JP2007061121A - Body motion analysis method, system and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a body motion analysis method, system and program capable of further improving analysis accuracy.
Three-dimensional motion measurement is performed on a marker J _i attached to a predetermined joint portion of the body, and the markers are connected by the position (x _i , y _i , z _i ) of the marker J _i with respect to a reference point J ₀ . Analyzing link behavior. Here, in advance by measuring the length in the stationary state of each link, the link reference length L _i. Then, each link length during body movement is measured as an actually measured value l _i . By comparing the link reference length L _i thus corresponding to the measured actual values l _i, and calculates the error [delta] L _i. Then, based on the calculated error δL _i , each measured marker position is corrected by the first correction so that the actual measurement value l _i is within a predetermined error range with respect to the link reference length L _i . .
[Selection] Figure 5

Description

  The present invention uses a multi-link model in which a predetermined part of the body is used as a reference point, a plurality of joint portions are marked with markers, and the links connecting the markers are connected in order. The present invention relates to a body motion analysis method, system, and program by three-dimensional motion measurement for analyzing motion. In particular, with the center of the head as a reference point, the club head of the golf club held by the neck, shoulder, elbow, wrist and hand is marked with a marker, respectively, and the neck-shoulder, shoulder-elbow, elbow-wrist and wrist- The present invention relates to a golf swing motion analysis method based on the DLT method for analyzing the motion of each link with respect to the reference point using a four-link model in which the links connecting the club heads are sequentially connected.

  In recent years, as the performance of electronic devices has improved dramatically and the measurement technology has advanced, research on physical movement has been rapidly developed. For example, in addition to ergonomics and biomechanics, a field called sports engineering has been established and researched. Sports engineering aims to improve the skills of sports and prevent obstacles arising from physical burdens in sports by developing tools that are indispensable for sports and elucidating the physical movements in engineering.

  In research on such body movements, research using a multi-link model in which a plurality of joint portions are marked with markers and each of the links connecting the markers is sequentially connected is known (for example, see Patent Document 1). ).

Many of these motion measurements are measured by a three-dimensional motion analysis method called a DLT (Direct Linear Transformation) method and a three-dimensional motion measurement apparatus using the same.
JP 2004-344418 A

  However, the body motion analysis method using the conventional multi-link model has the following problems. That is, since the marker to be measured by the three-dimensional motion measuring device is attached on the skin or clothes, it does not always match the actual joint position (movable point), and the marker position and the joint position are shifted due to the movement of the body. As a result, a measurement error may occur. In addition, if the marker is made small, it may cause observation failure and the like, and thus the marker needs to have a certain size. However, there is a case where such an error due to the size of the marker itself cannot be ignored.

  In order to reduce such an error, measures such as calculating an average value for a plurality of measurement results are also taken, but sufficient accuracy has not been secured.

  The present invention has been made to solve the problems of the prior art, and an object of the present invention is to provide a body motion analysis method, system, and program capable of further improving analysis accuracy.

  The body motion analysis method according to the present invention uses a multi-link model in which a plurality of joint portions are marked with a marker, and a plurality of links connecting the markers are sequentially connected with a predetermined point of the body as a reference point. A body motion analysis method by three-dimensional motion measurement for analyzing the motion of each link with respect to the reference point, wherein the link length measured in a stationary state is a link reference length, and an actual measurement value measured by three-dimensional motion measurement On the other hand, an error between the measured value and the link reference length is calculated, and the measured marker position is calculated for the measured marker position so that the measured value is within a predetermined error range with respect to the link reference length based on the error. 1 correction is performed.

  In the above method, a marker attached to a predetermined joint portion of the body is measured in a three-dimensional motion, and the motion of the link connecting the markers is analyzed from the position of the marker with respect to the reference point. Here, the length of each link in a stationary state is measured and used as the link reference length. And each link length at the time of a body movement is measured as an actual measurement value. The measured value thus measured is compared with the corresponding link reference length, and the error is calculated. Based on the calculated error, each measured marker position is corrected by the first correction so that the actually measured value is within a predetermined error range with respect to the link reference length.

  Thus, based on the fact that the length between the joints, that is, the length of the link between the markers is constant, each measured value of the measured link length approximately matches the link reference length. By correcting the marker position, the original position of each joint can be estimated with high accuracy, so that the body motion can be analyzed with high accuracy.

  Preferably, the first correction is performed by adding / subtracting 1 / W (where W is a real number of 2 or more) of the error to both ends of the link measured with respect to the marker positions at both ends of the link so that the error is reduced. It is determined whether or not the length of the corrected link is within the predetermined error range, and if it is not within the error range, the length of the corrected link and the link reference length are determined. After calculating the error, a first correction is performed on the corrected link length so that the error is reduced again until the corrected link length falls within the predetermined error range. The addition / subtraction and determination are repeated.

  In this case, each of the marker positions at both ends of the link is added or subtracted so that the error is reduced by 1 / W (for example, 1/2) of the error between the actually measured link length and the link reference length. Is corrected to be close to the link reference length. Then, it is determined whether or not the added or subtracted link length is within a predetermined error range. If the error is not within the error range, an error from the link reference length is calculated again for the corrected link length, and the first correction is repeated until it falls within the predetermined error range.

  Therefore, by repeating the first correction, the actually measured value of the measured link length can be more reliably corrected to a length that substantially matches the link reference length.

  Preferably, after the first correction, the second correction is performed on each measured marker position so that the operation of each marker after the first correction is smooth with respect to time.

  In this case, in addition to the first correction, the second correction is performed so as to satisfy the condition that the movement of the body moves smoothly in time series.

  Accordingly, by performing the second correction, it is possible to prevent the first correction from distorting the body motion trajectory (marker trajectory) and to analyze the body motion closer to the actual motion. it can.

  Preferably, in the second correction, the coordinate system of each marker is transformed into a spherical polar coordinate system, and then corrected so as to follow a predetermined higher-order equation.

  In this case, by converting the coordinate system of each marker into a spherical polar coordinate system, it is possible to prevent the condition of the constant link length in the first correction from being broken, and to correct it so as to follow a predetermined higher-order equation ( Perform curve fitting).

  Thereby, it is possible to obtain a link model that more closely resembles actual body movements while maximizing the effect of the first correction.

  Preferably, the three-dimensional motion measurement is performed by a DLT method.

  In this case, the measurement accuracy can be maintained high by using a general and established DLT (Direct Linear Transformation) method as a three-dimensional motion measurement method.

  In addition, the golf swing motion analysis method according to the present invention uses the center of the head as a reference point, and marks the neck, shoulder, elbow, wrist, and club head of the golf club held by the hand with markers, Golf swing motion analysis using the DLT method, which analyzes the motion of each link with respect to the reference point, using a four-link model in which shoulder, shoulder-elbow, elbow-wrist and wrist-club head links are connected in order. The link length measured in a stationary state is a link reference length, and an error between the actual measurement value and the link reference length is calculated with respect to the actual measurement value measured by three-dimensional motion measurement. Based on this, the first correction is performed on each measured marker position so that the actually measured value is within a predetermined error range with respect to the link reference length, and after the first correction, the first correction is performed. As the operation of each marker after correction becomes smooth with respect to time, it is characterized in performing a second correction for each marker position measured.

  Further, the body motion analysis system according to the present invention uses a multi-link model in which a plurality of joint portions are marked with markers using a predetermined point of the body as a reference point, and the links connecting the markers are sequentially connected. A body motion analysis system based on three-dimensional motion measurement for analyzing the motion of each link with respect to the reference point, defining a space as absolute coordinates and measuring the three-dimensional motion, and a stationary state Storage means for storing the link reference length, which is the length of the link measured in step 1, and the coordinates of each of the markers, and an actual measurement value of the length of each link by measurement is calculated from the coordinates of the markers, and the link reference An error calculation means for calculating an error between the actual measurement value and the link reference length in comparison with each of the lengths, and the actual measurement value based on the error is converted into the link reference length. First correction means for performing a first correction on each measured marker position so that it falls within a predetermined error range, and the operation of each marker after the first correction after the first correction. Is provided with second correction means for performing second correction on each measured marker position so that the time becomes smooth with respect to time.

  Furthermore, the body motion analysis program according to the present invention uses a multilink model in which a plurality of joint portions are marked with a predetermined point as a reference point, and a plurality of links connecting the markers are connected in order. A body motion analysis program based on three-dimensional motion measurement for analyzing the motion of each link with respect to the reference point, which is measured by a three-dimensional motion measurement device that defines a space as absolute coordinates and performs three-dimensional motion measurement. Storage means for storing the link reference length, which is the length of the link in a stationary state, and the coordinates of each of the markers, calculating an actual measurement value of each link length by measurement from the coordinates of the markers, and the link reference length Error calculation means for calculating an error between the actual measurement value and the link reference length, and the actual measurement value is calculated based on the error. First correction means for performing a first correction on each measured marker position so as to be within a predetermined error range with respect to the reference length, and after the first correction, after the first correction The marker functions as a second correction unit that performs a second correction on each measured marker position so that the operation of each marker is smooth with respect to time.

  According to the body motion analysis method, system, and program of the present invention, the measured link length is measured based on the fact that the length between the joints, that is, the length of the link between the markers is constant. By performing the first correction for each marker position so that the value substantially matches the link reference length, the original position of each joint can be estimated with high accuracy, so that the body motion can be analyzed with high accuracy. can do.

  Furthermore, by performing the second correction, it is possible to prevent the first correction from distorting the body motion trajectory (marker trajectory), and to perform a body motion analysis closer to the actual motion. it can.

The best mode for carrying out the present invention will be described below with reference to the accompanying drawings.
FIG. 1 is a diagram schematically showing a link model used in a body motion analysis system according to an embodiment of the present invention. FIG. 2 is a diagram showing a body motion analysis system in the present embodiment. In the present embodiment, the description will be based on the case where the motion analysis of the golf swing is applied. However, the present invention is not limited to this, and various motion analysis of the body can be applied without particularly changing the system. In addition, changes can be made without departing from the spirit of the present invention so that an optimum system is obtained according to each operation.

As shown in FIG. 1, the body motion analysis system according to the present embodiment uses a predetermined location (head) of the body of the subject M as a reference point J ₀ and a plurality of joint portions as markers J _i (i = 0, 1, 2...), And using the multi-link model in which the links L _i connecting the markers J _i are connected in order, the operation of each link L _i with respect to the reference point J ₀ is a three-dimensional operation. Analyze by measurement. More specifically, as shown in FIG. 1 in this embodiment, the subject M's neck (marker J ₁ ), shoulder (marker J ₂ ), elbow (marker J ₃ ), wrist (marker J ₄ ), and hand the gripped golf club of the club head (marker J ₅₎ marked by each marker, neck - shoulder (link L _1), the shoulder - elbow (link L _2), elbow - wrist (link L ₃₎ and wrist - Club Using a four-link model in which each of the links connecting the heads (link L ₄ ) is connected in order, the operation of each link L _i (i = 1, 2, 3, 4) with respect to the reference point J ₀ is a three-dimensional operation. Analyze by measurement. The marker J _i may be marked from the top of the garment, but is preferably marked from the top of the garment that is in close contact with the body, and more preferably directly from the garment, in order to reduce the generation error as much as possible. Size of the marker J _i, in order to reduce the possible measurement error, to the extent that can recognize camera 11 marker J _i which will be described later, it is preferable that as small as possible.

In order to analyze such a four-link model, the body motion analysis system, as shown in FIG. 2, defines a space S as absolute coordinates and performs a three-dimensional motion measurement, and a stationary each coordinate of a length of the link reference length L _i and the marker J _i of measured link L _i in the state _{(x i, y i, z} i) and the storage means 21 for storing said marker J _i The actual measured value l _i of the length of each link by measurement is calculated from the coordinates (x _i , y _i , z _i ) of the link, and compared with each of the link reference lengths L _i , the measured value l _i and the link are calculated. an error calculation means 22 for calculating an error [delta] L _i of the reference length L _i, so that the measured values l _i based on the error [delta] L _i is within a predetermined error range with respect to the link reference length L _i, First correction means 23 for performing a first correction on each measured marker position; Second correction means 24 for performing a second correction on each measured marker position so that the operation of each marker J _i after the first correction becomes smooth with respect to time t _m after the first correction; It is characterized by comprising.

In the present embodiment, as shown in FIG. 2, the three-dimensional motion measurement apparatus 1 is provided with two cameras (high-speed cameras) 11, and from the three-dimensional motion measurement means 25 of the computer 2 such as a personal computer. The measurement space S is controlled so that the two cameras 11 can be synchronized. The computer 2 has external storage means 26 such as a hard disk in which a body motion analysis program is stored, and causes the CPU 3 of the computer 2 to function as the three-dimensional motion measurement means 25. Note that the arrangement of two cameras 11, depending on the operation of various body, (position of the marker J _i) the measured joint position is placed optimized so as not to possible hidden.

  In the present embodiment, as shown in FIG. 2, the measurement space S has the golf club head impact position as the origin O, the flying ball line direction is the X axis, the direction perpendicular to the ground is the Y axis and the ground is horizontal and flying. It defines a coordinate space with the Z axis as the direction perpendicular to the spherical line (depth direction).

  In particular, since a golf swing is close to a state of swinging on a certain plane (referred to as swing plane P), it can be assumed that the golf swing is swinging on the swing plane P. Therefore, the analysis is performed by converting the three-dimensional coordinates of each marker calculated from the images photographed by the two cameras 11 into the two-dimensional coordinates on the swing plane P. FIG. 3 is a schematic diagram showing coordinate conversion to the swing plane.

As shown in FIGS. 2 and 3, the swing plane P is inclined by an angle α with respect to the horizontal plane. The inclination angle α of the swing plane P is determined by the least square method from the y coordinate and z coordinate values of the measurement point (marker J ₅ ) of the club head. The three-dimensional coordinates in the measurement space S of each marker J _i attached to the subject M are converted into two-dimensional coordinates on the swing plane P using the inclination angle α of the swing plane P determined in this way.

The measured three-dimensional coordinates of the marker J _i of the subject M are (x _i , y _i , z _i ), and the two-dimensional coordinates on the swing plane P after coordinate conversion are (X _i , Y _i ). The conversion equation is shown by the following equation. Here, β is an angle formed by the swing plane P and the straight line including the origin O and the marker J _i .

The angle, angular velocity, angular acceleration, and joint torque of each joint are calculated based on the position coordinates of each marker J _i subjected to the coordinate conversion as described above, and the motion analysis is performed.

In the present embodiment, a DLT (Direct Linear Transformation) method is used as a three-dimensional motion measurement method for calculating three-dimensional coordinates for each marker J _i from images taken by two cameras 11. In general, when performing three-dimensional motion measurement, for example, the distance from the camera to the subject M, the installation angle of the two cameras 11, the focal length of the camera lens, and the like need to be determined in advance. It is often difficult to obtain a numerical value accurately. On the other hand, in the DLT method, a point having a known three-dimensional coordinate called a control point is photographed together with the subject M, and the numerical value can be calculated backward from the image. Therefore, it is possible to easily calculate the three-dimensional coordinates of the marker J _i attached to the subject using the obtained numerical values without directly measuring the numerical values. Therefore, the measurement accuracy can be kept high by using the DLT method which is generally established as a method for measuring the three-dimensional motion.

In the present embodiment, the three-dimensional motion measuring means 25 calculates three-dimensional coordinates for each marker J _i from the images taken by the two cameras 11 using the DLT method. As shown in FIG. 1, the computer 2 has storage means 21 that is a memory for temporarily storing data and the like, and the calculated three-dimensional coordinates are stored in the storage means 21 and / or the external storage means 26.

  The computer 2 also has an input means 27 such as a mouse and a keyboard for the measurement person to make various settings / inputs, a monitor for displaying the results of the operation measurement / analysis, and an output means 28 such as a printer. The storage means 21, the external storage means 26, the input means 27, and the output means 28 are configured to be able to communicate with each other via a signal bus 29. Further, the CPU 3 of the computer 2 functions as an error calculation unit 22, a first correction unit 23, and a second correction unit 24, which will be described in detail below, according to a body motion analysis program.

  As described above, in the present embodiment, a single computer 2 can perform three-dimensional motion measurement and motion analysis (including the first and second corrections), but the existing three-dimensional motion measurement. Another apparatus may be connected to the apparatus / motion analysis system from the outside, and only the first and second correction processes may be performed in the computer.

Here, the measurement error correction process in the present embodiment will be described. As the correction method according to the present invention, each joint and between the length of the golf club, because it should be constant regardless of the time during operation, the time length of the link L _i between each marker J _i The first purpose is to make corrections so that they are constant. In addition, since the body movement of the subject M including the human is smooth and the joint position should move smoothly, the change in the position coordinate of each marker J _i is corrected so as to be smooth with respect to the time t _m. The second purpose is to do. Hereinafter, it demonstrates in order.

  FIG. 4 is a flowchart of the measurement error correction process in this embodiment.

First, the length of each link _{Li in} a stationary state is measured, and is read and stored as the link reference length _Li in the external storage means 26 or the storage means 21 (step S1). The link reference length L _i may be measured using the two cameras 11, or a value actually measured with a measure or the like may be input from the input unit 27. The reference point J ₀ provided on the head of the subject M as the starting point of the link L _i and the reference link L ₀ between the reference point J ₀ and the neck marker J ₁ are similarly processed. As a result, the reference link L ₀ and the links L ₁ , L ₂ ,..., L ₄ can be handled as a link model that is sequentially connected starting from the reference point J ₀ . Therefore, the coordinates of the markers J ₁ , J ₂ ,..., J ₄ are determined in order with the reference point J ₀ as a reference. The reference point J ₀ does not need to be provided on the head, but is preferably provided at a position where the coordinates do not change as much as possible during the operation. By treating the reference point J ₀ as a fixed point, each coordinate of each marker J _i can be determined more easily and reliably. Further, when the reference point J ₀ is handled as a fixed point, it is preferable that the average value of the coordinates of the reference point J ₀ for each time be the coordinate of the reference point J ₀ .

Next, the CPU 3 of the computer 2 uses the three-dimensional coordinates (x _i , y _i , z _i ), (x _{i + 1} , y _{i +} ) of the markers J _i , J _{i + 1} measured by the three-dimensional motion measuring means 25. ₁ , z _{i + 1} ) is read (step S2).

Then, the CPU 3 of the computer 2 calculates an actual measurement value l _i of the link length L _i of the measurement from the read three-dimensional coordinates (step S3). An actual measurement value l _i of the measured link length at the sampling time t _m is expressed by the following equation.

Subsequently, CPU 3 functions as an error calculating unit 22 calculates the error [delta] L _i = L _i -l _i is compared with the measured values l _i and link reference length L _i by the measurement (Step S4) .

The CPU 3 determines whether or not the calculated error δL _i is within a predetermined error range (step S5). In this embodiment, as shown in FIG. 4, an error tolerance of 0.01 mm or less is set as an error tolerance (0.01 mm is set as the error reference value), and a determination is made based on this. To do. That is, when the absolute value of the error δL _i is equal to or smaller than the error reference value 0.01 mm (No in step S5), the processing is continued without performing the first correction described later. On the other hand, when the absolute value of the error δL _i is larger than the error reference value 0.01 mm (Yes in step S5), the CPU 3 functions as the first correction unit 23, and the link actual measurement value l _i is the link reference value. Based on the calculated error δL _i so as to be within the above error range with respect to the length L _i , the first position coordinates of the markers J _i and J _{i + 1} at both ends of the link L _i are first. Is corrected (step S6).

For the markers J _i and J _{i + 1} that have undergone the first correction in step S6, the error δL _i at the corrected marker position is calculated again in steps S3 and S4, and the error is determined in step S5. Thus, the first correction processing (steps S3 to S6) is repeated until the calculated error δL _i falls within a predetermined error range (until No in step S5). That is, the measured positions of the markers J _i and J _{i + 1} are corrected by the first correction so that the actually measured link value l _i is within a predetermined error range with respect to the link reference length L _i . This first correction is performed every sampling time t _m .

  Here, the first correction in the present embodiment will be described more specifically. FIG. 5 is a diagram schematically showing a first correction method in the present embodiment.

In the present embodiment, the first correction is performed by setting the positions of the markers J _i and J _{i + 1} at both ends of the link L _i by ½ of the error δL _{i at} both ends of the measured actual value l _i of the link. Correction to the added positions J ′ _i and J ′ _{i + 1} (δL _i is positive (means addition) when L _i > l _i , and negative (means subtraction) when l _i > L _i. )).

That is, based on the coordinates of each marker J _{i (i} = 1,2,3,4) link coupled to the marker J _i L _i, the error [delta] L _i in L _i-1, the [delta] L _i-1, each Each coordinate component is corrected as follows.

However, the marker J ₅ in the club head, not a marker between links, based on only one of the link L _4, is corrected as follows.

  In this case, each of the marker positions at both ends of the link is added by ½ of the error between the actual measured link length and the link reference length, and corrected so that the link length is substantially equal to the link reference length. Is done.

Therefore, the actually measured value l _i of the measured link length can be made to substantially coincide with the link reference length L _i by the first correction.

Thus, the length of the joint, i.e., the marker J link length of L _i between _i is based on a constant, the length of the measurement link measured value l _i link reference length L as substantially coincides with the _i position of each marker _{J i (x i, y i} , z i) by correcting the, since it is possible to estimate the original position of each joint with high accuracy, the motion of the body Analysis can be performed with high accuracy.

If the absolute value of the measurement error [delta] L _i for the link L _i is equal to or less than the error reference value 0.01mm of the reference value 0.01mm or less is the case and the first correction of the error (No in step S5), CPU 3 Functions as the second correction means 24 and performs the second correction for each measured marker position so that the operation of each marker J _i within the error range is smooth with respect to the time t _m . In this case, the second correction is performed so as to satisfy the condition that the movement of the body moves smoothly in time series. That is, in the first correction process, correction is performed independently at each sampling time t _m , so that the three-dimensional coordinates of each marker J _i are not continuous in time series. Therefore, the coordinates of each marker J _i are made continuous with respect to time, and the link length does not change at that time (so that the effect of the first correction is not diminished). ,to correct.

In the present embodiment, in the second correction, the coordinate system of each marker J _i is coordinate-transformed into a spherical polar coordinate system, and then corrected so as to follow a predetermined higher-order equation. In this case, the coordinate system of each marker J _i is converted into a spherical polar coordinate system, thereby preventing the condition of the constant link length in the first correction from being broken and correcting so as to follow a predetermined higher-order equation. (Curve fitting is performed). FIG. 6 is a flowchart of the second correction process in this embodiment. FIG. 7 is a diagram showing a spherical polar coordinate system used for the second correction processing in the present embodiment.

ただし、０≦θ_i≦π，０≦φ_i≦２πである。 First, as shown in FIG. 7, each link L _i is converted into a spherical polar coordinate system with the origin of the coordinate system of the marker J _i at each link end (ie, (x _i , y _i , z _i ) → (r _i , θ _i , φ _i )) (step S71). In this case, the coordinates (r _i , θ _i , φ _i ) of the spherical polar coordinate system can be expressed as follows using the original coordinate system (x _i , y _i , z _i ).

However, 0 ≦ θ _i ≦ π and 0 ≦ φ _i ≦ 2π.

Accordingly, at this time, the r _i component in the spherical polar coordinate system becomes the length of the link L _i , and therefore, if the remaining θ _i and φ _i components are corrected so as to be continuous with respect to time, the length of the link L _i Can be corrected without changing (without diminishing the effect of the first correction).

As for the time series change of the marker J _i converted into the spherical polar coordinate system in this way, a predetermined higher order equation is input from the input means 27 or a predetermined higher order equation stored in advance in the external storage means 26 is selected. As a result, the CPU 3 reads the higher order equation (step S72).

A high-order equation can be expressed as follows in the case of n-th order.

  The coefficients of such higher order equations are obtained using the least square method. The order n may be experimentally set in advance or may be automatically determined by the CPU.

The CPU 3 performs approximation (curve fitting) on the temporal change of the θ _i component and the φ _i component using the higher-order equation (step S73). For curve fitting in steps S72 and S73, specifically, an existing curve fitting program can be employed. After the approximation, the original three-dimensional Cartesian coordinate system is reconverted for the subsequent motion analysis process (step S74), and the second correction process is terminated.

  By performing the second correction as described above, the first correction prevents the body movement trajectory (marker trajectory) from being distorted and maximizes the effect of the first correction. However, since the correction can be made so that the link model is closer to the actual motion, the body motion analysis closer to the actual motion can be performed.

In the first correction in the present embodiment, the positions of the markers J _i and J _{i + 1} at both ends of the link L _i are set to 1/2 of the error δL _{i at} both ends of the measured actual value l _i of the link. However, the position is not limited to ½, and may be corrected to a position obtained by adding 1 / W of error δLi (W is a real number of 2 or more). It is also possible to reduce the number of corrections until convergence within a predetermined error range by adding a value smaller than ½ as the correction value to reduce the correction width and performing finer correction.

Further, each marker end portion is set such that the ratio of the correction value at one end J _i of the marker of the link Li to the correction value at the other end J _{i + 1} of the marker is U: 1−U (0 <U <1). The positions of J _i and J _{i + 1} may be corrected.

Further, when there is a fixed point that is not related to the time change in the marker J _i , the other end part is set so that the link length is constant in order from the link having the fixed point as an end part with the fixed point as a reference. It is good also as correcting by determining a marker position and determining the marker position used as the other edge part of an adjacent link in order after that.

Further, a configuration in which only the first correction is performed and the second correction is not performed may be employed. By not performing the second correction, the calculation speed for correction can be increased. However, as the made The more the number of links L _i, since an error is likely to occur, it is preferred to use a second correction.

  Here, in the body motion analysis system according to the present invention, the accuracy evaluation of the four-link model in the above embodiment will be described as an example in comparison with a comparative example.

<Accuracy evaluation method>
Assuming an ideal swing motion that does not include errors in advance, errors assumed in Monte Carlo simulation were randomly generated and inserted into an ideal swing model, and many motion analyzes were performed. Next, evaluation was made by comparing the difference in accuracy by the correction method with the analysis result of the ideal swing motion. The number of trials in the Monte Carlo simulation was 1000.

<Evaluation method by angle>
When the correction (first and second correction) processing of the present invention is performed (example), only when the second correction processing is performed (reference example), and when there is no correction (comparative example), Monte Carlo -Using simulation, we investigated the effect of errors on joint angles, and obtained and compared and evaluated probability density functions for each.

<Evaluation method by joint torque>
In order to investigate the influence on the joint torque, a comparison was made using the value at the maximum value of each joint torque. The ratio between the maximum value (τ _{o, max} ) of each joint model in the ideal model and the maximum value (τ _{e, max} ) of each joint model in the error model was defined by the following equation and analyzed using Monte Carlo simulation. A probability density function was obtained from the variation at this time and evaluated.

<Evaluation result by angle>
FIG. 8 is a diagram showing each probability density distribution of angular errors in the shoulder joint. FIG. 9 is a diagram showing each probability density distribution of the angle error at the wrist joint.

  When there is no correction, the influence of the error is received as it is, and the variation of the error is large. On the other hand, when the correction according to the present invention is performed, the probability that the value is close to the ideal value is higher than that in the case where the correction is not performed, and the variation is very small. Compared to the case of only the second correction, a result in which the probability near the ideal value is higher and the variation is smaller is obtained. Therefore, the accuracy can be further improved by performing the first correction. Similarly, in FIG. 9, the accuracy can be further improved by performing the correction of the present invention.

<Evaluation results by joint torque>
FIG. 10 is a diagram showing a probability density distribution related to a ratio of maximum values of joint torque in the wrist joint.

  Compared to the case of only the second correction, the joint torque maximum value ratio rτ has a larger probability distribution close to 1 and the variation becomes considerably smaller when the first correction and the second correction are performed. became. Therefore, the accuracy can be further improved by performing the first correction for the joint torque.

  The body motion analysis method, system, and program according to the present invention improve the skill of sports and improve sports skills by developing empirically and sensoryly discussed tools for human body movements and engineering the body movements. It can be used to prevent disability caused by physical burden. For example, golf club motion analysis, tennis swing motion analysis, throwing form analysis, or the like can be performed to develop golf clubs or tennis rackets, or to obtain suggestions for improving swings or forms. It can also be used for physical therapy such as rehabilitation by analyzing walking. Furthermore, it can be used not only for humans but also for motion analysis of animals and the like.

It is a figure showing roughly a link model used for a body movement analysis system concerning one embodiment of the present invention. It is a figure which shows the body movement analysis system in this embodiment. It is the schematic which shows the coordinate transformation to a swing plane. It is a flowchart about the measurement error correction process in this embodiment. It is a figure which shows schematically the 1st correction method in this embodiment. It is a flowchart about the 2nd correction process in this embodiment. It is a figure which shows the spherical polar coordinate system used for the 2nd correction process in this embodiment. It is a figure which shows each probability density distribution of the angle error in a neck joint. It is a figure which shows each probability density distribution of the angle error in a shoulder joint. It is a figure which shows the probability density distribution regarding the ratio of the maximum value of the joint torque in a wrist joint.

Explanation of symbols

1 three-dimensional operation measuring device 2 computer 21 storage unit 22 the error calculating unit 23 the correction means 24 correction means J ₀ reference point J _i marker L _i link l _i actual value S measurement space length of the link t _m sampling time [delta] L _i error

Claims (8)

Using a multi-link model in which a predetermined part of the body is used as a reference point, a plurality of joints are marked with markers, and the links connecting the markers are sequentially connected, the movement of each link with respect to the reference point is analyzed. A body motion analysis method based on 3D motion measurement,
The link length measured in a stationary state is defined as a link reference length, and an error between the actual measurement value and the link reference length is calculated for the actual measurement value measured by three-dimensional motion measurement.
A body motion analysis method characterized in that first correction is performed on each measured marker position so that an actual measurement value falls within a predetermined error range with respect to a link reference length based on the error. The first correction is performed by adding / subtracting 1 / W (where W is a real number of 2 or more) of the error to both ends of the link measured with respect to the marker positions at both ends of the link so that the error is reduced. Is what
It is determined whether or not the corrected link length is within the predetermined error range. If the corrected link length is not within the predetermined error range, the error is calculated after calculating the error between the corrected link length and the link reference length. The first correction is performed again so that the error is reduced again with respect to the corrected link length, and the addition / subtraction and determination are repeated until the corrected link length falls within the predetermined error range. The body motion analysis method according to claim 1, wherein:   The second correction is performed on each measured marker position after the first correction so that the operation of each marker after the first correction is smooth with respect to time. 3. The body motion analysis method according to 2.   4. The body motion analysis method according to claim 3, wherein the second correction is performed by converting the coordinate system of each marker into a spherical polar coordinate system, and then correcting the coordinate system along a predetermined higher order equation.   The body motion analysis method according to claim 1, wherein the three-dimensional motion measurement is performed by a DLT method. Mark the club head of the golf club gripped by the neck, shoulder, elbow, wrist and hand with the center of the head as a reference point, respectively, and neck-shoulder, shoulder-elbow, elbow-wrist and wrist-club head A golf swing motion analysis method based on the DLT method for analyzing the motion of each link with respect to the reference point using a four-link model in which each of the links connected to each other is sequentially connected,
The link length measured in a stationary state is defined as a link reference length, and an error between the actual measurement value and the link reference length is calculated for the actual measurement value measured by three-dimensional motion measurement.
Based on the error, a first correction is performed on each measured marker position so that the actual measurement value is within a predetermined error range with respect to the link reference length,
After the first correction, the second correction is performed on each measured marker position so that the operation of each marker after the first correction is smooth with respect to time. analysis method. Using a multi-link model in which a predetermined part of the body is used as a reference point, a plurality of joints are marked with markers, and the links connecting the markers are sequentially connected, the movement of each link with respect to the reference point is analyzed. A body motion analysis system based on 3D motion measurement,
A three-dimensional motion measurement device that defines a space as absolute coordinates and performs three-dimensional motion measurement;
Storage means for storing a link reference length, which is a link length measured in a stationary state, and coordinates of each of the markers;
Calculating an actual measurement value of the length of each link by measurement from the coordinates of the marker, comparing with each of the link reference length, an error calculating means for calculating an error between the actual measurement value and the link reference length;
First correction means for performing first correction on each measured marker position so that the actual measurement value is within a predetermined error range with respect to the link reference length based on the error;
After the first correction, there is provided second correction means for performing a second correction on each measured marker position so that the operation of each marker after the first correction is smooth with respect to time. Body motion analysis system characterized by Using a multi-link model in which a predetermined part of the body is used as a reference point, a plurality of joints are marked with markers, and the links connecting the markers are sequentially connected, the movement of each link with respect to the reference point is analyzed. A body motion analysis program based on 3D motion measurement,
Computer
Storage means for defining a space as an absolute coordinate and storing a link reference length that is a length of a link in a stationary state measured in a three-dimensional motion measurement apparatus that performs three-dimensional motion measurement and each coordinate of the marker;
An error calculation unit that calculates an actual measurement value of each link length by measurement from the coordinates of the marker, and calculates an error between the actual measurement value and the link reference length in comparison with each of the link reference lengths;
First correction means for performing a first correction on each measured marker position so that the actual measurement value is within a predetermined error range with respect to the link reference length based on the error, and the first correction Thereafter, the movement of the body is made to function as second correction means for performing second correction on each measured marker position so that the movement of each marker after the first correction is smooth with respect to time. Analysis program.

Images