Intermediate of	Endpoint
					1	Neck part	Upper part of the spinal column	Neck	Head with a rotatable shaft
2	Upper part of the spine	Mid-spine	Upper part of the spine	Neck
					3	Upper part of right vertebral column	Neck	Upper part of the spinal column	Right shoulder
4	Upper part of left vertebral column	Neck	Upper part of the spine	Left shoulder
					5	Middle part of the spine	Upper part of the spinal column	Middle part of the spine	Bottom of spinal column
6	Bottom of right vertebral column	Middle part of the spine	Bottom of spinal column	Right hip
					7	Bottom of left spine	Middle part of the spine	Bottom of spinal column	Left hip
8	Right wrist	Right elbow	Right wrist	Right thumb
					9	Right elbow	Right shoulder	Right elbow	Right wrist
10	Right shoulder	Upper part of the spinal column	Right shoulder	Right elbow
					11	Right hip	Bottom of spinal column	Right hip	Right knee
12	Right knee	Right hip	Right knee	Right ankle
					13	Right ankle	Right knee	Right ankle	Right foot
14	Left wrist	Left elbow	Left wrist	Left thumb
					15	Left elbow	Left shoulder	Left elbow	Left wrist
16	Left shoulder	Upper part of the spine	Left shoulder	Left elbow
					17	Left hip	Bottom of spinal column	Left hip	Left knee
18	Left knee	Left hip	Left knee	Left ankle
					19	Left ankle	Left knee	Left ankle	Left foot

The joint angle definition shown in Table 2 is also used for the distal immobilization of the lower limbs (e.g., deep squat, lateral bending, etc.) and the proximal immobilization of the four limbs.

In the action of upper limb far fixation (such as pull-up, handstand, etc.), the joint angle definition adopts the limb far end mark point as the starting point of the joint angle, and is specifically defined as shown in table 3:

TABLE 3 definition of joint angle

Serial number	Name (R)	Starting point	Intermediate (II)	Terminal point
					1	Neck part	Upper part of the spine	Neck	Head with a rotatable shaft
2	Upper part of the spinal column	Mid-spine	Upper part of the spinal column	Neck
					3	Upper part of right vertebral column	Neck	Upper part of the spine	Right shoulder
4	Upper part of left vertebral column	Neck	Upper part of the spinal column	Left shoulder
					5	Middle part of the spine	Upper part of the spine	Middle part of the spine	Bottom of spinal column
6	Bottom of right vertebral column	Middle part of the spine	Bottom of spinal column	Right hip
					7	Bottom of left spine	Middle part of the spine	Bottom of spinal column	Left hip
8	Right wrist	Right thumb	Right wrist	Right elbow
					9	Right elbow	Right wrist	Right elbow	Right shoulder
10	Right shoulder	Right elbow	Right shoulder	Upper part of the spinal column
					11	Right hip	Bottom of spinal column	Right hip	Right knee
12	Right knee	Right hip	Right knee	Right ankle
					13	Right ankle	Right knee	Right ankle	Right foot
14	Left wrist	Left thumb	Left wrist	Left elbow
					15	Left elbow	Left wrist	Left elbow	Left shoulder
16	Left shoulder	Left elbow	Left shoulder	Upper part of the spine
					17	Left hip	Bottom of spinal column	Left hip	Left knee
18	Left knee	Left hip	Left knee	Left ankle
					19	Left ankle	Left knee	Left ankle	Left foot

Subject body joint angle calculation

Here, the joint angle data of the tester's body may be calculated according to a conventional mathematical algorithm using the spatial coordinate data. For example, the mark point iota is used _ik0 、ι _ik And iota _ik1 Spatial angle θ representing joint point k ^k ，ι _ik0 And iota _ik1 Two marked points representing the adjacent joint points, and the joint angles of the two marked points are calculated as follows:

ι _ik0 ＝(x _ik0 ，y _ik0 ，z _ik0 ) Denotes the coordinates of the ik0 th mark point

ι _ik ＝(x _ik ，y _ik ，z _ik ) Denotes the coordinates of the ik th mark point

ι _ik1 ＝(x _ik1 ，y _ik1 ，z _ik1 ) Denotes the coordinates of the ik1 st mark point

Vector m = (x) ₁ ，y ₁ ，z ₁ ),n＝(x ₂ ，y ₂ ，z ₂ )

(1)x ₁ ，y ₁ ，z ₁ ＝(x _ik0 -x _ik )，(y _ik0 -y _ik )，(z _ik0 -z _ik )

(2)x ₂ ，y ₂ ，z ₂ ＝(x _ik1 -x _ik )，(y _ik1 -y _ik )，(z _ik1 -z _ik )

Extracting the space coordinates of the body joint mark points of the tester, reducing the quality of the testing action of the tester by modeling, calculating the motion trail and the angles of 19 joint angles of the body joint mark points of the tester, and storing the image information of the tester in the testing process.

The joint angle at the k-th joint angle in m frames is recorded as

The angular velocity and the angular acceleration of the joint angle are respectively calculated and recorded as

The space coordinate R (1) corresponding to the time sequence of the mark point, the angle theta (2) corresponding to the time sequence of the joint angle, the angular velocity omega (3) and the angular acceleration alpha (4) are extracted and calculated by the camera and the system.

In order to improve the accuracy of the evaluation, the data obtained by calculation can be preprocessed in the step.

Processing of spatial coordinate data:

after the identification of the mark points, data denoising (specifically, denoising by using a wavelet threshold method) can be carried out, then the acceleration of the mark points is obtained according to the coordinates and the time sequence of the mark points, the limit value of the data is set according to the upper limit of the acceleration which can be reached by a human body and the discussion result of an expert, and if the limit value is exceeded, the data is deleted. Through the treatment, the marking point sequence which is more accordant with the human body can be obtained.

During testing, a tester can do three times per action according to system prompts so as to collect data for many times and improve the testing accuracy. And the system carries out subsequent grading calculation on the three sections of data obtained after collection and processing.

Joint (spatial) angle data processing:

during testing, the tester performs three times per action to obtain three-segment space mark point sequence data, and three-segment joint angle sequence data are obtained through point-to-angle calculation. After the joint angle (angle) is obtained through calculation, the angular velocity and the angular acceleration corresponding to the three sections are obtained through time sequence calculation. And setting a threshold according to the discussion of experts and project personnel, and eliminating unreasonable angular acceleration, thereby reducing the occurrence frequency of the unreasonable angular acceleration and obtaining a sequence which is more in line with the activity mode of the joints of the human body.

The joint angular velocity records the instantaneous angular velocity (ω = Δ θ/Δ t), and the angular acceleration is calculated by the angular velocity (α = Δ ω/Δ t).

Twelve data including three-section space mark point sequences, joint angles, joint angular velocities and joint angular accelerations can be obtained in each action, each data is recorded from the system prompt sound, and the recording is stopped when the prompt sound is finished.

Step 104: and respectively evaluating the joint activity (activity), the motion completion quality (stability) and the body consistency level (symmetry) of the completed motion according to the space coordinate data and the joint angle data, and calculating to obtain a joint activity score, a motion stability score and a body symmetry score.

In order to improve the accuracy of the evaluation, the following data selection processing may be performed on the spatial coordinate data and the joint angle data in this step:

1. for joint mobility (mobility)

The joint space angle score calculation method comprises the following steps:

the system calculates the difference between the maximum value and the minimum value of the multiple data (the three data in the above example) obtained after collection and calculation, selects the difference between the multiple data to compare, and then carries out subsequent score calculation according to the maximum difference.

2. For action completion quality (stability)

Spatial coordinate data (joint spatial position) score calculation method:

the system calculates the difference between the maximum value and the minimum value of the multiple data (the three data in the above example) obtained after collection and calculation, and performs subsequent score calculation (representing the variation range between the initial action and the final action) according to the maximum difference after selecting the difference between the multiple data for comparison.

The joint angular velocity score calculation method comprises the following steps:

and calculating the angular velocity data of the corresponding joint through the angular velocity calculation of the system joint, calculating to obtain the average value of each group of data (namely the average value of the angular velocity of the joint of the test action of the tester), and performing subsequent score calculation according to the difference value between the average value of the angular velocity of the joint and the average value of the angular velocity of the joint corresponding to the standard action.

The joint angular acceleration score calculation method comprises the following steps:

the system acquires and calculates a joint angular acceleration data sequence, measures the curve similarity of the tested action and the standard action of the tester, calculates the similarity of the two sequences by adopting an Edit Distance on Real sequence (EDR), obtains the similarity degree between the change of the joint angular acceleration and the standard action when the tester performs the action, and performs subsequent score calculation.

3. Level of physical uniformity of the finished action (symmetry)

Spatial coordinate data (joint spatial position) score calculation method:

the system calculates the difference between the maximum value and the minimum value of the acquired and calculated multiple data (the three data in the above example), selects the difference between the multiple data to compare, and then performs subsequent score calculation (representing the variation amplitude between the initial action and the final action) according to the maximum difference.

The joint space angle score calculation method comprises the following steps:

the system calculates the difference between the maximum value and the minimum value of the acquired and calculated data (the three times of data in the above example) for multiple times, and calculates the maximum difference of each group of data. And performing difference calculation again by using the maximum difference value of each side of the left side and the right side for multiple times to obtain the angle difference value of the joints on the left side and the right side, averaging the left angle difference value and the right angle difference value for multiple times, and performing subsequent score calculation (average value of the multiple difference values) according to the average value.

and calculating angular velocity data of the corresponding joints on the left side and the right side through the angular velocity of the system joints, calculating to obtain an average value of each group of data (namely the average value of the angular velocities of the joints on the left side and the right side of the test action of the tester), and performing subsequent score calculation according to the difference value of the average values of the angular velocities of the joints on the left side and the right side.

the system respectively measures the curve similarity of the acquired and calculated angular acceleration data tracks of the left and right joints, calculates the similarity of two groups of sequences by adopting an edit distance method (EDR), and obtains the similarity of the angular acceleration of the left and right limb joints when the tester performs the action, so as to perform subsequent score calculation.

In this step, the joint mobility (mobility), the motion completion quality (stability), and the body consistency level (symmetry) of the completed motion are evaluated and calculated to obtain corresponding scores, and various methods that will be easily conceived by those skilled in the art may be employed, which will be described in detail later.

To sum up, the body health assessment method based on joint point identification provided by the embodiment of the present invention includes firstly acquiring a human body image of a tester completing a preset test action, which is acquired by a depth camera, then extracting spatial coordinate data of a mark point of a body joint of the tester from the human body image, then calculating joint angle data of the body of the tester according to the spatial coordinate data, wherein the joint angle data includes an angle, an angular velocity and an angular acceleration, and finally respectively evaluating joint activity, action completion quality and body consistency level of completed action according to the spatial coordinate data and the joint angle data, and calculating to obtain a joint activity score, an action stability score and a body symmetry score. Therefore, the joint movement degree (movement degree), the movement completion quality (stability) and the body consistency level (symmetry) for completing the movement are evaluated, and the evaluation is comprehensive; in addition, the 3D human body image is used, a three-dimensional human body model is established, data comparison is carried out in the depth direction (Z axis) of the space, and the evaluation accuracy is high.

As an alternative embodiment, the evaluating the joint activity, the motion completion quality and the body consistency level of the completed motion according to the spatial coordinate data and the joint angle data respectively, and calculating a joint activity score, a motion stability score and a body symmetry score (step 104), may include:

step 105: and evaluating the body health according to the joint activity degree score, the action stability score and the body symmetry score, and calculating to obtain a body health evaluation score.

Therefore, the body health evaluation score is comprehensively obtained according to the three scores (the joint activity score, the action stability score and the body symmetry score), so that a comprehensive evaluation is given to the user, the use habit of the user is met, and the use experience of the user is better.

As another alternative embodiment, the evaluating the joint activity, the motion completion quality, and the body consistency level of the completed motion according to the spatial coordinate data and the joint angle data, and calculating a joint activity score, a motion stability score, and a body symmetry score (step 104), may include:

step 1041: for the joint motion degree, projecting the angle of the joint to three planes, namely a coronal plane, a sagittal plane and a horizontal plane to obtain a projection angle variation range;

in the present invention, "degree of joint movement" (degree of movement) is used to evaluate the extent and degree of joint movement function impairment, and is one of the bases for specifying a movement plan and evaluating a movement training effect. The joint activity degree examination is evaluated according to the joint (activity) angle (without using other data, see fig. 4), the system can adopt a neutral position method 0 degree method commonly used in the world, an anatomical limb position is taken as a 0 degree position, and the joint activity angle is recorded according to joint space angle projection angles of the limb in flexion, extension, adduction, abduction, internal rotation, external rotation and the like on a coronal plane, a sagittal plane and a horizontal plane motion plane respectively.

According to three projection angles of joint space angles in a coronal plane (X axis and Y axis), a sagittal plane (Y axis and Z axis) and a horizontal plane (X axis and Z axis), as shown in fig. 3, the system evaluates different motion modes of joints, and the motion angle ranges of each joint and the corresponding joint are shown in the following table 4:

TABLE 4 articulation patterns and ranges of motion

Step 1042: and comparing the projection angle change range with the normal movement range of the joint, and grading by adopting a percentile method to obtain the joint mobility score.

In the step, when the joint mobility is judged, the angle changes of different motion modes (such as bending, rotating and the like) of the joint are judged when a tester performs test actions according to the angle changes of the projection angles of the joint space angles on the coronal plane, the sagittal plane and the horizontal plane, so that the joint mobility is scored in the motion process.

Judging the situation of limited movement by combining the normal movement range of the joint, taking various measurement results of the flexion movement range of the knee joint as an example, and indicating no limited movement by the movement range of 0-150 degrees; the range of motion is 20-150 degrees to indicate limited extension; the range of motion is 0-120 degrees, which indicates that the flexion is limited; 20-120 indicates that knee flexion and extension are both limited. And according to the motion range angles of the joints with limitation, no limitation and different motion modes, which are given by experts, grading standard division is carried out on the measured joint motion degree.

In specific implementation, the percentile method can be adopted to complete modeling of indexes of each joint according to data in an expert database and a system database. The percentile method generally establishes standards of different grades according to percentiles such as P3, P10, P25, P35, P50, P65, P75, P90, P95 and P97, and the system refers to experts and multiple research references and sets the percentiles such as P3, P10, P35, P60, P75, P85 and P95 to establish the standards of different grades.

The joint motion angle is divided into range scores, namely the motion angle is the highest score in the normal range, and exceeds or is lower than the normal range, and the joint motion range is abnormal, such as excessive joint motion and insufficient joint motion, which is closely related to muscle force imbalance, bone lesion, soft tissues around joints and the like. The system thus stratifies and quantifies the range of joint motion based on expert and multiple research results, as well as a large number of samples in a database.

Thus, the joint movement score can be obtained through the above steps 1041 to 1042.

As still another alternative, the evaluating the joint activity, the motion completion quality, and the body consistency level of the completed motion according to the spatial coordinate data and the joint angle data, and calculating a joint activity score, a motion stability score, and a body symmetry score (step 104) may include:

step 1041': for the motion completion quality, establishing a Bayesian network structure chart by using the space coordinate data, the angular velocity and the angular acceleration of the joint;

step 1042': setting a critical point for each index evaluation in the Bayesian network structure chart, and defining the critical point as 60 points according to 100 points;

step 1043': establishing an initial condition probability table according to the test data;

in a step 1044': and calculating the test data by adopting a PCMHS algorithm to obtain a motion stability score.

Thus, the motion stability score can be obtained through the above steps 1041 '-1044'.

In the present invention, the test of "quality of motion completion" (stability) is to evaluate the stability of the body of the tester, and the spatial position of the joint point, the angular acceleration of the joint, and the angular velocity of the joint, which are collected and calculated by the camera and the system, are used for evaluation (without using the angle of the joint, see fig. 4).

The joint mark points (joint point space positions) represent the similarity degree between the final position of the test action of the tester and the action mark points in the standard action library; the joint space angular velocity represents the degree of similarity between the average velocity of the movement completion of the tester and the movement in the standard movement library; the angular acceleration of the joint space angle represents the degree of similarity between the change of the action joint angle of the tester and the action in the standard action library, so that the mutual coordination of muscle groups around the corresponding joint and ligaments of the tester in the motion process, the signal conduction of proprioception in the joint and the control capability of the joint on the muscles are judged.

Poor joint stability indicates that the joint has the problems of unbalanced peripheral muscle strength, poor proprioceptive function, uncoordinated transmission of a power chain and the like, and the problems can cause unstable joint and weakened muscle cooperative contraction action to increase the risk of joint injury, such as meniscus injury, ankle sprain and other lesions.

As another alternative embodiment, the evaluating the joint activity, the motion completion quality, and the body consistency level of the completed motion according to the spatial coordinate data and the joint angle data, and calculating a joint activity score, a motion stability score, and a body symmetry score (step 104), may further include:

step 1041": establishing a Bayesian network structure chart by utilizing the spatial coordinate data, the angle of the joint, the angular velocity and the angular acceleration for the body consistency level of finishing the action;

step 1041": setting a critical point for each index evaluation in the Bayesian network structure chart, and defining the critical point as 60 points according to 100 points;

step 1041": establishing an initial condition probability table according to the test data;

step 1041": and (4) calculating the test data by adopting a PCMHS algorithm to obtain a body symmetry score.

Thus, the body symmetry score can be obtained through the above steps 1041 "-1044".

In the invention, the test of the body consistency level (symmetry) of the finished action is to evaluate the body symmetry of the tester, and the similarity of the four points on the left and right limbs of the body is evaluated by using the joint point space position, the joint space angle, the joint space angular velocity and the angular acceleration (all data need to be used, see fig. 4) acquired and calculated by a camera and a system. The larger the difference between the position, angle, angular velocity, and angular acceleration at the same time in space is, the lower the body symmetry is.

The following further describes the specific procedures of the above steps 1041'-1044' and 1041"-1044", as follows:

first, a bayesian network structure diagram of the motion completion quality (stability) and the body consistency level (symmetry) of the completed motion is established, as shown in fig. 5;

secondly, setting a critical point for each index evaluation in the Bayesian network structure chart, specifically, for each evaluation method, setting the critical point according to a judgment standard and judgment of a professional, and defining the score of the critical point as 60 points according to 100 points;

then, the continuous data are formulated according to a rating table, 60 points are divided into 1 standard deviation downwards on average, the theoretical worst data are taken as 0 point, the limited area average value is taken as 30 points, the high risk area average value is taken as 75 points, the low risk area average value is taken as 85 points, the normal area average value is taken as 95 points, and the theoretical best data are taken as 100 points; respectively listing corresponding scores of different numerical values according to 0-60, 60-80, 80-90 and 90-100 subareas and equal step length to be used as an initial table, namely a conditional probability table;

finally, the probability of each item appearing on a different score is calculated and recorded as P (X) _i ) A conditional probability table is obtained. P (X) _i ) The representative point item X scores a probability of i.

The setting rule of the critical value ensures that 60 minutes of the system of the evaluation method is the 'limited' judgment standard of each test method and is also the controller of the system of the evaluation method, and the robustness (the robustness is the transliteration of Robus and refers to the noise resistance in the system) of the method is ensured, including two main aspects of (1) the accuracy of the test result and (2) the stability of the system operation.

From fig. 6, it can be seen that the motion completion quality and body consistency level conform to Directed Acyclic Graph (DAG) features. Data has no direction and no ring structure, and can not return to any node in the structure from the node.

Meanwhile, the structure of the method is a special DAG structure, and only one father node can be found in the two structures respectively, namely the probability of the action completion quality and the probability of the action completion body consistency level. Because of this particular data structure, the DAG structure for the quality of action completion and the level of physical consistency in bayesian in fig. 5 can be converted to a data distribution of a tree structure (a graph that is connected and has no loops is called a tree).

And (3) for the test data to be processed according to the tree structure:

1) The core of the test and evaluation results of the levels of the action completion quality and the action completion body consistency level is a probability problem, test data are processed according to a tree structure, the position of the test data in a tree database model is easier to query, and whether the test data are in a high risk area is further judged.

2) The data processing mode adopts an improved PCMHS algorithm in the MCMC algorithm, takes different test method categories as parallel Markov chains in a data structure, and has a convergence characteristic.

MCMC: abbreviation of Markov Chain Monte Carlo algorithm (Markov Chain Monte Carlo, MCMC).

PCMHS: the algorithm is an improved algorithm aiming at the conditions that the data fusion is reduced and the convergence speed in a data structure is slowed down after the data volume of the traditional MH algorithm is greatly increased, and is characterized in that a plurality of MH samples are simultaneously carried out, and a plurality of Markov chains which are parallel and converge on a certain constant (such as Boltzmann constant) distribution are constructed.

MH algorithm: it is an abbreviation of Metropolis Hasting, together with another gibbs algorithm, one of the two classical algorithms of Markov Chain Monte Carlo (MCMC).

Markov chain: markov Chains (MC) are stochastic processes (stochastic processes) in probability theory and mathematical statistics with Markov properties (Markov property) and existing within discrete index sets (index sets) and state spaces (state spaces).

3) The method can add newly obtained incremental data after Bayesian network modeling to a database of a tree structure (namely, DAG structure of action completion quality and body consistency level in Bayes in FIG. 5 is converted into data distribution of a tree (a connected graph without a loop is called as a tree)) structure in an iterative manner, and dynamically adjust probability based on the incremental data, so that the method keeps high correlation with an actual measurement result and approaches to a real situation.

Incremental data: incremental data is a noun of the database that corresponds to the full amount of data. The full data is all the data in the data table; incremental data refers to newly acquired data after the last export. All data obtained through testing in the method can participate in the adjustment of the testing method of the index, so that the method is ensured to have higher conformity with the actual condition, and the testing result is ensured to be more accurate.

The method adopts a PCMHS algorithm to test data and comprises the following two steps: initialization and iterative sampling.

The formula is illustrated below:

1. PCMHS initialization

(1) And converting the Bayes data network (figure 5) with the directed acyclic graph structure into a spanning tree structure with the maximum weight value through a maximum spanning tree algorithm. The node similarity is used as the weight of each edge and is the ratio of the number of the common neighbor nodes of the two nodes to the number of all the neighbor nodes. And selecting and adding the n-1 edge with the maximum weight value to the set T one by one, and finally generating a maximum spanning tree containing n nodes and the n-1 edge.

The steps of constructing the maximum spanning tree are as follows:

1) In the initialization set V _T = { x } and E _T {}: for a given directed acyclic network graph G (V, E), G is a directed acyclic graph, V is a set of nodes, and E is a set of edgesI.e., the set of arcs, T is the spanning tree, and X is the starting node of the spanning tree and is also the only parent node in the spanning tree structure.

2) Selecting the edge with the maximum weight value from the edge set E to be added into the set ET, wherein the node u is the set V _T Element (V) of (1), node V ∈ V, but V is not at V _T In (1).

3) Repeating the process of 2) until all nodes u are added to the set V _T In (1).

(2) By judging the mutual information MI (X) according to a given threshold value epsilon, the calculation formula is:

"∑": summing the symbols; log logarithmic sign; the "P" probability.

Through the two steps (1) and (2), the samples with quality better than that of the initial samples generated completely randomly can be obtained, and the smooth distribution of the sampling process is closer to the smooth distribution.

2. Iterative process for PCMHS

The parallel sampling algorithm PCMHS is used for designing a recommended distribution of single-chain sampling based on the sampling of arcs in the overall data, and designing a recommended distribution of double-chain intersection based on the sampling of local substructures. The proposed distribution calculation formula for arc sampling and substructure sampling is as follows:

(1)π _ij is node X in the posterior distribution P (G | D) of the network structure _i To X _J The probability of existence of a directed arc in between; a. The _ij Is node X in the current sample population _i To X _J There are individual numbers of directed arcs in between.

Hypothesis vector (pi) _ij ,π _ji ,1-π _ij -π _ji ) And the posterior probability dive super coefficient is (-1 + A) under the condition of Dirichlet prior distribution with super coefficients of 1 _ij ，1+A _ji ，1+г-A _ij -A _ji ) Dirichlet distribution,. Pi. _ij And pi _ji The posterior estimates of (a) are:

where r is 2 pi, dirichlet distribution is Dirichlet distribution, a continuous multivariate probability distribution family with vectors of positive and real numbers as parameters is used as prior distribution in bayesian statistics.

The meaning of a posterior distribution P (G | D) is the probability that the directed acyclic graph of the network structure belongs to a posterior distribution in a given sample set, P being the probability sign, G being the directed acyclic graph sign, D being the given sample set sign.

(2)K _y Is the probability of the parent-child structure corresponding to the node subset Y in the posterior distribution P (G | D) of the network structure, B _y Is the number of individuals in the current sample population for which such parent-child structures exist. If the probability parameter (K) _y ，1-K _y ) Obeying Dirichlet prior distribution with over-coefficient of 1, the posterior distribution of the probability parameter obeys over-coefficient (1 + B) _y ，1-B _y ) In the case of a Dirichlet distribution of phi _y The posterior estimation formula of (distribution function of y) is:

(3) And (3) aiming at the arc sampling and the substructure sampling in the steps (1) and whether the newly obtained incremental data are accepted or not, calculating the acceptance probability according to the following formula.

1) Assume that the current state is G ^(c) Then according to the recommendation distribution R (G) ⁽ⁿ⁾ |G ^(c) ) Generating the next state G ⁽ⁿ⁾ ；

2) The new data acceptance probability calculation formula is as follows:

i.e. 1 to

Of the measured value (c).

The resulting probability is converted into a percentile score, i.e., a stability and symmetry score.

As still another alternative, the evaluating the joint activity, the motion completion quality and the body consistency level of the completed motion according to the spatial coordinate data and the joint angle data respectively, and calculating a joint activity score, a motion stability score and a body symmetry score (step 104), may include:

step 105': and evaluating the sports injury risk according to the joint activity score, the action stability score and the body symmetry score, and calculating to obtain the sports injury risk score.

In this embodiment, the risk of the athletic injury is also evaluated, and an athletic injury risk score is calculated, so that the evaluation result is more comprehensive.

Further, the evaluating the sports injury risk according to the joint activity score, the motion stability score and the body symmetry score, and calculating a sports injury risk score (step 105'), may include:

step 1051': and taking the joint activity degree score, the action stability score and the body symmetry score as evaluation indexes, determining the weight coefficient of each evaluation index by an AHP-CRITIC mixed weighting method, carrying out comprehensive scoring, and calculating to obtain a sports injury risk score.

When a human body moves, a motion chain of the human body is formed by dynamic and ordered linkage of a plurality of links such as the head, the upper limbs, the trunk and the lower limbs in the state change process of the human body depending on a mechanical motion rule of a motion system consisting of bones, joints, muscles and nerves. The phenomenon that parts of links are relatively weak due to insufficient muscle strength, fatigue and injury in each link or links in the kinematic chain is called as 'weak link'. The action 'weak link' takes the asymmetry of the action, poor stability and the compensation of the occurrence of the action as main expression forms.

When a certain part of the body is overused, the kinetic chain of the part can have a weak balance state. In long-term movement, each link in the part is continuously lost, once weak balance is broken, the structure of the area is abnormal, local damage can be caused, compensatory action is further caused, and the damage forms a vicious circle. In addition, the phenomenon of habitual side can occur in the movements due to the asymmetry of the functionality of part of the movements during the movement, and the phenomenon of uneven muscle strength of one side of the limbs at other links and the opposite side of the movement chain can occur due to different frequent use of the limbs at two sides for a long time, so that the balance state is lost, and the movement injury is caused seriously.

Therefore, the frequent use of a certain link or links and one-sided limbs for a long time can cause the corresponding link to be over-tired, thereby causing the phenomenon of 'weak link' of action, and generating injury risk or athletic injury.

In this embodiment, the joint activity, the action completion quality, and the body consistency level score of completing the action are used as evaluation indexes, and a weight coefficient of each evaluation index is determined by an Analytic Hierarchy Process (AHP), an index weight determination method (CRITIC) based on index correlation, and an AHP-CRITIC hybrid weighting method, and a comprehensive score is used as a final evaluation index. The specific evaluation method may be as follows:

subjective and objective combination weight index:

the subjective weighting method (AHP) has more advantages than the objective weighting method (entropy weighting method) in determining the weight according to the intention of a decision maker, but has relatively poor objectivity and relatively strong subjectivity; the objective weighting method has objective advantages, but cannot reflect the degree of importance of decision makers to different indexes, and has certain weight and degree opposite to the actual indexes.

Therefore, when the weight of the index is assigned, the intrinsic statistical rules and authority values between the index data should be considered. Here, a reasonable decision index weighting method is provided, that is, a combined weighting method combining subjective weighting (AHP) and objective weighting (CRITIC) is adopted to make up for the deficiency of single weighting.

Subjective weighting method:

analytic Hierarchy Process (AHP) analyzes multiple target information by using mathematical logic thinking, information of the same level is matched and compared in the Process of processing the information, the proportion of relative importance among the information is calculated by taking a certain information in the previous level as a comparison criterion, a pair comparison judgment matrix is constructed, and the weight proportion of each information relative to the information of the previous level is calculated.

1) Constructing a decision matrix

The system refers to expert opinions, judges the importance among all indexes, determines the priority of all indexes as symmetry > stability > activity degree, and constructs a judgment matrix (shown in a table 5) for pair comparison.

TABLE 5 determination priority matrix for pairwise comparison of sports injury risks

Weight index	Symmetry of a fluid	Stability of	Degree of motion
				Symmetry property	1	2	3
Stability of	1/2	1	2
				Degree of motion	1/3	1/2	1

2) Consistency check

And solving the maximum characteristic value of the judgment matrix, calculating a consistency ratio CR value according to a formula, and considering that the judgment matrix passes consistency test when the CR is less than 0.1.

CR＝CI/RI

3) Solving for weights

For the judgment matrix passing the consistency test, the normalized characteristic vector is used as the weight vector to obtain the weight alpha _j (j＝1,2,…,n)。

Objective weighting method:

the objective assignment method uses CRITIC (criterion impact high intercritical Correlation) method, which comprehensively measures the objective weight of an index based on the contrast strength of an evaluation index and the conflict between indexes. And considering the variability of the indexes and the correlation among the indexes, and performing scientific evaluation by using the objective attributes of the data.

The total number of m samples, n indexes, x in the system _ij The value of the jth evaluation index of the ith sample is represented, and the evaluation matrix can be represented as:

the objective weight is calculated as follows:

1) Index normalization processing:

forward direction index (the larger the index value used, the better):

reverse index (smaller index value used is better):

2) Calculating the mean value

And standard deviation s _j ：

3) Calculating the coefficient of variation:

4) Calculating a correlation coefficient matrix:

ρ _ij ＝cov(y _k ,y _l )/(s _k ,s _j )，k＝1,2,…,n；l＝1,2,…,n

in the formula: ρ is a unit of a gradient _ij Cov (y) being the correlation coefficient between the i-th index and the j-th index _k ,y _l ) Representing the covariance between the kth index and the l index.

5) Calculating the information content of the index:

6) Determining the objective weight as:

calculating comprehensive weight:

to make the integrated weight ω _i As much as possible connectNear to

Solving the optimization model to obtain the comprehensive weight as follows:

as another alternative embodiment, the evaluating the sport injury risk according to the joint activity score, the motion stability score and the body symmetry score, and calculating a sport injury risk score (step 105'), may include:

step 106': and evaluating the physical health according to the joint activity degree score, the action stability score, the body symmetry score and the sports injury risk score, and calculating to obtain a physical health evaluation score.

Therefore, the body health assessment score is comprehensively obtained according to the four scores (the joint activity score, the action stability score, the body symmetry score and the sports injury risk score), so that a comprehensive evaluation is given to the user, the use habit of the user is met, and the use experience of the user is better.

Further, the evaluating the physical health according to the joint activity score, the motion stability score, the body symmetry score and the sports injury risk score, and calculating a physical health evaluation score (step 106') may include:

step 1061': and determining the weight coefficient of each evaluation index by using the joint activity score, the action stability score, the body symmetry score and the sports injury risk score as evaluation indexes, fusing according to the minimum discrimination information principle to obtain comprehensive weight, and calculating to obtain a body health evaluation score.

In this embodiment, for the comprehensive evaluation problem of physical health, four scores, i.e., joint mobility, motion completion quality, motion consistency level and motion damage risk, are used, a weight coefficient of each evaluation index is determined by using an Analytic Hierarchy Process (AHP) and an index weight determination method of index correlation (CRITIC) hybrid weighting method, and the weight coefficients are fused according to the minimum discrimination information principle to obtain a comprehensive weight. The specific evaluation method may be as follows:

subjective weighting method:

analytic Hierarchy Process (AHP) is a subjective weighting method whose main steps are described below.

1) Constructing a decision matrix

The judgment of the importance among the indexes is carried out according to the scale of 1-9 shown in Table 6 by referring to the expert opinions, and the judgment result is expressed by a ratio to construct a judgment matrix.

TABLE 6 meaning of scale values

Scale	Means of
		1	Of equal importance when compared to two elements
3	The former being slightly more important than the latter in comparison with the two elements
		5	The former is significantly more important than the latter when compared with the two elements
7	The former is extremely important than the latter in comparison with two elements
		9	Comparison of two elements, the formerMore important than the latter
2,4,6,8	Intermediate value representing the above-mentioned adjacency judgment

Is provided with n indexes x ₁ ，x ₂ ，…，x _n Subjectively sorting the indexes according to the principle that the importance degree is not reduced, determining a scale value and recording the corresponding scale as t _i If the transitivity according to the degree of importance is applied to other elements in the decision matrix, the final decision matrix R is as follows.

And (3) expressing the product of all elements in the ith row of the matrix R, so that the subjective weight of each index in the total score can be quantitatively determined as follows:

objective weighting method:

and calculating the joint activity, the action completion quality, the body consistency level of completed action and the weight coefficient of the sports injury risk score by using a CRITIC method.

AHP-CRITIC hybrid weighting method:

respectively calculating four weight coefficients according to AHP and CRITIC, and obtaining the weight coefficients according to the composite weight

Get the composite weight of the four scores。

After the evaluation score is given, a targeted exercise training plan can be given according to the physical exercise condition of the tester, and specifically, a rule model can be generated through the training plan according to the joint activity, the body symmetry, the action stability and the exercise injury risk of main body links of the tester, so that a targeted periodic exercise training scheme is generated.

The following illustrates the joint point recognition-based physical health assessment method of the present invention:

before evaluation the test person was told to wear a sports garment that was slightly tight, comfortable, and did not interfere with the natural movement towards the body, and that the material of the garment was as non-reflective as possible. If any discomfort, pain, occurs during the test, the staff can be instructed and the test stopped by himself.

The tester is explicitly informed of the evaluation purpose, the general flow and the attention. The basic condition of the testee is known and collected, the basic condition of the body of the testee, the previous injury condition and the like are preliminarily determined, and the potential risk in the evaluation process is eliminated.

The tester evaluates the voice and video guide of the action according to the system, and completes five action tests (wall angel, over-top deep squat, single-foot balance, trunk rotation and side bow step) along with the video.

The whole evaluation process is about ten minutes approximately, a tester finishes actions according to video guidance in the test process, the actions are finished naturally as far as possible on the premise of not causing pain and discomfort, and each action is complete and is kept consistent with the video guidance as far as possible.

Each action is repeated for three times, the prompt follows the guide video rhythm as much as possible, the test action is complete as much as possible, the body is kept from shaking as much as possible in the test process, and the gravity center of the body is kept stable.

And after the test action is finished, the system displays an evaluation report. The report first displays the overall assessment, and then scores the tester for four dimensions of activity, stability, symmetry, and risk of athletic injury. And then the report is displayed according to human body links. The human body is divided into five movement links including neck, shoulder upper limbs, trunk, pelvis and lower limbs, scores of the five movement links in three dimensions of activity, stability and symmetry are displayed, the three movement links are evaluated in a grading mode, and the three movement links are divided into three grades according to excellence, goodness and poor results. Meanwhile, the evaluation report also comprises evaluation of static posture, sports injury risk, insufficient/tense muscle force, joint activity trend graph, activity angle and the like.

According to the test evaluation result report, the system provides a complete six-time training plan, and improvement can be performed according to the current weak part of a tester. The tester can view the exercise action at the WeChat public number and perform exercises according to the demonstration action.

In the interface display figure for assessment results one (activity risk screening results):

activity-activity amplitude, physical fatigue, the higher the activity the better.

Stability-muscle strength, quality of action, the higher the stability the better.

Symmetry-body symmetry, the better the symmetry the lower the risk of movement.

Radar plot on the right: representing the comparison of three dimensions of activity, stability and symmetry.

Evaluation and grading: excellent (100-85); good (84-75); typically (74-60); to be increased (60 or less).

In this example, the final screening results are shown in fig. 7, and it can be seen from the evaluation test results that the degree of activity of the testers is good, which indicates that the degree of physical fatigue is low and the amplitude and flexibility of the activity are high. The stability score was general, indicating that the muscle strength was at an intermediate level. The general symmetry shows that the body is symmetrical, and the coordination ability is general, which relates to the potential motion risk level. Stability and symmetry need be progressively improved through the exercise of pertinence, improve the motion ability, avoid the injury risk.

In the interface display of the evaluation results two (motor fitness scores):

the left side is five parts of the body: neck, shoulder, upper limbs, torso, pelvis, and lower limbs, respectively. This part is five body parts, and is evaluated in three dimensions of mobility, stability, and symmetry.

In this example, the last given motor fitness score is shown in fig. 8, and it can be seen first from the evaluation that the stability of the neck of the test person is good and the mobility of the shoulder, back, upper limbs and pelvis is in a good range. This indicates that the neck has better muscle strength, and the shoulder upper limbs and pelvis have better flexibility, large range of motion and lower fatigue. On the other hand, the lower limbs have low mobility, stability and symmetry, which means that the lower limbs have weak strength, fatigued muscles and high activity risk, and the weak parts need to be exercised and improved.

To sum up, the body health assessment method based on joint point recognition according to the embodiment of the present invention uses technologies such as visual recognition, big data, edge calculation, cloud calculation, and database, and uses sports measurement, sports training, functional anatomy, and sports rehabilitation as professional bases, and automatically evaluates the "sports fitness" level of the body (including activity degree of main joints, motion stability, body symmetry, and risk of sports injury of main body links) of the subject by completing the quality of standard motions for the subject, and can automatically generate a set of solutions of a targeted training plan according to the evaluation result.

The above description is only for the specific embodiment of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention are included in the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A physical health assessment method based on joint point identification is characterized by comprising the following steps:

acquiring a human body image of a tester completing a preset test action, which is acquired by a depth camera;

calculating joint angle data of the body of the tester according to the space coordinate data, wherein the joint angle data comprise an angle, an angular velocity and an angular acceleration;

2. The method of claim 1, wherein the subject's body joint marker points comprise at least 30 joint marker points, each being: head, neck, upper spine, middle spine, bottom spine, left/right shoulders, left/right elbows, left/right wrists, left/right hands, left/right hips, left/right knees, left/right ankles, left/right feet, left/right fingertips, left/right thumbs, nose, left/right eyes, left/right ears.

3. The method of claim 1, wherein the joint angle data comprises data for at least 19 joint angles, the joint angles comprising: neck, mid-spine, left/right upper spine, mid-spine, left/right bottom spine, left/right shoulder, left/right elbow, left/right wrist, left/right hip, left/right knee, left/right ankle.

4. The method of claim 1, wherein the evaluating joint activity, motion completion quality, and body consistency level of completed motion from the spatial coordinate data and joint angle data, respectively, and calculating a joint activity score, a motion stability score, and a body symmetry score comprises:

5. The method of claim 1, wherein the evaluating joint activity, motion completion quality, and body consistency level of completed motion from the spatial coordinate data and joint angle data, respectively, and calculating a joint activity score, a motion stability score, and a body symmetry score comprises:

establishing an initial condition probability table according to the test data;

and calculating the test data by adopting a PCMHS algorithm to obtain a motion stability score.

6. The method of claim 1, wherein said evaluating joint activity, motion completion quality, and body consistency level of completed motion from said spatial coordinate data and joint angle data, respectively, and calculating a joint activity score, a motion stability score, and a body symmetry score comprises:

establishing an initial condition probability table according to the test data;

7. The method of claim 1, wherein the evaluating the joint activity, the motion completion quality, and the body consistency level of the completed motion from the spatial coordinate data and the joint angle data, respectively, and calculating a joint activity score, a motion stability score, and a body symmetry score, thereafter comprises:

8. The method of claim 7, wherein the evaluating a risk of athletic damage based on the joint activity score, the motion stability score, and the body symmetry score, and wherein calculating the athletic damage risk score comprises:

9. The method of claim 7, wherein the assessing a risk of athletic injury based on the joint motion score, the motion stability score, and the body symmetry score, and wherein calculating the risk of athletic injury score comprises:

10. The method of claim 9, wherein the evaluating health of the person based on the joint activity score, the motion stability score, the body symmetry score, and the athletic injury risk score comprises: