JPS6338478A - Apparatus for analyzing motion operation - Google Patents

Abstract

(57) [Abstract] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION (19) Object of the Invention (Field of Industrial Application) The present invention relates to an apparatus for analyzing human motor movements. More specifically, the present invention relates to a device suitable for use in analyzing human athletic movements, such as golf swings.

(Prior art) There are many conventional golf practice devices that use specific equipment to regulate the movement of the hands and feet to create the ideal swing form for practice players. (Japanese Unexamined Patent Publications No. 60-34474, No. 61-16754), the swing itself was focused on human movement, and a certain form, motion,
The problem is that there are few types of rhythms, and many people are unable to improve or escape slumps because they are unable to overcome slumps and deficiencies based on human movement characteristics such as form, motion, and rhythm. By the way, is the above-mentioned practice device that restrains the body etc. a golf ball? - The device involved is complicated, and there is also the problem that when the device is removed, it gives a strange feeling during the actual play and the feeling of the swing is different, and the results of practice are not fully utilized. There is.

In general, the conditions for hitting a golf ball correctly include having the right grip, taking the right stance, keeping your eye on the ball, always performing the swing motion with a certain rhythm, and There are various conditions such as not moving up and down. Among these conditions, grip and stance can be seen concretely through one's own eyes, so one can confirm the suitability of the grip and stance based on knowledge gained from other people's comments and instruction manuals. You can easily admit your mistakes. However, golf after entering the backswing? −
Conditions based on human motor movements and sensory elements, such as the movement of a golfer or the rhythm of a swing, are not something that can be easily learned through guidance from others, and it is not possible to move the body in accordance with the instructions. In the end, there is no other way than to become aware of it through your own efforts and practice.

The conditions for a golf swing to be performed correctly are a series of movements from address, back ink, downswing, impact, follow-through, and finish, with the correct rhythm and timing. It is an extremely small thing to be carried out so that it passes along a certain orbit. In particular, the twisting of the body from address to the top of the backswing, the shift of weight, the tempo until this is completed, the pause at the bottom of the backswing, the timing of starting the downswing, and the follow-through are uneven and uneven. It can be said that a correct swing can be obtained if the body movements, tempo, and timing during this period are smooth and can be performed at a constant rate.

Therefore, attempts have been made to develop a training device that analyzes the human VJ movement of a series of swings from take-back to impact and follow-through, determines its merits, and corrects it to a standard swing form. For example, as a swing diagnostic device, as shown in FIG. 9, an input device 10 with a built-in variable resistor attached to the golfer's body.
1 and a golf club 102 are connected by a 7-m 103 to detect a swing trajectory (Japanese Patent Application Laid-Open No. 59-225, 0
No. 82) has been proposed. This swing diagnostic device is
The movement of the golf club 102 is transmitted to the human power device 101 via the rigid arm 103, the movement is converted into a voltage change and extracted, and after computer image processing, the trajectory of the swing is expressed as a line drawing. It diagnoses the swing by comparing it with the normal swing trajectory.

Additionally, a device is being considered that uses a standard television camera and video equipment to simultaneously check the golfer's swing from the front, overhead, and back of the golfer (Japanese Unexamined Patent Publication No. 61
-11,070@). This device allows the swing to be visualized three-dimensionally in the mind of the practitioner by switching the image output so that the movement directions of each video image match on the screen using a mirror image switching device.

Furthermore, as shown in Figure 10, we focused on the relationship between the weight shift between the left and right feet during the swing and the form, and recorded the changes in weight distribution during the swing and the swing motion simultaneously to determine the swing motion. There is also a golf swing analysis device that displays a weight shift transition 202 superimposed on an image 201 on a display 203 so that the quality of swing motion and weight shift can be analyzed by taking into account the timing of both (Japanese Patent Application Laid-open No. (Sho 60-24.864).

(Problems to be Solved by the Invention) However, the swing diagnosis device of JP-A-59-225.082 requires the input device 101 to be attached to the body, and furthermore, the input device 101 and the golf club 10 must be attached to the body.
2 are connected by a rigid arm 103,
It is the same as the conventional method in that the swing is performed under a certain degree of restraint, and the feeling is different from that during actual play. In addition, although it is said that the trajectory of the swing can be obtained, the grip end of the golf club and the golfer's
Since the input device is connected to the chest input device, what is actually obtained is a composite of the grip end and body movement, which is completely different from the swing trajectory. Moreover, even if you compare this trajectory with the trajectory of an exemplary professional golfer, you will notice that there is a difference between the two. How to move the parts to get the best shot, and how to move them is bad.
It is extremely difficult to cut Jl.

In addition, in the case of the video device for golf practice disclosed in JP-A-61-11,070, it is easier to reproduce the form and check the movement each time than when using a single video device, but it is It is not possible to compare and analyze the average repeated swings and specific swings, including the differences in swing form between times and bad times, good shots, bad shots, and frequent shots, etc., and just vaguely observe one's own swing. As much as possible, it is no better than the conventional deo-lined bandits.

Furthermore, in the case of the golf swing analysis device disclosed in JP-A No. 60-24,864, although it is possible to analyze the relationship between weight shift and swing, in other respects it is at the same level as the above-mentioned video device. It is something.

In addition, all of the conventional golf practice devices described above use the swing motion of a certain professional golfer as an ideal image, or the swing motion based on a theory held by the manufacturer as an ideal image, and this is the only correct one. This is a comparative diagnosis of a practitioner's form, etc. However, unlike machines and robots, humans have many degrees of freedom in movement and have unique movement characteristics such as individual physique, muscle strength, flexibility, etc. -v1
It can be said that it is inherently impossible to practice or correct the trajectory to something that approximates it. Moreover,
Each conventional training device has a different element, such as the rhythm of the swing, the trajectory of the swing, or the shift of weight, but use this as the only standard to obtain the correct or ideal swing. However, even if you master any one of them, it will not result in the best swing, nor can you say that even if you lack one of them, you will not be able to produce a nice shot. Normally, it is said that to get the best shot, swing rhythm, swing speed, swing layering, weight shift, etc. are optimal, but these are not independent but are interrelated. The best shot is not achieved just because any one element is not in its best condition or even if any other element is the best. In addition, these best conditions are not one uniform condition that applies to everyone, but are unique to each golfer and are determined by the individual movement characteristics. Is this a professional golf thing? - It is clear that not all golfers have the same form and swing motion, but rather each has its own unique swing that is the best for that person and cannot be imitated by anyone else.

Furthermore, even if one attempts to imitate the correct exemplary swing of a professional golfer, it is practically impossible to accurately trace the spatial trajectory of another person who has different movement characteristics. Normally, when we look at other people's desirable swings, we do not only see the trajectory of one particular point, but we also find characteristics in vague overall movements, such as shoulder movements. Even when looking at a golf ball, it is not possible to grasp the spatial trajectory of the shoulder accurately and quantitatively by attaching a lamp to the shoulder of a golfer. It's just a matter of understanding. In other words, when you refer to someone else's swing, you are simply capturing the general characteristics of the movement, recognizing it in a vague form, and then following it in a vague form, which is difficult to grasp visually. I realized that I judge rest based on the characteristics of the human movement, that is, the state close to a static state such as the start and end of the movement. Even if this movement can be quantitatively captured using measuring instruments, humans have unique movement characteristics and no two people have exactly the same movement characteristics, and the data is Although it is one of the factors that determine complex movements, it cannot be said to be accurate and sufficient data to understand the whole, so it is difficult to apply it as is and may even lead to incorrect judgment.
In the end, it can only be traced in a rough and ambiguous manner based on its own movement characteristics. In order to control human motor movements, which are complicated and meaningless to quantify, and apply them to human motor movements, which are also complicated to control, it is necessary to find out the characteristics of motor movements in an ambiguous manner. It is natural and accurate to deal with it based on your own motor behavior characteristics.

This is true in other sports as well.

Furthermore, correcting the swing motion of an individual alpha based on the swing of a specific person, such as a professional golfer, ignores the unique movement characteristics of that golfer. Concentrate all your attention on restraining or guiding the golf club to correct or imitate the movement 3-
Even if it goes well when you are in a V position, it is undeniable that once you are freed from these constraints, you will unconsciously shift to a swing based on each person's unique motor behavior. As a result, as practice progresses and the technique improves, each person usually achieves a stable motion based on their motor movement characteristics and establishes a unique form motion.

The point is, no matter how your form or swing is compared to the general theory, how can you maintain the best swing conditions that lead to your best shot during the competition? It is important to obtain a form and swing motion based on the characteristics, and to include not only good shots but also bad shots and so-so shots, so that the average repeated shots are no different from your best shots. A swing once created within one's own motor characteristics can be re-achieved, and it is much more effective to bring the swing closer to it and solidify it than to change one's own movement based on the swing of someone else with different motor characteristics. This is because it is considered to be easily possible.

Therefore, rather than using other people's movement characteristics as a standard, we decided to create our own best swing motion.
We devised a method that allows players to check their swings on the 44i, objectively and unambiguously understand how to eliminate variations in shots, and get closer to their own best swings.

Of course, it goes without saying that this line spacing method can also be applied to other sports such as batting, tennis, and swimming, as well as rehabilitation.

A first object of the present invention is to provide a device that can analyze human motor movements in a vague and ambiguous state. A second object of the present invention is to provide an apparatus that can perform comparative analysis of averagely repeated exercise movements using one's own optimal exercise movement as a standard. Furthermore, a third object of the present invention is to provide an apparatus that can perform comparative analysis of a specific exercise motion using one's own optimal exercise motion as a standard. The fourth purpose of this invention is to provide an analysis device that can simultaneously perform the above-mentioned analyzes in a region where the locus of motion is dry in space.

II. Structure of the Invention (Means for Solving Problems) In order to achieve the above object, the motor motion analysis device of the present invention, as shown in FIG. A marker 1 exhibiting an area with a density or temperature level equal to or higher than a predetermined value, an image input device 2 that captures the exercise movement of the subject as a digital image, and an input image that converts the input image into a binary image using the density or temperature level value of the marker as a threshold. A binary image processing unit 3 converts the binary image, and an image reconstruction processing unit 4 adds the binary image for one operation and reconstructs it into a direct image at a density level or pseudo color that is proportional to the frequency of appearance of one frame. and an image output device 9 that outputs the processed image, and the movement characteristics of the movement are determined by a spatial existence probability distribution image.The movement movement of the subject is determined by setting 1 density or temperature level of the marker as a threshold value. Only the motion data of the part to be analyzed as a value image is captured, and the images for one motion are added together to create a static multi-direction image with shading or color corresponding to the frequency of appearance. As an image!'14, it should be displayed as an image, so in bright areas (or vice versa), the subject's moving parts remain stationary for a long time, whereas in darker areas or completely dark areas, the static time is short or quick. It can be qualitatively determined that the object is moving, as well as the direction and speed of the movement.

Therefore, the motion of a predetermined region can be analyzed from this spatial existence probability distribution still image.

Further, as shown in FIG. 2, the present invention allows the above-mentioned motion motion analysis device to calculate the average spatial existence probability from all still multivalued images for a predetermined number of motions output from the image reconstruction processing unit 4. Average spatial existence probability distribution averaging processing unit 5 for calculating distribution
and continuous spatial existence probability distribution averaging that averages a group of static multi-value images corresponding to the optimal movement motion selected based on the subject's subjectivity from the static multi-value images output from the image reconstruction processing unit. It incorporates a processing unit 6, a comparison image reconstruction processing unit 7 which compares and calculates the average spatial existence probability distribution data and the optimal spatial existence probability distribution data to find the distribution difference, and an image output device 9 which outputs the processed image. , the uneven distribution of the optimal motion and average motion in the motion is determined from the spatial existence probability distribution image. If a spatial existence probability distribution appears in the output image, it indicates that the movement characteristics of the relevant movement of the subject are not constant, and the magnitude, direction, frequency, etc. of uneven distribution of movement can be determined depending on the degree of existence and distribution state. Can be analyzed.

Furthermore, as shown in FIG. 3, the present invention averages a group of multi-valued images corresponding to the optimal exercise motion selected based on the subject's subjective opinion from among the sequentially input images to the above-mentioned analysis device. an optimal spatial existence probability distribution averaging processing unit 6, optimal spatial existence probability distribution data output from the averaging processing unit, and spatial existence probability distribution data corresponding to a specific movement motion from the image reconstruction processing unit 4; A comparative image reconstruction processing unit 8 that calculates the distribution difference by comparing and calculating the distribution difference, and an image output device 9 that outputs the processed image are incorporated, and the maldistribution of a certain exercise movement with respect to the optimal movement movement at the current moment of the movement is incorporated. I'm trying to find out.

In this case, when the appearance of a spatial existence probability distribution is recognized in the output image, there is a difference between the subject's current optimal movement for the relevant movement and a specific movement, and the degree of existence and distribution state It is possible to analyze the magnitude, direction, frequency, etc. of uneven distribution of movements.

Further, in the present invention, a marker of one of the three primary colors of light is used, an image input device is used to capture the subject's motor movements as a digital color image, and the input image is subjected to image processing for each color image signal. By doing so, it is possible to detect even close areas where a number of pieces of motion data overlap on the screen at the same time.

(Example) Hereinafter, embodiments of the present invention will be described in detail based on the drawings.

FIG. 4 is a block diagram showing an embodiment of the motion analysis device of the present invention. The motor movement analysis system of the present invention includes a marker 1 that vaguely displays the movement part of the subject to be analyzed as an area of brightness equal to or higher than a predetermined value at the 9 degree level of a grayscale image, and a digital image of the movement movement of the subject. An image input device 2 to capture an image, a computer 10 that performs predetermined image processing/arithmetic processing on the input image to obtain a spatial existence probability distribution image, an image output device 9 that outputs the processed image, and a monitor for monitoring each process. It consists of incidental equipment such as

Marker 1 is for displaying as a level above a certain standard value in order to make it easier to capture only the movement part of the subject to be analyzed as image data, and is used to display a level above a certain reference value as a threshold value for the binary image processing unit. If a brightness or a temperature level is used as a threshold value, it includes all exothermic substances and the like that exceed that temperature level. Usually, the marker 1 is in the form of a patch that can be easily attached to and removed from the motion analysis area of the subject, or in the form of a marker that is sewn, colored, or woven into clothes, hats, sports equipment, etc. in advance. Furthermore, this marker may be one that exhibits a brightness above a predetermined concentration level when illuminated, or one that emits light by itself, and has a minimum area that makes it easy to detect the appearance frequency, that is, the spatial existence probability distribution. Although it is possible to use a device of any shape or size, such as a light emitting diode, it is preferable to use a device with a wide area that presents a vague area. In this specification, the term "subject" refers not only to humans, but also to general objects to be analyzed, such as animals, exercise equipment, clothing, hats, robots, and exercise machines.

The image input device 2 is an industrial television camera 21 (hereinafter abbreviated as ITV camera) which is usually easy to handle and inexpensive.
A device equipped with a D converter 23 is often used. However, in some cases, it is also possible to use image input means other than ITV, such as an infrared ITV camera, and in this case, a pyrogen is used as the marker 1. The most common image tube is the vidicon, and other photodiode arrays and COD
It is also possible to use a solid-state image carrier such as the following. The analog image signal from the ITV camera 21 passes through a camera control unit (hereinafter referred to as CCU) 22 that performs various controls such as input brightness adjustment, and is time-divisionally processed by a synchronization signal from a synchronization signal generator 24. After A/
Quantized (digitized) in the D converter 23,
The signal is output as a digital image signal to the binary image processing section 3 of the computer 10 either as it is or via an interface or the like. Note that the A/D converter 23 performs signal conversion in synchronization with the synchronization signal from the synchronization signal generator 24.

The Δ/D converter 23 converts the sampled image signal into
It has a function of digitizing, that is, suspending, depending on the brightness, and in the case of this embodiment, has a resolution of 8 bits. Since the Δ/D converter 23 is required to be high-speed due to the large amount of processing data, a parallel comparison type converter is used in this embodiment.

However, if the speed of quantization is not an issue, a circuit may be used in which a sample hold circuit or the like is used in the previous stage to temporarily hold the data, and then quantization is performed by a Δ/D converter. In addition, this image input device 1 performs sampling to obtain 256 x 256 pixels in relation to processing time, but it is not limited to this, and if the image is manually input or black and white data, 256 x 256 x 8 bit pixels are performed. The resolution is also sufficient. When a monochrome camera is used as the TV camera 21, an image signal is used as the input image signal, and when a color camera is used, a luminance signal or each color signal is used as the input image signal.

The image input device 1 is activated by a sensor composed of a light emitting diode and a phototransistor (not shown) placed at a position intersecting the trajectory of the movement (a position intersecting the trajectory of the golf club in the case of a golf swing), for example. This is carried out using a foot switch or the like operated by the person himself/herself, and when this sensor detects the start of the subject's exercise movement, it is configured to send an activation signal to the image input device 1. However, the startup switch for starting up the image input device 1 does not need to be limited to the configuration described above.
Mechanical switches and software are used to detect changes in the subject's motor movements, and changes over a predetermined amount within a predetermined time are detected.
The configuration may be such that when the motion data is detected, the motion motion data is processed. Note that a video and monitor 91 are connected to the CCtJ 22, and the imitated exercise motion is reproduced as necessary.

The computer 10 that processes the digital image data captured by the image input device 1 described above is a well-known computer, and is basically a ROM 13 into which a program is loaded.
, a central processing unit (hereinafter abbreviated as CPU) 11, a buffer (corresponding to RAM) 12 that stores image data before processing and processed image data in these processing units, and an input/output unit (Ilo>14). and performs predetermined image processing and arithmetic processing under control according to a program written in the ROM 13.

In the case of the analysis system shown in FIG. 1, the CPU 11 has a basic configuration consisting of a binary image processing section 3 and an image reconstruction processing section 4 that calculates hO for one operation of this binary image and reconstructs it into a still multivalued image. are doing. The binary image alpha processing section 3 binarizes the digital image data sent from the Δ/D converter 23 using the required polishing level as a threshold value, and converts it into a binary image. In the case of this embodiment, the beta type is adopted for the suspension, and it is divided into density levels of 8 for black and 255 for white. Therefore, the density level, for example, the threshold value, which is clearly distinguishable from the background and other moving parts (for example, clothing and skin) is set to 200 at an 8-bit density level. In this case, if you use marker 1 with a brightness level of 200, even if there is a difference in brightness between markers in different areas or markers in the same area due to lighting, the level of All of the 200 or more items are determined to have data and are imported. This processed data is c
This is determined by the internal logic of the puil, and if it is a negative logic, 0 is written, and if it is a positive logic, a 1 is written into the buffer 12. This binary image processing operation is performed every time one frame is input, and continues until the stop sensor detects that the subject's motor movement has stopped (i6 configuration. In this example, the sensor used is the same as the activation sensor (7p). The data will be processed. Therefore, the binary image of each frame is sequentially written into a predetermined address of the buffer 12.

The image reconstruction processing unit 4 is an image processing unit that adds 2 1 Fri images that are sequentially input for one operation and reconstructs them into a still multivalued image consisting of a grayscale image or a pseudocolor image that is proportional to the frequency of appearance of one frame. Therefore, the binary image of the operating valve must be added by sequentially adding the binary image of the next frame in synchronization with the binary image of the previous frame stored in the addition address of the buffer 12. It has the functions that can be used. This image is a still multivalued image that shows information that the brightness of the active part vaguely displayed by the marker is proportional to the frequency of appearance, that is, a spatial existence probability distribution in a distribution state with a certain density level.

In addition, the central processing unit of the other analysis system shown in FIG. 2 incorporates the comparison image reconstruction processing unit 7 into the above-mentioned analysis device shown in FIG. 1, and analyzes the difference between the average motion motion and the optimal motion motion. . In addition to the above-mentioned binary image processing unit 3 and image reconstruction processing unit 4, the CPU 11 includes an average spatial existence probability distribution averaging processing unit 5 controlled by a program stored in the ROM 13, and an optimal spatial existence probability distribution processing unit 5. It has an averaging processing section 6 and a comparison image reconstruction processing section 7 as its basic configuration, and uses the subjective judgment of the subject input from the voice sensor 15 or the detection result of the measuring device through a well-known pattern recognition circuit. It has the ability to calculate the distribution difference between a still multi-valued image obtained by the optimal movement motion selected by subjective judgment based on the above-mentioned subjective judgment and a still multi-valued image obtained by averaging all the movement movements. That is, it has a function of detecting the uneven distribution of the spatial probability distribution of the optimal motion and the spatial probability distribution of the average motion stored in the addition address. Note that the audio sensor section 15 has a function of detecting the subject's utterances, and usually uses a sound collecting microphone or the like.

The output of the audio sensor section 15 is converted into a digital signal via an A/D converter 16, and is input to the input/output section 14 of the computer. The average spatial existence probability distribution averaging processing unit 5 calculates an average distribution by adding and subtracting all the static multivalued images of the valves having a predetermined number of operations outputted from the image reconstruction processing unit 4. The optimal spatial existence probability distribution averaging processing unit 6 averages the multivalued images for the optimal exercise motion selected from all the multivalued images based on the subject's subjectivity. In the case of this embodiment, the selection of the optimal motor movement is performed by specifying the address of the buffer 112 to be stored based on the subjective opinion of the subject each time a series of motor movements is completed, either by voice or by operating a switch by the subject. It is said that For example, when the exercise action is a golf swing, the test subject judges his/her own swing subjectively or with a measuring device and then vocalizes the result, or inputs the measurement result into a computer as an electrical signal. In the case of audio, if a predetermined word is considered to be the best, it will generate a sound such as "Very Good" or "Nice Shot 1." At this time, a pattern recognition circuit analyzes this sound pattern and creates a sound that has been assigned in advance. A still multivalued image signal obtained by a series of movement operations is stored in the address of the buffer 12. That is, data is stored.It goes without saying that the address assignment by the pattern recognition circuit can be repeated in a number of ways. In this embodiment, the address instruction for storing multivalued image signals has been explained only using a pattern recognition circuit, but the invention is not limited to this, and an appropriate number of mechanical switches or other switches can be used. In addition, the selection of the specified address is not limited to direct input using the subject's voice input or machine operation, but can also be configured to select the ball flight distance, trajectory, flight distance, etc. from the ball launch direction, angle, ball rotation, etc. Known devices for estimating flight direction, etc., or detecting the speed, face angle, trajectory, hitting position, etc. of a golf club at the time of impact, and estimating flight distance, etc., such as JP-A-58-190.456, JP-A-Sho. 59-
141.965, JP-A No. 60-48.762, etc., the quality of the swing result is determined according to the level of the subject at that time, and the buffer 112 to be stored is selected accordingly. Also good. At the same time, the multivalued image obtained by the series of movement operations is also written to the addition address of the buffer 12.

The buffer 12 has a refresh memory function, and is composed of SRAM or DRAM.

In each of the averaging processing units 5 and 6, the averaged average existence probability distribution data (multivalued image) and the optimal spatial existence probability distribution data are calculated in the comparison image reconstruction processing unit 7 at the next stage. The deviation is required. That is, the comparison image reconstruction processing section 7 calculates the outputs from both the averaging processing sections 5 and 6 to obtain an ifi difference accordingly.

In addition, the central processing unit of the other analysis system shown in FIG. 3, except for the above-mentioned average spatial existence probability distribution averaging processing unit 5 shown in FIG. The image data is directly input to the comparison image reconstruction processing unit 8, and the average optimal spatial existence probability distribution selected based on the subject's subjectivity and the just-completed motor movement or buffer 1 are compared.
Spatial existence JAI rate weight distribution of a specific movement action read from 2! 5 differences are calculated, and the current necessity for one's own optimal exercise movement is displayed as an image to show the uneven distribution of specific exercise movements.

Also, wood buff? Although a method is adopted in which control is performed by the CPU 11, it is also possible to adopt a direct memory access (hereinafter referred to as DMA) method. C.P.
Buffer 1 control by U11 requires processing time because it requires program execution time, but circuit control becomes easier. On the other hand, in the DMA method, the digital image sent from the A/D converter 23 is directly stored in the buffer 12, so threshold processing is performed in the buffer 12.
It needs to be executed when data is read. However, it has the advantage that binary image data can be written to the buffer 12 at high speed.

Furthermore, a magnetic disk, for example, is employed as the storage device 19 that stores the image signals processed by the computer as fixed data.

The disk 19 is controlled by the CPU 11 (R configuration) and is used to store the required still multilevel image. This disk 19 is a hard disk having an appropriate memory capacity according to the data amount of the multilevel image , floppy disk, and the required image data is C.
It is configured to be written and read under the control of PLll. Further, this disk 19 stores multivalued image data corresponding to past exercise movements of the subject, and can be read out by the CPU 11 as needed.

A hard copy device or a soft copy device such as a CRT is usually used as the image output branch @10 that outputs the image information processed in the image reconstruction processing section 4 or the comparative image reconstruction processing sections 2 to 8. . The hard copy device]O is controlled by the cPU], and the buffer? ] 2 has a function of printing out a spatial existence probability distribution image consisting of a multivalued image stored in .

The hard copy device 10 is preferably a dot printer, a heat insulating printer, or a color printer for pseudo-color display to make the distribution state easier to see. The monitor 92 is connected to the CPU 11 via the D/A converter 17, and under the control of the CPU 11, the buffer 1
It has a function of displaying the spatial existence probability distribution image data stored in 2 or the disk 19, that is, the multivalued image data for one motion. In this embodiment, the monitor 92 employs a CRT display. In addition, the terminal 18 is connected to the CPU 11, and the buffer 1
It serves as an input/output terminal for transmitting image data stored on the disk 2 and the disk 19 to the outside. That is, the motor motion analysis device of the present invention can be placed on a network, and by being controlled by a host, the motor motion analysis device of the present invention can be used to analyze a test subject's motor motion using a motor motion analysis device or various computers located at other locations. The structure is as follows. If the distance between terminal devices in the network is long, an appropriate model device may be used. The communication format is not particularly limited either, and may be considered individually depending on the network used.

In this example, only one CPU"l'l was explained, but a parallel CPU system in which one CPU is used to input the exercise motion of the subject to the buffer aco 2 and other data processing is performed by other CPIJs is also available. In this case, the processing speed of various data becomes high, and other logical processing becomes possible.

Also, the reason why the CCU 22 has three outputs, R, G, and B, is to take into consideration the case where the ITV 21 is changed to one that can input color words, and the marker 1
Using the three primary colors of light, red, blue, and green, we sample these R, G, and B image signals and simultaneously perform motion motion analysis of different parts while preventing data overlap.
It is made in quru@ construction.

Hereinafter, analysis of the golf swing motion using the present motion motion analysis device will be explained in detail based on the flowcharts of FIGS. 5 to 7.

First, the subject appropriately determines his or her stance on a predetermined position or a subjectively determined position. A golf pole is fixed at a predetermined position, and the subject's position and motion with respect to the pole are analyzed using this ball as a reference. On the other hand, the subject attaches a marker 1 representing a vague area of a predetermined density level or higher to a region whose exercise behavior is to be analyzed, such as the shoulder, elbow, head, knee, hand, or lower back. Marker 1 has an 8-bit beta type concentration level that can be clearly distinguished from the body and background, for example, 200
However, the value is not particularly limited to this value.

First, the subject controls the power supply control, internal OS software, etc. to turn the device into rsTARTJ. "Activation sensor ON?J detects whether the subject has started a movement or not, and when the activation sensor is turned ON, it sends an activation signal to the CCU 22 (Step 1).

"Image human power J is operated by the ON signal of the activation sensor, and is inputted as an analog image that is the subject's exercise movement data via the input device I-cho V21 and CCLJ22. This input image is sent to the A/D converter 23. The sample is sampled into 256 x 256 pixels per scanning line and converted into a digital image signal with an 8-bit density level (Steps 2 and 3). This is performed sequentially until detection, and the input image data of several + frames is sequentially captured as a digital image.This digital input image is subjected to threshold 1 direct processing with a threshold of about 200 at a certain density level, for example, 8-bit density. 2
The image is converted into an ff1 image (step 4). That is, only information about the movement of a vague area of the active part displayed by the marker 1 is captured as image data, and the movement of other areas, the background, and the part are discarded. This "2Hani image processing" is performed by 2Ifi, which is composed of CPU11 software.
This is performed in the image processing section 3. "Data temporary store" is to temporarily store binary image data in the internal RAM or at a predetermined address in the buffer 112.
Controlled by U11 (step 5). Therefore, the exercise action images transferred from the image input device 2 are sequentially converted into binary images and stored in the memory. ``The stop sensor ONJ detects whether the subject's exercise movement has ended or not. If the exercise movement continues, it branches before step 2, and when it detects that the exercise movement has ended, it branches to the next step. Proceed to step .
Stop sensor ONJ is a routine that functions by the stop sensor. In such a flowchart, the image data for the number + frames collected while the subject's exercise movement continues are sequentially stored in the internal RAM or buffer 1 as binarized image data until the stop sensor is turned on.
2 will be remembered. When the stop sensor detects that the subject's exercise movement has ended, the process proceeds to step 7. This F
"Store to addition address" is the buffer specified by the addition address allocated in advance to buffer 5? 1
The binary image data temporarily stored in the internal RA~1 or the buffer 12 is transferred to the 7th address of 2. Further, the buffer memory data at this addition address is incremented one after another if the binary image data is H (positive logic). Therefore, when golf swings are performed 12T times and the trajectory of the same movement is followed, the memory data in the buffer 12 becomes large (step 7). That is, 2 IFB
By adding the images, a still multivalued image with a density level or color depending on the frequency of appearance in the space is obtained. The next step, ``result output,'' reads the data in the buffer 12 specified by the addition address, and uses the H/C (hard copy) device 9 and monitor 92, which are image output devices, to display a static or static image during a certain motor movement of the subject. Information in a state close to this, that is, spatial existence i-fold distribution, is output as a grayscale image with brightness proportional to the frequency of appearance or a pseudo-color image specified by the frequency of appearance. This "result output" is performed by the image reconstruction processing unit 4 configured by software of the CPU 11 (step 8). After performing the above IP raccoon operation, it reaches "END" and ends the operation.

Next, the operation of other systems will be explained in detail based on the flowchart shown in FIG. 1st glance, same flow-
The same routine name will be given to each routine, and the explanation will be omitted.

In this case as well, a marker 1 with a brightness of at least a predetermined concentration level is attached to the part of the subject whose motor movement is to be analyzed.

"Voice pattern recognition" in step 10 uses a well-known pattern recognition circuit to compare the input voice data and the stored voice data to determine which data they match. , the address in the buffer 12 at which the binary image data is to be stored is determined based on the subject's voice. The address of this buffer 12 is pre-assigned to vl. For example, when the voice ``Berry, Gutsudo'' is recognized, the oooo address is set as the first address, and when the voice ``Gutsutoko'' is recognized, the address is set to 01.
00 address is assigned as the first address. If the subject determines that the current swing is the optimal exercise motion based on his/her subjective opinion, he/she generates a sound saying "Verify" (step 10). ``Berry-GOOD?J'' is a routine that uses a pattern recognition circuit to determine whether or not the current swing is an optimal motor movement.If the subject's voice pattern is [Berry-GOOD?
Branch to ``Store to Very Good Address'' in step 2, and if it is not [very good], rGOO in step 13
D? Brunch at j. [Store to very good address] has the function of storing binary image data to the very good address of the buffer 12 (step 1).
2). Also, step 13 rGOOD? J is a routine that determines whether the subject's voice is "gutsudo" or not.
In the case of [Good], the process branches to step 14, ``Store to GOOD address'', and otherwise, the process branches to step 15, ``Normal?''. Step 14
Store J stores the binary image data in the rGOOD address of the buffer 12 at the good address (the address specified by the voice saying good). Also, step 1
5 [Normal? If the branch is to J, it is determined whether the voice is "Normal" or not. If the voice is [Normal], it is branched to "S1 to normal address". In step 16), the first voice is If it is not [Normal], step 17 "Branch to store J to pad address. Step 16 1 Store to normal address" stores the binary image data at the normal address of buffer 12 (the address instructed by the voice saying "normal"). )
This is a routine that stores ``Store to pad address'' in step 17 is a routine for storing the binarized density level data in the pad address of the buffer 12 (address designated by a voice called ``hat''). "Store to Berry GOOD address", rGOO
Store 1 to D address “Store J to normal address
Alternatively, the binary image data that has flowed through the store J routine to the second address in step 1B is processed with the data at the memory address of the buffer 12 indicated by the addition address in step 1B. Ru. This [store to addition address] is controlled by the software of the CPU 11. However, "store to added address" means that the binary image data is added for each movement movement, whereas "store to very added address", rGOO
“Store to D address” “Store to normal address”
and [Store to bad address] are routines that load newly fetched 22-value image data. Step 19, "Specified number of times completed?" is a routine that determines whether the number of swing operations has completed the specified number of times, and if the number of swings has not reached the specified number of times, before the step 1, "Activation sensor ON?" If the specified number of times has been reached, the process branches to step 20, ``averaging''.

"Averaging" is the buffer 12 indicated by the addition address.
This is a routine that averages the data, that is, the multivalued image,
The optimum motion selected based on the average spatial existence probability distribution and the subject's subjectivity can be obtained from all still multivalued images for a predetermined number of motions. This "average ecology" is performed by the average spatial existence probability distribution averaging processing section 6. The "deviation calculation" is a calculation of the data in the buffer 112 specified by the addition address and the data in the buffer 12 specified by the verification address, and is executed by the comparison image reconstruction processing section 7. . This calculation result is then sent to H via the “result output”.
The data displayed and printed by the image output device 9 is printed out by the /C device 9 or displayed on the monitor 92 (step 21.22). Then, the average movement is compared with the own optimal movement movement (which is also averaged), and the distribution difference is found. That is,
The difference between the optimal motor action and the average motor action can be detected as an existence probability distribution proportional to the frequency of appearance.

Further, another analysis will be explained based on the flowchart of FIG. Note that only the parts that are different from the above flowchart will be explained.

This analysis device converts the input image of each movement into a binary image, temporarily stores it in the buffer 112, and constructs a multivalued image by adding the images.The analysis device then converts the input image of each movement into a binary image, composes a multivalued image by adding the images, and then converts the input image into a binary image to form a multivalued image. The image reconstruction processing unit 8 compares the latest input data or specific data read out from the buffer 12 with the static multilevel image data corresponding to the above-mentioned optimal operation. They differ in that they perform comparison operations.

Step 30 “Very Gutsdo? J is the previous step 1
This is a routine that determines whether or not the current swing was the optimal motion based on the voice pattern recognition of 0. If it is the optimal motion, it branches to step 31 "Store to Berrygood addition address" and stores the optimal motion. If it doesn't work, go to step 32, "Is there verified data?"
Brunch on. The binary image data stored in the veri-gts address is added with past data (step 31), and then averaged and reconstructed into an averaged multi-valued image (step 33), which is then stored in the veri-gts address. (Step 34). Step 32, ``Very good data available'', determines whether static multi-valued image data of the optimal exercise movement selected by the subject based on subjectivity has been written to the address of the buffer 712 indicated by the Helly Grad address. This is a routine. If there is past veri-gtsd data, load it from the veri-gtsd address (step 35), compare it with the current swing data, and calculate the spatial existence probability distribution difference (step 3).
6). If there is only one pitching data, that data is used, and if there is data on the number of pitches, the average value is used. Then, the calculation result is sent to the outside via "result output" in step 37. On the other hand, if there is no very good data, the process jumps after step 37 and reaches "end", where the operation ends. Through the above processing, it is possible to output the difference between the movement movement to be analyzed and the self-optimal movement movement as an image of the existence 11'"rate molecule ji" proportional to the frequency of appearance.

1 Effect of the Invention As is clear from the above explanation, the present invention provides a marker that exhibits an area of concentration or temperature level equal to or higher than a predetermined value in a region where the exercise behavior of the subject is to be analyzed, and An image human-powered device that captures movements as digital images;
a binary image processing unit that converts the input image into a binary image using the temperature or temperature level value of the marker as a threshold;
an image reconstruction processing unit that calculates 0 for one operation of a value image and reconstructs it into a still multivalued image with a temperature level or pseudo color proportional to the frequency of appearance of one frame; and an image output device that outputs the processed image. Since the motion characteristics are determined using 5 images of spatially existing stochastic molecules, information in which the difference in brightness ratio and hue is proportional to the degree of appearance, that is, the spatially existing probability distribution can be obtained from the image. This spatial existence probability distribution 5
is displayed in an ambiguous form, with the front and back of the movement being close to a stationary state, or the analysis of human movement movements that assumes subjectivity and ambiguity may result in the movement being perceived in ambiguous terms. It is more accurate/able to understand the movement more accurately than using fixed expressions.

Therefore, if this motion motion analysis system is used to analyze the golf swing, it will be possible to obtain the results related to the golf swing as shown in Fig. The motion characteristics are expressed as a spatial existence probability distribution image, in other words, the feature quantities of human motion that can be visually perceived are captured in a state close to a static state such as the start and end of the motion. From this image, the overall movements of the practitioner's head, feet, shield, golf club, etc. can be determined. For example, according to the same figure, the left foot (front side), which is the pivot foot, ■
It is observed that the ball is firmly fixed and moves in the same position as the right foot swings. In addition, it can be seen that with the backswing, the left leg ■ side moves slightly to the lower right, while the right leg ■ side moves slightly to the upper right, and the torsion of the upper body is easily compensated for.

In addition, the image of the golf club is clearly visible at the top position of the backswing, and an afterimage is observed leading up to it, indicating that the takeback was gradual until it reached the top, and sufficient "preparation" was taken. After that, the image of the golf club shows that the swing is not stopped until the end position, and the image of the golf club is rapidly dropped.
It turns out that they are completely friends. In other words, it can be seen that a suitable swing rhythm of slowly taking back and quickly swinging down is being executed. And, while the image shows a state in which the player is leaning over J and T, it is clearly seen that the position of the grip end at the time of finishing ■ is approximately at the shoulder position ■ and is pulled back after the follow-through. I can understand what you are trying to do. In addition, the image (2) at the center of the body has a horizontally long, almost horizontal distribution, and it is recognized that the body is rotating in the horizontal plane, that is, the body is not moving up or down. Furthermore, the head [stock] is
The distribution is concentrated in one point and no movement is observed.

I can understand that he did not take his eyes off the ball ■ when he was hit. Therefore, when these analyzes are taken together, it can be concluded that this swing has no major drawbacks. By the way, this image was taken by professional golfer Al Guiber (AL,
This is an analysis of the swing of GEIBERGER.

The present invention also provides an average spatial existence probability distribution that calculates an average spatial existence probability distribution from all static multivalued images for a predetermined number of movements output from the image reconstruction processing unit in the above-mentioned movement motion analysis device. B. An averaging processing section and an optimal space for averaging a group of still multi-valued images corresponding to the optimal exercise motion selected based on the subject's subjectivity from the still multi-valued images output from the image reconstruction processing section. an existence probability distribution averaging processing unit, the average spatial existence probability distribution data and the maximum
The comparative image reconstruction processing unit that compares and calculates the distribution difference with the existence probability distribution data and the image output device that outputs the processed image are refined, and the existence of the optimal motion motion and the average motion motion that are interlocked are narrowed down. It is determined from the 1LIi existence probability distribution image. If the output image has a certain spatial presence, it indicates that the movement duration of the relevant movement of the subject is constant, and the uneven distribution of the movement can be determined depending on the distribution state of the degree of existence. Size, direction, frequency, etc. can be analyzed.

Furthermore, the present invention provides an optimal spatial existence probability distribution image that averages a group of multi-valued images corresponding to the optimal movement motion selected based on the subjectivity of the subject from among the images sequentially inputted to the above-mentioned analysis device. an averaging processing unit; and a comparative image reconstruction processing unit that compares and calculates the distribution difference between the optimal spatial existence probability distribution output from the averaging processing unit and the spatial existence probability distribution corresponding to a specific movement motion. and an image output device that outputs the processed image, and the maldistribution of a certain specific movement with respect to the current optimal movement of the movement is determined. In this case, when the appearance of a spatial existence probability distribution is recognized in the output image, there is a difference between the subject's current optimal movement for the relevant movement and a specific movement, and the degree of existence and distribution state indicate that It is possible to analyze the magnitude, direction, frequency, etc. of uneven distribution of

Furthermore, the present invention uses markers of any of the three primary colors of light, uses an image input device to capture the subject's motor movements as digital color images, and processes the input images for each color image signal. By doing so, it becomes possible to detect even close areas where a number of motion data overlap on the screen at the same time. As a result of the operation, when the binary images are added, markers of different colors are assigned to parts on almost the same trajectory that are weighted on the images to prevent data from being overlapped.

[Brief explanation of the drawing]

Fig. 1 is a functional block diagram showing the basic configuration of the motion analysis device of the present invention, Fig. 2 is a functional block diagram of the 46th invention,
FIG. 3 is a functional block diagram of another invention, FIG. 4 is a block diagram showing an embodiment of the motion analysis device according to the present invention, and FIGS. 5 to 7 are flowcharts of each invention. FIG. 8 is an explanatory diagram showing an example of a still multivalued image and a golf swing at 41, and FIGS. 9 and 10 are front views showing an example of a conventional golf practice device. 1...Marker, 2...Image input processing, 3...2
Value image processing section, 4... Image regrooving processing section, 5... Average spatial existence probability distribution averaging processing section, 6... Optimal spatial existence probability distribution averaging processing section, 7... Comparison Image reconstruction processing section, 8... Comparison image reconstruction processing section, 9... Image output device. Patent applicant Nihon Furnace Kogyo Co., Ltd. Figure 5 Figure 9 Figure 10

Claims (6)

[Claims] (1) A marker that exhibits a region with a concentration level or temperature level above a predetermined value in a region where the exercise movement of the subject is to be analyzed; an image input device that captures the exercise movement of the subject as a digital image; and an image input device that captures the exercise movement of the subject as a digital image; A binary image processing unit that converts a binary image into a binary image using a density or temperature level value as a threshold, and a static multivalued image of a density level or pseudo color that is proportional to the frequency of appearance of one frame by adding this binary image for one operation. A motion motion analysis device comprising an image reconstruction processing section that reconstructs an image and an image output device that outputs the processed image, and characterized in that the motion characteristics of the motion are determined using a spatial existence probability distribution image. (2) While displaying the marker in one of the three primary colors of light, the image input device is used to capture the subject's motor movements as a digital color image, and the input image is image-processed for each color image signal. A motion motion analysis device according to claim 1. (3) a marker that exhibits a region with a concentration level or temperature level above a predetermined value in a region where the exercise movement of the subject is to be analyzed; an image input device that captures the exercise movement of the subject as a digital image; and an image input device that captures the exercise movement of the subject as a digital image; A binary image processing unit that converts a binary image into a binary image using a density or temperature level value as a threshold, and a static multivalued image of a density level or pseudo color that is proportional to the frequency of appearance of one frame by adding this binary image for one operation. An image reconstruction processing unit that reconstructs an image, and an average spatial existence probability distribution averaging processing unit that calculates an average spatial existence probability distribution from all still multivalued images for a predetermined number of operations output from the processing unit. and an optimal spatial existence probability distribution averaging process that averages a group of static multivalued images corresponding to the optimal movement motion selected based on the subject's subjectivity from the static multivalued images output from the image reconstruction processing unit. a comparative image reconstruction processing section that compares and calculates the distribution difference between the average spatial existence probability distribution data and the optimal spatial existence probability distribution data; and an image output device that outputs the processed image. A motion motion analysis device characterized in that the uneven distribution of optimal motion motion and average motion motion is determined from a spatial existence probability distribution image. (4) While displaying the marker in one of the three primary colors of light, the image input device is used to capture the subject's motor movements as a digital color image, and the input image is image-processed for each color image signal. A motion motion analysis device according to claim 3. (5) a marker that exhibits an area with a concentration level or temperature level above a predetermined value in a region where the exercise movement of the subject is to be analyzed; an image input device that captures the exercise movement of the subject as a digital image; and an image input device that captures the exercise movement of the subject as a digital image; A binary image processing unit that converts a binary image into a binary image using a density or temperature level value as a threshold, and a static multivalued image of a density level or pseudo color that is proportional to the frequency of appearance of one frame by adding this binary image for one operation. An image reconstruction processing unit that reconstructs an image, and an optimal spatial existence probability distribution average that averages a group of multivalued images corresponding to the optimal movement motion selected from sequentially input images based on the subject's subjective opinion. an averaging processing unit, and calculating a comparison between the optimal spatial existence probability distribution outputted from the averaging processing unit and the spatial existence probability distribution corresponding to a specific motion action from the image reconstruction processing unit, and calculating a distribution difference therebetween. An exercise motion comprising a comparative image reconstruction processing unit and an image output device that outputs the processed image, and characterized in that the maldistribution of the current motion with respect to the self-optimal motion of the motion is determined from a spatial existence probability distribution image. Analysis device. (6) While displaying the marker in one of the three primary colors of light, the image input device is used to capture the subject's motor movements as a digital color image, and the input image is image-processed for each color image signal. A motion motion analysis device according to claim 5.