KR102226927B1 - Using the gesture and EMG sensing data, motion analysis system of free weight exercise by hybrid detection method - Google Patents

Abstract

The present invention relates to a free weight exercise motion analysis system using a hybrid detection method using a motion sensor and an EMG sensor, and more particularly, a complex motion motion using a motion sensor and a muscle sensor. It relates to a free weight motion motion analysis and feedback system that enables a free weight motion with a relatively wide range of motion through analysis in a safe manner.

Description

Using the gesture and EMG sensing data, motion analysis system of free weight exercise by hybrid detection method}

The present invention relates to a free weight exercise motion analysis system using a hybrid detection method using a motion sensor and an EMG sensor, and more particularly, a complex motion motion using a motion sensor and a muscle sensor. It relates to a free weight motion motion analysis and feedback system that enables a free weight motion with a relatively wide range of motion through analysis in a safe manner.

Recently, as income level has improved and interest in appearance has increased, the number of people who want to manage their body through constant exercise is increasing, and they spend a lot of time in fitness or other sports facilities to wear clothes that show their body or to show off their legs. Actually.

Exercises related to posture correction or weight training can be largely divided into mechanical exercise and free weight exercise.

The most effective mechanical exercise is the bench press, and the free weight exercise includes dead lift and squat, and these three are called the three major exercise of fitness exercise.

Among the various body parts, squat exercises are practiced a lot to save the hip line and leg line. In foreign countries, the buzzword'shut up and squat' spreads so that squat is loved by people all over the world, and it is called the'king of exercise'. It is enough to be called.

Squat exercise is a motion of sitting and standing until the thigh is level with the knee. It is performed as a basic exercise for lower body exercise in yoga academy, Pilates academy, sports hospital, and rehabilitation hospital as well as a gym.

Squat exercise is excellent for correcting legs and strengthening muscle strength, but it is a high-intensity exercise that is highly likely to cause body injuries, so you must work out in the correct posture and in a state of being prepared for injuries.

The correct squat posture is to sit until the knee and thigh line is perpendicular, taking care not to bring the knee forward in front of the second toe, but the deformed knee and pelvis movement and improper squat posture during the squat exercise may cause back pain and knee pain. I can.

Therefore, correct posture is a very important factor in squat exercise, and recently, squat is used not only as exercise but also as a medical test method to evaluate incorrect lower body movement and overall posture.

Since the devices for squat exercise on the market are simply composed of a weight and a body to increase the load on the body, there is a limit to confirming the alignment of the body and analyzing the movement or giving feedback for correct movement.

The tool that can analyze movement during squat exercise (FMS KIT) is not objective because it is checked only by the therapist's eyes, and there are problems with consistency, such as relying on the therapist's personal competence, and incurring a lot of cost during personal training There is a drawback.

Incorrect alignment of the legs during squat exercise increases the load on the knee joint and causes knee injury.If the center of weight is tilted forward, the injury to the lower back increases.Therefore, correct alignment of the body during squat exercise is important to prevent body damage. Do.

Therefore, a system that can analyze incorrect movements and give feedback so that correct movements can be performed is needed to prevent injury and correct body alignment during squat exercise.Accordingly, exercise achievement through human interface function and ICT function for exercise interest There is a need for a way to maximize exercise effect through recording and database of effects, and posture monitoring.

In the end, the present invention proposes a system capable of inducing correct body alignment and preventing damage to the body by analyzing the above-described squat exercise as well as the free weight exercise motion and feeding back to the exerciser. In addition, the generated free weight motion model can be used as basic data for robot motion control, VR environment design, and solution development for establishing optimal motion motion.

Korean Patent Publication No. 10-2016-0094060

Therefore, the present invention was devised to solve the above conventional problems,

An object of the present invention is to enable a safe free weight exercise through a complex exercise motion analysis using a motion sensor and a muscle sensor.

In order to achieve the problem to be solved by the present invention, the free weight exercise motion analysis system by a hybrid detection method using a motion sensor and a muscle sensor according to an embodiment of the present invention,

It is configured to include two or more cameras, through which the depth map is calculated, and the motion motion state analysis device 100 for analyzing the motion motion state,

An EMG biological signal device 200 configured to include an EMG band and to filter and amplify the fine electrical signal obtained from the EMG band and provide it to an analysis control device through wireless communication;

By comprising an analysis control device 300 for optimizing exercise information by synthesizing the exercise motion state information of the free weight exercise obtained from the exercise motion state analysis device and the EMG biological signal information obtained from the EMG biological signal device, The problem of the present invention is solved.

The free weight exercise motion analysis system according to the hybrid detection method using a motion sensor and a muscle sensor according to the present invention,

Motion sensor and muscle sensor are used to provide the effect of enabling safe free weight movement through complex movement motion analysis, and can be used for intelligent robot or motion pattern analysis and motion process analysis using stored digital data. do.

Specifically, it is possible to provide safe free weight exercise through complex exercise motion analysis using a gesture recognition sensor and a muscle sensor (EMG, electromyogram). As is important, safe free weight exercise is possible by limiting the allowable range of the human body and the feedback signal of the mechanical load, and the muscles after a certain exercise time through numerical monitoring of the body balance and range of motion, which are important in free weight exercise. It provides a human body monitoring feedback function that supports safe exercise by measuring the amount of exercise by accumulating the fatigue level of the body, and developing an intelligent device by acquiring a variety of data on the exercise program to maximize the exercise effect through stimulation of various muscles. It can be used as data for the exercise, and it can be used as exercise posture correction data by securing data to correct the error pattern of free weight exercise.

1 is an overall configuration diagram of a system for analyzing a motion of a free weight motion using a hybrid detection method using a motion sensor and a muscle sensor according to an embodiment of the present invention.
2 is a block diagram of a motion motion state analysis apparatus of a free weight motion motion analysis system using a hybrid detection method using a motion sensor and a muscle sensor according to an embodiment of the present invention.
3 is an exemplary diagram of acquiring a stereo image using two cameras.
4 is an exemplary diagram for finding a corresponding point using two cameras.
5 is an exemplary view of a pattern using structured light.
6 is an exemplary diagram of a human body joint for motion recognition.
7 is an exemplary diagram showing motion recognition of a right-hand push-up and a left-hand wave shape with two sample operations.
8 is a block diagram of an EMG biological signal device of a system for analyzing a motion of a free weight motion using a hybrid detection method using a motion sensor and a muscle sensor according to an embodiment of the present invention.
9 is a circuit diagram of a differential amplifier of a free weight motion motion analysis system using a hybrid detection method using a motion sensor and a muscle sensor according to an embodiment of the present invention.
10 is a conceptual configuration diagram of an EMG biological signal device of a free weight exercise motion analysis system using a hybrid detection method using a motion sensor and a muscle sensor according to an embodiment of the present invention.
FIG. 11 is a synthesis of motion motion and EMG biosignal data for the free weight motion of the analysis control device of the free weight motion motion analysis system using a hybrid detection method using a motion sensor and a muscle sensor according to an embodiment of the present invention. This is a flow chart showing the process of acquiring and optimizing data.
12 is a ground loop circuit diagram of a system for analyzing a motion of a free weight motion according to a hybrid detection method using a motion sensor and a muscle sensor according to an embodiment of the present invention.

A free weight exercise motion analysis system using a hybrid detection method using a motion sensor and a muscle sensor according to an embodiment of the present invention,

It is configured to include two or more cameras, through which the depth map is calculated, and the motion motion state analysis device 100 for analyzing the motion motion state,

An EMG biological signal device 200 configured to include an EMG band and to filter and amplify the fine electrical signal obtained from the EMG band and provide it to an analysis control device through wireless communication;

It is configured to include an analysis control device 300 for optimizing exercise information by synthesizing the exercise motion state information of the free weight exercise obtained from the exercise motion state analysis device and the EMG biological signal information obtained from the EMG biological signal device. It is characterized.

At this time, the analysis control device 300,

It is characterized by detecting and optimizing the exercise motion state information and EMG bio-signal information of the free weight exercise, converting it into a database, monitoring it, and giving feedback to the user, so that it is possible to monitor a safe exercise with a large motion range.

At this time, the motion motion state analysis device 100,

At least two or more cameras for photographing an exercise image of an athlete; And

A depth map calculation unit for calculating a parallax map by determining a corresponding point having the same characteristic in the images obtained from the two cameras, and calculating a depth map by converting the calculated parallax map into a distance; And

It characterized in that it comprises a; motion motion state analysis unit for analyzing the motion motion state of the image input through the camera.

At this time, the EMG biological signal device 200,

EMG band for acquiring a microelectrical signal that is attached to the muscle of the exerciser and flowing through the muscle; and

Filtering amplification unit for filtering the fine electrical signal obtained from the EMG band and amplifying the filtered signal; And

And an embedded unit for acquiring the amplified signal and providing it to the analysis control device through wireless communication.

Hereinafter, it will be described in detail through an embodiment of a system for analyzing a free weight motion motion using a hybrid detection method using a motion sensor and a muscle sensor according to the present invention.

The squat exercise described in the present invention is an important free weight exercise for training the muscles of the lower extremities and strengthening the trunk and lower body muscles. There is a need for a device that can analyze incorrect movements for alignment of the body and give feedback so that correct movements can be performed.

To this end, a free weight motion analysis system using a hybrid detection device using a motion sensor and a muscle sensor provides a function of monitoring the motion state of a free weight with a wide range of motion.

In addition, the stored digital data can be used in future intelligent robots or motion pattern analysis and motion processes.

In addition, if the leg alignment is not correct during squat exercise, the load on the knee joint increases, causing injury to the knee area, and if the center of weight is tilted forward, it increases the injury to the lower back. Alignment is an important factor.

In order to check this, the motion sensor and the muscle sensor as described above are used.

1 is an overall configuration diagram of a system for analyzing a motion of a free weight motion using a hybrid detection method using a motion sensor and a muscle sensor according to an embodiment of the present invention.

As shown in Fig. 1, the free weight exercise motion analysis system based on a hybrid detection method using a motion sensor and a muscle sensor is largely a motion motion state analysis device 100, an EMG biological signal device 200, and an analysis It is configured to include a control device (300).

In the following, the motion motion state analysis device 100 is to be defined as a motion sensor, and the EMG biological signal device 200 is to be defined as a muscle sensor.

The motion motion state analysis apparatus 100 includes two or more cameras, calculates a depth map through this, and analyzes the motion motion state.

In addition, the EMG biological signal device 200 includes an EMG band, and performs a function of filtering and amplifying the fine electrical signal obtained from the EMG band and providing it to an analysis control device through wireless communication. .

Accordingly, the analysis control device 300 optimizes the exercise information by synthesizing the motion motion state information of the free weight exercise obtained from the motion motion state analysis device and the EMG biological signal information obtained from the EMG biological signal device.

The free weight exercise described in the present invention is a squat, dead lift, push-up, etc., and, unlike a machine exercise moving in a fixed trajectory, the shaking of each segment of the body is controlled. Therefore, it is effective in increasing muscle coordination and muscle mass and reducing body fat, but there is a risk of injury due to failure to control exercise or weight.

Injuries caused by free weight exercise are largely caused by functional limitations and body imbalance, which leads to decrease in exercise performance and musculoskeletal pain, so free weight training that analyzes faulty movement patterns and prevents them in advance. An evaluation model for this is required.

In addition, free weight exercise is a complex joint exercise that mobilizes multiple joints at the same time, not just an exercise that improves major muscles, and is a rapid development of the healthcare industry such as medical devices, VR, telemedicine services, analysis of momentum based on big data, and establishment of a robot motion system. Requires a systematic, objective, and numerically quantified analysis method of free weight training.

2 is a block diagram of a motion motion state analysis apparatus of a free weight motion motion analysis system using a hybrid detection method using a motion sensor and a muscle sensor according to an embodiment of the present invention.

As shown in Figure 2, the motion motion state analysis device 100,

At least two or more cameras 110 for photographing an exercise image of an athlete; And

A depth map calculator 120 for calculating a parallax map by determining a corresponding point having the same characteristic in the images acquired from the two cameras, and calculating a depth map by converting the calculated parallax map into a distance; And

It characterized in that it comprises a; motion motion state analysis unit 130 for analyzing the motion motion state of the image input through the camera.

Among the image application technologies using computers, 3D vision processing technology plays an important role, and is one of the very important applied technologies in analyzing motion movements.

Gestures expressed as movement movements mean physical, voice, and emotional movements or expressions that function as symbolic, and are expressed as gestures, and the dictionary meaning is the use of limbs or body as a means of expression, and also thoughts, emotions, and attitudes. It can be defined as the movement of the body or limb to express or emphasize.

This motion recognition technology is a technology in which a computer or a robot autonomously analyzes and recognizes human behavior, and automatically analyzes how each part of the body changes over time and interprets the change in an abstract sense. Any movement to manipulate the computer is true.

There are two methods of acquiring the depth of an image for recognizing a motion motion.

Among them, the passive method refers to an approach that acquires multiple images using two or more cameras and analyzes them to find out depth information. The active method shoots a pattern or wave on the scene and analyzes it. How to find out.

Two or more cameras 110 may be configured as image sensors, the image sensor being a semiconductor that converts photons into electrons and displays them on a display or stores them in a storage device, and a light-receiving element that converts a light-receiving signal into an electrical signal, It consists of a pixel circuit part that amplifies and compresses the converted electrical signal and an ASIC part that processes image signals by converting the preprocessed analog signal to digital, and there are types such as CCD, CMOS, and CIS (Contact Image Sensor).

The stereo image for recognizing motion motion is a technique that extracts 3D images and information using two or more cameras by imitating the human visual system.It is similar to the principle that a person measures the distance using two eyes. Facilitating the problem of finding a corresponding point by keeping the eyes of the person in mind.

3 is an exemplary diagram of acquiring a stereo image using two cameras.

4 is an exemplary diagram for finding a corresponding point using two cameras.

As shown in FIG. 3, in the step of making a stereo system using two cameras, radiation distortion and tangential distortion of the lens are first removed, and the angle and distance between the two cameras are adjusted.

That is, the depth map calculation unit 120 calculates a parallax map by determining a corresponding point having the same characteristic in an image obtained from two cameras, and calculates a depth map by converting the calculated parallax map into a distance.

Specifically, a corresponding point having the same characteristics is found in the A-camera and B-camera images, and a parallax map is obtained through this.

The parallax map is converted to a distance using a trigonometric method based on the camera position, and this process is called reprojection, and a depth map is obtained through this.

라 할 수 있으며, 삼각비를 적용하면, 물체까지의 거리(깊이) Z_W는

가 된다.If the distance to the object is Z,

If you apply the trigonometric ratio, the distance (depth) Z _W to the object is

Becomes.

Xi and f are known, but the value Xw is a condition of unknown value, so Z is _W, but can not confirm the Z _W which extends to the stereo geometry with two cameras.

Active sensor In stereo image, it only plays a role of passively receiving light information coming into the camera, but in the case of active sensor, it plays a role of finding the distance from the returning signal by actively projecting something onto the scene.

Structured light replaces a camera with a projector in stereo, and the projector projects a sophisticated pattern designed to make it easier to find a corresponding point (see Fig. 4).

Therefore, the calculation of the depth using the corresponding point is as Equation 1 below.

(수식1)

(Equation 1)

Patterns used by structured light Every point has a unique pattern to distinguish it from others, and it is easy to find a corresponding point.

As for the pattern used by the structured light, the difference between the corresponding point of the black and white pattern and the corresponding three-color pattern can be confirmed through an example (see FIG. 5)

5 is an exemplary view of a pattern using structured light.

In the case of Microsoftware's keynet, infrared rays are projected to make it invisible.

In addition, a 480*640 RGB image and a depth image can be simultaneously acquired at 30 fps, and the depth of the acquired image can be measured approximately 0.8 to 3.0 meters only in an indoor environment.

The viewing angle is about 57° horizontally and 43° vertically, and can be tilted up and down by 27°.

The motion recognition sensor using a keynet sensor emits infrared rays from an infrared illuminator, scans the subject through infrared reflection, and recognizes the depth of infrared rays reflected by the object through CMOS camera recognition technology, and performs image processing to detect the sensor. Recognize the distance from the play through (see Fig. 6).

6 is an exemplary diagram of a human body joint for motion recognition.

In addition, the color recognized by the color image CMOS camera and the above information are collected, and the motion is processed in the form of a skeleton to recognize real-time player motion information.

Image sensors can be largely divided into CCD (Charge Coupled Device) type and CMOS (Complementary Metal Oxide Semiconductor) type according to their structure.

In a CCD-type image sensor with 4 pixels, when an image is formed on the surface of the image sensor by an optical lens, each pixel creates a number of signal electron groups proportional to the intensity of light entering the position for a specified time (1 field). .

The biggest feature of the CCD type is that a group of signal electrons generated in each pixel by light is moved to the output section by a pulse applied to the gate of the CCD structure, converted into a voltage at the output section, and then sent out sequentially.

That is, the number of electrons generated in each pixel becomes information about light up to the output unit.

Functions outside the CCD type image sensor, such as CDS (Correlated Double Sampling), AGC (Auto Gain Control), and A/D converter (Analog to Digital Converter), can be integrated into the chip in the CMOS type image sensor.

In addition, the TG (Timing Generator), which generates the chip driving signal, is also integrated inside the chip, and because the driving voltage is low, all functions can be converted into one-chip, which is the biggest advantage of a CMOS image sensor.

On the other hand, according to an additional aspect, the motion motion state analysis device 100 is characterized in that it is configured to further include an IR light camera.

That is, if the two cameras are for obtaining a stereo image, the IR light camera is for analyzing a depth.

Of course, it can be obtained through two stereo cameras, but by additionally configuring the IR light camera as necessary, it is possible to analyze the depth more accurately.

The IR light camera is a camera device that collects radiant energy emitted by an object without supplying light from the outside, visualizes it through appropriate conversion, and makes it easier to recognize.

It is different from the existing imaging equipment that observes by the presence or absence of visible light or the difference in the intensity of reflected light.

Since the difference in radiant energy emitted per unit area and per unit time, that is, radiant emittance or exitance, is imaged, observation is possible even at night when there is no light at all.

The configuration of the thermal infrared camera varies depending on the wavelength used, the required performance, the detector, and the infrared detection method.

The thermal infrared camera consists of an optical lens part, an infrared sensor part (μ-Bolometer/ROIC), and an image realization part.

The HCI application program (Human-computer Interaction application) applied from the viewpoint of robotics is mainly used to utilize a pre-defined set of joint motions in the motion recognition process.

Dynamic time warping (DTW) is a template matching algorithm.

r(i), i=1, 2, 3, 4... i, t(j), j = 1, 2, 3, 4... j Assumes a reference vector and a test vector, respectively.

로 표현할 수 있다.This is the information about the position vector.

It can be expressed as

The depth map calculator calculates and compares the distances with all reference vectors for the test vector, and can calculate the position vector by matching the reference vector with the smallest distance and the test vector.

The motion motion state analysis unit 130 performs a function of analyzing the motion motion state of the image input through the camera.

That is, the motion sequence is a concatenation of N of these feature vectors.

The pre-processing step removes the variation of the shape vector due to the size and position of the person in the field of view (FOV) of the camera.

The distance with the weight of the human joint can be represented by a specific motion classification.

는 관절(joint)별 가중치를 나타낸다. here,

Represents the weight for each joint.

Here, FIG. 7 is an exemplary diagram showing motion recognition of a right-hand push-up and a left-hand wave shape by two sample operations.

That is, the motion motion state analysis unit 130 analyzes the motion motion state of the image input through the camera.

8 is a block diagram of an EMG biological signal device of a system for analyzing a motion of a free weight motion using a hybrid detection method using a motion sensor and a muscle sensor according to an embodiment of the present invention.

As shown in Figure 8, the EMG biological signal device 200,

EMG band 210 for acquiring a microelectric signal that is attached to the muscle of the exerciser and flowing through the muscle; And

Filtering amplification unit 220 for filtering the fine electrical signal obtained from the EMG band and amplifying the filtered signal; And

And an embedded unit 230 for acquiring the amplified signal and providing it to the analysis control device through wireless communication.

Muscles are controlled by nerves, and microscopic currents are always flowing.

Measuring the change in muscle state by examining the electrical activity of these muscles is called an electromyogram.

There are two types of EMG tests: invasive and non-invasive methods, and non-invasive methods include superficial EMG, which detects muscle signals by attaching electrodes to the skin.

Surface electromyography (sEMG) is a measurement method in which electrodes are attached to the skin surface. Unlike needle electromyography (nEMG), which measures only one motor unit action potential trains (MUAPs), it is a painless and comfortable group of muscle movement units. It is possible to quantitatively analyze the overall synergistic activity of the musculoskeletal system, which is highly utilized in the rehabilitation field related to musculoskeletal diseases.

EMG can diagnose functional abnormalities rather than structural characteristics of muscles.

Particularly, the most common musculoskeletal disease, RSI (Repetitive Strain Injury) caused by repeated muscle use, WRMD (Work Related Musculoskeletal Disorder) caused by excessive work, Myofascial Pain Syndrome, which causes pain in a specific muscle area called trigger point, CFIDS (Chronic Fatigue & Immune Dysfunction Syndrome), Fibromyalgia, etc. The use of surface EMG is high in diagnosis and rehabilitation training.

Among these human biological signals, EMG signal processing is used in the field of rehabilitation engineering to obtain human motion and force information, and real-time signal processing is possible by extracting and analyzing EMG signals generated according to human movement, and obtaining bio signals. The signal generated through the miniaturized, modularized EMG sensor suitable for this can be processed together with the motion motion sensor to secure muscle fatigue and digitalized data to process information.

The arrangement of electrodes for EMG measurement is to monitor only the activity of the relevant muscle area as the distance between the reference electrode and the measurement electrode decreases (specific-mode), and as the distance increases, the general-mode of various muscle areas is monitored. Can be monitored.

The electrodes can be attached to any location where the muscle is located. During free weight exercise, electrodes are attached to the muscle part used for exercise, and the facial muscle part can be mainly used for studies related to emotions and emotions such as stress.

Surface EMG methods include RMS, power spectrum, median frequency, mean frequency, RMS-probability distribution, RMS cross-correlation, and MVIC normalization. Etc.

In particular, RMS reflects muscle contraction or muscle tone, Median Frequency reflects muscle fatigue information, and cross-correlation between muscle regions provides information on muscle contraction timing.

Utilizing this quantitative analysis of EMG can more accurately identify EMG characteristics of normal and abnormal muscles.

Recently, more various analysis methods such as wavelet analysis, principal component analysis, nonlinear dynamics analysis, high-order statistical analysis, and Laplacian mapping have been introduced, and technologies that enhance sensitivity and specificity in clinical use of surface EMG are being utilized.

In order to sense the EMG, it includes an EMG band unit 210 for acquiring a bio-signal, a filtering amplification unit 220, and an embedded unit 230.

At this time, the embedded unit 230 includes an AD conversion stage and a transmission stage capable of generating a digital signal and transmitting it wirelessly through Bluetooth communication.

As an electrode to be attached to the skin for signal measurement, an Ag/Agcl electrode, which is widely used in surface electromyography and has excellent sensitivity, is used, and a bipolar snap electrode and a snap electrode for grounding are used to detect bipolar electrical signals (mainly EMG) from the electrode. Fine surface EMG signals were obtained precisely.

The EMG signal measured in the human muscle epidermis appears as a very fine electrical signal with hundreds to thousands of μV, so in order to transmit such a minute EMG signal to a general embedded part (DSP, Microprocessor, etc.), the signal level that can be processed by DSP EMG signal needs to be amplified.

For this, a differential amplifier is used to amplify the micro signal measured from the muscle by about 1000 times or more.

The circuit and operation of the differential amplifier are shown in FIG. 9.

For example, when an input voltage is simultaneously applied to an inverting input (-) terminal and a non-inverting input (+) terminal of an OPAMP, it is configured as a differential amplifier.

In general, useful information in human surface EMG signals is known to be distributed between 8 and 500 Hz, so the amplified signals are analog filtered.

Design and apply a band pass filter by applying a high pass filter (HPF) with a cutoff frequency of 8Hz and a low pass filter (LPF) with a cutoff frequency of 500Hz.

The filtered signal is now detected as an analog signal of the order of hundreds to thousands of mV that can be read by the DSP.

The digitally converted EMG signal has the advantage of easy signal processing.

As shown in FIG. 10, microelectrical signals flowing through the muscles detected through multiple EMG electrodes attached to the band are transmitted to the embedded unit 230 through filtering and amplification, and this is transmitted to the embedded unit 230 through Bluetooth wireless communication. 300).

The transmitted signal is integrated together with the motion motion signal and recognized as motion monitoring data.

Based on the accumulated data, not only simple condition determination, but also data for each user so that it can be used as data data for intelligent exercise movements, is read as a feedback signal.

On the other hand, the analysis control device 300 of the present invention is for optimizing the exercise information by synthesizing the exercise motion state information of the free weight exercise obtained from the exercise motion state analysis device and the EMG biosignal information obtained from the EMG biological signal device. Will perform the function.

That is, multi-channel signal motion recognition technology (MSGR, Multi-signal gesture recognition framework) is applied, which is achieved by synthesizing two or more signals, and for this purpose, the analysis control device constitutes a signal synthesis unit.

으로 나타낼 수 있고, 각각의 추출된 디지털신호, 즉, 센서에서 추출된 벡터

에 대해서 운동동작 상태 이미지는

의 벡터로 나타내고 EMG 생체바이오 신호는

로 나타낸다. At this time, the signal synthesis unit is the state equation

Can be expressed as, and each extracted digital signal, that is, a vector extracted from the sensor

About the motion motion state image

And the EMG bio-bio signal is

Represented by

At this time, the multi-channel signal shape vector is as shown in Equation 2

(수식2)

(Equation 2)

In addition, a process of obtaining and optimizing data by synthesizing the motion motion for the free weight exercise and EMG biometric data is shown in FIG. 11.

As described above, when the analysis control device 300 is configured, it detects and optimizes the motion motion sensing and bio-signals, converts it into a database, monitors it, and provides feedback to the user, thereby ensuring high safety by monitoring a safe motion with a large motion range. It becomes.

Meanwhile, as shown in FIG. 1, in order to ensure the accuracy of measurement, in addition to the specification of a simple data sheet, various parameters must be considered.

Especially in noisy locations or outdoors, understanding the environment in which the application operates is an important factor for accurate measurements.

Common examples of noise that adversely affect signals are ground loops, high in-phase voltages, and electromagnetic radiation. Techniques used to remove noise from the measurement system include appropriate shielding, cable connection, and termination.

The in-phase voltage is a voltage that exists at both the positive (+) terminal and the negative (-) terminal of the instrumentation amplifier, and the maximum operating voltage depends on the isolation circuit, not the input range of the amplifier.

It is rare for the in-phase voltage to consist only of DC levels, and most in-phase voltage sources contain AC components as well as DC offsets.

In particular, when the low level analog signal passes through the instrumentation amplifier of the DAQ device, the noise appears severely.

Sources of AC noise can be classified into capacitive, inductive, and radiative types according to the coupling mechanism, and DAQ equipment with high CMRR over a wide frequency range is used. If selected, the overall noise of the system can be lowered.

Meanwhile, as shown in FIG. 12, the ground loop is one of the most commonly seen noise sources in data acquisition applications.

A ground loop occurs when the ground phases of two terminals connected in a circuit are different from each other. If the ground phases are different, current flows in the interconnections, resulting in an offset error.

Insulation barriers can be utilized to ensure stability in current loop applications.

Digital logic can also be affected by a noisy environment, and may lead to incorrect on/off values or accidental triggers, so using 24V logic to increase the noise margin can be less affected by noise.

Through the analysis control device 300 of the present invention described so far, by detecting and optimizing the exercise operation state information and EMG bio-signal information of the free weight exercise, converting it into a database, monitoring it, and giving feedback to the user, a safe exercise with a large motion range You will be able to monitor it.

Therefore, the motion sensor and the muscle sensor are used to provide the effect of enabling safe free weight movement through complex movement motion analysis, and can be used for intelligent robot or motion pattern analysis and motion process analysis using stored digital data. You will be able to.

For example, applying to virtual reality (Vertual Reality) for exercise and rehabilitation treatment, using the free weight training evaluation model as a resource for national health DB based on Big Data, as a basic data for rehabilitation treatment related software and applications, in telemedicine systems. It can be used for rehabilitation exercises, correction exercises, exercise prescriptions, etc., and can be used to establish a shape system for humanoid robot motions.

Those skilled in the art to which the present invention with the above contents pertains will understand that the present invention can be implemented in other specific forms without changing the technical spirit or essential features of the present invention. Therefore, the embodiments described above are exemplified in all respects and should be understood as non-limiting.

100: motion motion state analysis device
200: EMG biological signal device
300: analysis control device

Claims (4)

In the free weight exercise motion analysis system using a hybrid detection method using a motion sensor and a muscle sensor,
At least two or more cameras 110 for photographing an exercise image of an athlete; And
A depth map calculator 120 for calculating a parallax map by determining a corresponding point having the same characteristic in the images obtained from the two cameras, and calculating a depth map by converting the calculated parallax map into a distance; And
A motion motion condition analysis unit 130 configured to analyze motion motion condition of an image input through the camera; and
EMG band 210 for acquiring a microelectric signal that is attached to the muscle of the exerciser and flowing through the muscle; And
Filtering amplification unit 220 for filtering the fine electrical signal obtained from the EMG band and amplifying the filtered signal; And
An EMG biological signal device 200 comprising; an embedded unit 230 for acquiring the amplified signal and providing it to an analysis control device through wireless communication,
Consisting of including an analysis control device 300 for optimizing exercise information by synthesizing the exercise motion state information of the free weight exercise obtained from the exercise motion state analysis device and the EMG biological signal information obtained from the EMG biological signal device,
The depth map calculation unit 120,
Characterized in that the position vector is calculated by matching the reference vector with the smallest distance and the test vector by calculating and comparing the distances with all reference vectors for the test vector using Equation 1 below,
The motion motion state analysis unit 130,
It is characterized in that it can be expressed as a specific motion classification by referring to a distance having a weight of a human joint using Equation 2 below,
Analysis control device 300,
It is characterized by synthesizing two or more signals by configuring a signal synthesis unit and applying a multi-channel signal motion recognition technology (MSGR, Multi-signal gesture recognition framework).
The signal synthesis unit is a state equation for the continuation of the tense of the input image.

The vector extracted from the sensor, which is each extracted digital signal, can be represented by

About the motion motion state image

And the EMG bio-bio signal is

Characterized in that it is represented by,
(X is a vector,

Is a T-dimensional vector,

Is the state vector of the motion motion detected by the camera,

Is the state vector of the motor motion detected in the EMG)
At this time, the multi-channel signal shape vector is a free weight exercise motion analysis system by a hybrid detection method using a motion sensor and a muscle sensor, characterized in that using Equation 3 below.

(Equation 1)
(The f _n is a function for the position vector, X ₁ ,Y ₁ ,Z ₁ ,X ₂ ,Y ₂ ,Z ₂ ,...X ₆ ,Y ₆ ,Z ₆ are position vector values)

(

Is the weight per joint,

Is the location information including the human joint weight, Equation 2)

(Equation 3)

delete delete delete

Images