KR101613055B1 - Method of motion reasoning by adjusting biological signal using noise and motion reasoning device for the same - Google Patents
Method of motion reasoning by adjusting biological signal using noise and motion reasoning device for the same Download PDFInfo
- Publication number
- KR101613055B1 KR101613055B1 KR1020150092159A KR20150092159A KR101613055B1 KR 101613055 B1 KR101613055 B1 KR 101613055B1 KR 1020150092159 A KR1020150092159 A KR 1020150092159A KR 20150092159 A KR20150092159 A KR 20150092159A KR 101613055 B1 KR101613055 B1 KR 101613055B1
- Authority
- KR
- South Korea
- Prior art keywords
- probability model
- value
- signal
- ratio
- sensor
- Prior art date
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
- G06F3/011—Arrangements for interaction with the human body, e.g. for user immersion in virtual reality
- G06F3/015—Input arrangements based on nervous system activity detection, e.g. brain waves [EEG] detection, electromyograms [EMG] detection, electrodermal response detection
Landscapes
- Engineering & Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Neurosurgery (AREA)
- General Health & Medical Sciences (AREA)
- Neurology (AREA)
- Health & Medical Sciences (AREA)
- Dermatology (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Biomedical Technology (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The present invention discloses a method of calibrating a biological signal detected by a biological signal sensor and inferring an operation. The operation inference method according to the present invention includes: a probability model generation step of generating a probability model for each operation corresponding to a user operation command and detecting and storing a first pause signal strength (S1); Detecting a second pause signal strength (S2) in an operation monitoring mode; (R) of the first pause signal intensity (S1) and the second pause signal strength (S2), and corrects the correction value using the feature factor value (F2) detected by the user operation and the ratio ≪ / RTI > And inferring the operation by comparing the corrected feature value (F2 ') with the stored probability model.
According to the present invention, in the process of generating a probability model, the feature parameter value is corrected using the ratio of the stored noise signal intensity and the noise signal intensity detected in the operation monitoring mode, or the probability model is corrected, Accurate motion inference can be performed even in a situation. In addition, since the user does not have to perform a troublesome training process to calibrate the probability model, the usability is greatly improved.
Description
The present invention relates to a method for inferring a user's action using a bio-signal detected by a user operation, and more specifically, it relates to a method for inferring a user's action by using a noise, And accurately estimating the motion of the user.
In general, a biological signal refers to an electrical or magnetic signal generated in the human body. Typically, an electrocardiogram (EMG), an electrocardiogram (ECG), an electroencephalogram (EEG), a safety graph (EOG) And the like.
Such bio-signals have been mainly used for therapeutic or rehabilitative purposes in the medical field. However, in recent years, there has been a problem that the user's operation in the human-computer interface (HCI) or the human-machine interface The range of application is increasing for the purpose of controlling the operation of computers, machines, robots, etc.
In order to control a computer, a machine, a robot, or the like using such a bio-signal, it is very important to accurately not only detect a bio-signal accurately but also accurately infer the user's operation intention from the detected bio-signal.
In recent years, a specific operation corresponding to an operation command is repeated several times to several tens of times in a state where a user wears or attaches a bio-signal sensor (e.g., an electromyogram sensor) A training process is performed to generate a probability model for each operation and a sensor, and a method of inferring an operation based on a detected feature parameter value against a probability model when a user performs an operation is widely used.
However, various kinds of noise are included in the signal detected by the bio-signal sensor.
For example, a signal detected by an electromyographic sensor attached to a user's wrist or forearm may include weak signals not only in the EMG signal but also in other parts of the body such as a pulse signal and an EEG signal, May also be included.
As described above, other bio-signals and electromagnetic waves other than the EMG signal are a kind of noise. The noise is contained in a signal detected at an active time when the user performs an operation, (Relaxation time).
However, it is very important to solve the noise problem in order to increase the accuracy of the operation reasoning, since it is difficult to accurately extract the feature value necessary for the operation reasoning if the noise includes such noise excessively.
Recently, various methods for removing noise included in a biological signal have been introduced. As an example,
As another example,
Since it is very difficult to completely remove the noise included in the bio-signal, it is necessary to study a method of maximizing the accuracy of the operation reasoning based on the signal including the noise.
On the other hand, the noise included in the bio-signals is not always constant and varies depending on various factors such as the physical condition of the user, the skin condition, the contact state of the sensor and the skin, and the surrounding environment. Therefore, even if the same user performs the same operation, the feature value of the detection signal changes every time, so that the accuracy of the operation reasoning can not be maintained constant.
For example, FIG. 1 shows a normal distribution (Gaussian) probability model serving as a basis for operational reasoning. The probability model of characteristic factor values (for example, line length and line length) varies greatly depending on the SNR Respectively.
If the noise level is changed, the probability model that is the basis of the motion reasoning is changed. Therefore, if the existing probability model is used as it is, the accuracy of the motion reasoning is greatly degraded.
Therefore, in order to precisely infer operation, a probability model must be newly set according to the noise level. In order to set a new probability model, the user must perform a training process of repeating a plurality of operations corresponding to an operation command every several to several tens of times. There is a problem that is very cumbersome and inconvenient. In addition, since the noise level varies from time to time due to various factors, it is practically impossible to perform such a calibration procedure every time.
Therefore, it is necessary to provide a new method that can maintain or improve the accuracy of motion inference even when the size of noise changes.
SUMMARY OF THE INVENTION The present invention has been made in view of the above problems, and it is an object of the present invention to correct the feature value and / or probability model of a signal detected by a bio-signal sensor using noise, .
According to an aspect of the present invention, there is provided a method for inferring an operation by correcting a bio-signal detected by a bio-signal sensor, the method comprising: generating a probability model for each operation corresponding to a user operation command; A probability model generating step of detecting and storing the pause signal strength S1; Detecting a second pause signal strength (S2) in an operation monitoring mode; (R) of the first pause signal intensity (S1) and the second pause signal strength (S2), and corrects the correction value using the feature factor value (F2) detected by the user operation and the ratio ≪ / RTI > And inferring the operation by comparing the corrected feature value F2 'with the stored probability model, thereby providing a method for inferring the operation by correcting the bio-signal.
In the operation inference method according to an aspect of the present invention, the ratio r is S2 / S1, and the calibrated characteristic factor value F2 'is obtained by multiplying the characteristic factor value F2 by the ratio r And multiplying it by the inverse number. The corrected feature value F2 'may be calculated by multiplying the feature value F2 by the reciprocal of the ratio r and the weight.
According to another aspect of the present invention, there is provided a method for inferring an operation by correcting a bio-signal detected by a bio-signal sensor, comprising the steps of: detecting and storing a characteristic factor value F1 for each operation corresponding to a user operation command; A probability model generating step of generating a probability model parameter using the feature parameter value F1 and detecting and storing the first pause signal strength S1; Detecting a second pause signal strength (S2) in an operation monitoring mode; (R) of the first pause signal intensity (S1) and the second paused signal intensity (S2), and calculates a corrected characteristic factor value (r) by using the characteristic factor value (F1) F1 '); A probability model calibration step of correcting the stored probability model using the corrected feature value F1 '; And correcting the bio-signal including the step of inferring an operation by comparing the feature value (F2) detected by the user operation with the corrected probability model.
In the operation inference method according to another aspect of the present invention, the ratio r is S2 / S1, the corrected characteristic factor value F1 'is calculated by multiplying the characteristic factor value F1 by the ratio r. . In this case, the corrected feature value F1 'is calculated by multiplying the feature value F1 by the ratio r and the weight.
The bio-signal sensor may include a plurality of sensors, and the ratio r may be calculated for each sensor.
According to another aspect of the present invention, there is provided an operation reasoning device for calibrating a user's operation by correcting a bio-signal detected by a bio-signal sensor, the device comprising: a probability generator for generating and storing a probability model for each sensor, Model generation means; Noise detection means for detecting a first pause signal strength (S1) in a probability model generation mode and for detecting a second pause signal strength (S2) in a motion monitoring mode; Noise ratio calculating means for calculating a ratio (r) of the first pause signal intensity (S1) and the second pause signal strength (S2); Calibration means for calculating a feature factor value F2 'calibrated using the feature factor value F2 detected by the user's operation and the ratio r; And an operation reasoning means for inferring an operation by comparing the corrected feature value (F2 ') with the stored probability model.
According to another aspect of the present invention, there is provided an operation reasoning device for correcting a biological signal detected by a bio-signal sensor to deduce a user's operation, characterized in that a characteristic factor value F1 for each sensor is detected A probability model generating means for generating and storing a probability model parameter using the feature parameter value F1; Noise detection means for detecting a first pause signal strength (S1) in a probability model generation mode and for detecting a second pause signal strength (S2) in a motion monitoring mode; Noise ratio calculating means for calculating a ratio (r) of the first pause signal intensity (S1) and the second pause signal strength (S2); The feature parameter value F1 'corrected using the feature parameter value F1 and the ratio r is calculated and the probability model parameter is updated using the corrected feature parameter value F1' Correcting means for correcting the model; And an operation reasoning means for inferring an operation by comparing the feature value (F2) detected by the user operation with the corrected probability model.
According to the present invention, in the process of generating a probability model, the feature parameter value is corrected using the ratio of the stored noise signal intensity and the noise signal intensity detected in the operation monitoring mode, or the probability model is corrected, Accurate motion inference can be performed even in a situation. In addition, since the user does not have to perform a troublesome training process to calibrate the probability model, the usability is greatly improved.
1 is a diagram showing a normal distribution probability model according to a signal-to-noise ratio (SNR)
2 is a graph showing a relationship between a signal-to-noise ratio (SNR) and a characteristic parameter value of a biological signal in an operating unit and a resting unit
3 is a flowchart showing a method of generating a probability model according to an embodiment of the present invention.
4 is a flowchart showing an operation reasoning method according to the first embodiment of the present invention
5 is a flowchart showing an operation reasoning method according to a second embodiment of the present invention
6 is a diagram illustrating a case where a normal distribution probability model is calibrated using noise;
7 is a schematic diagram of an operation reasoning device using bio-signals according to an embodiment of the present invention
Hereinafter, preferred embodiments of the present invention will be described with reference to the drawings.
2 is a graph showing a relationship between a feature value (e.g., line length, line length) and a signal-to-noise ratio (SNR) of a signal detected by a bio-signal sensor (e.g., an electromyogram sensor) . Here, the resting signal corresponds to noise since it is a signal that is detected when the user does not perform any operation.
However, in FIG. 2, it can be seen that there is a relatively constant correlation between the characteristic value of the idle period and the characteristic value of the idle period in the section where the SNR is relatively large. That is, when the noise increases, the characteristic factor value of the operating unit also increases proportionally.
The present invention is based on this point, and is characterized in that a feature parameter value is corrected using a ratio of a noise signal intensity detected in an operation monitoring mode waiting for input of a user operation command and a noise signal intensity detected at the time of setting a probability model, To a method capable of maintaining or improving the accuracy of motion inference despite variations in noise.
The operation inference method according to the present invention is based on a probability model, and therefore, a probability model that is a basis of operation inference must first be set.
As shown in FIG. 3, in order to generate the probability model, the user first switches the operation reasoning device to the probability model generation mode with the bio-signal sensor (e.g., an electromyogram sensor) mounted. (ST11)
When switched to the probability model generation mode, the motion reasoning device detects and stores the first idle signal power (S1) using a pause signal detected by all the sensors before the user action is input. The first pause signal strength S1 corresponds to the noise signal strength at the time of generation of the probability model. Needless to say, the first pause signal strength S1 must be detected and stored for each sensor.
However, the timing of detecting the pause signal is not particularly limited. Accordingly, the first pause signal strength S1 may be a signal intensity at a specific point in time before the user's operation is input, or may be a mean, a median, or the like of the signal strength sampled during the set- It is possible. It may be a signal intensity at a specific point in time after the user's operation for training is terminated or may be a mean, median, etc. of the signal strength sampled during a set time after the user's operation is terminated . (ST12)
Then, the user must perform a training process of repeating a specific operation corresponding to the operation command several times to several tens of times. The operation reasoning device acquires and stores a signal generated by each operation from all connected bio-signal sensors. (ST13, ST14)
Then, the operation reasoning device stores the characteristic factor value (F1) of each operation using the signal obtained in the above step. Here, the characteristic factor may be an average value of the detected signal, an entropy, a signal intensity, a line length, and the like. (ST15, ST16)
If there are a plurality of operations corresponding to the operation command, a training process of repeating the operation corresponding to another operation command several times to several tens of times should be performed, and similarly, the characteristic factor value F1 of each operation And extracts it. (ST17, ST18)
Then, by using the extracted feature parameter values, a probability model for each sensor of each operation is generated, and parameter values of each probability model are stored. For example, when m operations are inferred using n sensors, n * m probability models should be created and parameter values should be stored for each probability model. (ST19)
On the other hand, there are the normal distribution (Gaussian) probability model, the Poisson distribution probability model, and the gamma distribution probability model in the probability distribution model. In the normal distribution probability model, The variance is used as a parameter, and in the gamma distribution probability model, the gamma value of the feature parameter is used as a parameter.
There are two methods for inferring the user's behavior by using the probability model generated through the above-described process, the first pause signal strength (S1) per sensor, and the feature value (F1) In the following, each method will be described in detail.
1. First Embodiment
The operation inference method according to the first embodiment of the present invention is a method of inferring an operation by correcting the feature factor value F2 detected from each bio-signal sensor in the operation monitoring mode without correcting the probability model.
4, first, the operation reasoning device detects a signal from all the sensors in the operation monitoring mode and detects and stores a second pause signal power S2. The second pause signal strength S2 is regarded as the noise intensity at the time when the user action is input. It is needless to say that the second pause signal strength S2 must be detected and stored for each sensor.
The timing of detecting the second pause signal strength S2 is not particularly limited. Accordingly, the second pause signal strength S2 may be a signal strength detected at a specific point in time before the user operation is input in the operation monitoring mode, a mean of the signal strength sampled during the set time before the user operation is input, Median, etc. < / RTI > It may also be a signal strength detected at a specific point in time after the user's operation is terminated, or a mean, median, etc. of the signal strength sampled during the set-up period after the user's operation is terminated. (ST31)
Then, when a user wearing a bio-signal sensor (e.g., an electromyogram sensor) performs a specific operation corresponding to an operation command, the operation reasoning device detects a signal from all connected bio-signal sensors, And stores the characteristic factor value F2 for each sensor. Here, the characteristic factor may be an average value of the detected signal, an entropy, a signal intensity, a line length, and the like. (ST32, ST33, ST34)
The operation reasoning device then calculates the ratio r of the first pause signal strength S1 and the second pause signal strength S2. Hereinafter, the ratio r will be defined as S2 / S1. Since the time of calculating the ratio r is not particularly limited, it may be calculated immediately after the second pause signal strength S2 is detected in step ST31.
Meanwhile, since the biological signal sensors to the first and second pause signal strength (S1.S2) it is also detected by each of sensor n when n individual, the ratio (r) is a ratio r 1 to r n obtained for each sensor. (ST35)
Subsequently, the operation reasoning device multiplies the inverse numbers (1 / r 1 to 1 / r n ) of the per-sensor ratios r 1 to r n to the detected feature factor values
The accuracy of the operation reasoning may not be reduced or improved even though the noise ratio r is reflected in the calculated characteristic factor value F2 depending on the type and complexity of the operation. In this case, if the data for various noise levels are accumulated for each sensor and the weights necessary for calibration of the characteristic factor value F2 are empirically obtained and stored, the characteristic factor value F2 can be more accurately corrected have. That is, if the specific operation or the sensor is set to multiply the predetermined weight (alpha) in addition to the reciprocal of the ratio (1 / r), the accuracy of the operation reasoning can be further improved. (ST36)
Next, user behavior is deduced by comparing the calibrated characteristic factor value (F2 ') for each sensor with a previously set probability model. (ST37)
According to the first embodiment of the present invention, by correcting the feature factor value F2 detected by the user operation to the new feature factor value F2 'using the noise ratio r, There is an advantage that the probability model of the model can be used as it is without correction.
2. Second Embodiment
The operation reasoning method according to the second embodiment of the present invention corrects the existing probability model using the noise ratio r and compares the detected feature value F2 detected in the operation monitoring mode with the corrected probability model .
Referring to FIG. 5, in this case, the operation reasoning device detects signals from all the sensors in the operation monitoring mode and detects and stores the second idle signal power (S2) as in the first embodiment. The second pause signal strength S2 is also detected and stored for each sensor, and the detection point of the second pause signal strength S2 is as described above. (ST51)
Then, when a user wearing a bio-signal sensor (e.g., an electromyogram sensor) performs a specific operation corresponding to an operation command, the operation reasoning device detects a signal from all connected bio-signal sensors, And stores the characteristic factor value F2 for each sensor. Here, the characteristic factor may be an average value of the detected signal, an entropy, a signal intensity, a line length, and the like. (ST52, ST53, ST54)
The operation reasoning device then calculates the ratio r of the first pause signal strength S1 and the second pause signal strength S2. Since the time of calculating the ratio r is not particularly limited, it may be calculated after the second pause signal strength S2 is detected in step ST51.
In this case, the ratio r includes ratios r 1 to r n obtained for each sensor when there are n biological signal sensors. (ST55)
Then, the operation reasoning device multiplies the sensor characteristic factor value F1 of each operation stored in the existing probability model setting process by the sensor ratios r 1 to r n , respectively, '). That is, F1 '= F1 * r *?. Here, β is a weight value set for each operation and / or sensor, as in α, and β = 1. (ST56)
Meanwhile, since the parameters of the existing probability model are calculated by using the feature parameter value (F1) of each motion and sensor, when the existing feature parameter value (F1) is changed, the existing probability model parameter And the probability model can be calibrated.
Therefore, in the second embodiment of the present invention, by using the sensor characteristic factor value F1 'of each operation calibrated from the existing characteristic factor value F1, The probability model is corrected by changing to the model parameter.
6 is a diagram showing how the normal distribution probability models for the three operations are corrected in different forms according to the noise magnitudes. (ST57)
When the probability model is corrected in this manner, the user action is inferred by comparing the feature value (F2) per sensor detected by the user operation with the corrected probability model in the operation monitoring mode. (ST58)
According to the second embodiment of the present invention, the feature factor value F2 for each sensor detected by the user operation is compared with the corrected probability model using the noise without comparing with the existing probability model, It is possible to prevent degradation of the accuracy of the signal.
Meanwhile, even when a new operation of the user is inputted after correcting the probability model according to the second embodiment of the present invention, the above two operation reasoning methods can be selectively performed. That is, as in the first embodiment, an operation can be deduced by comparing the corrected feature value F2 'with the probability model using the noise ratio r, or by correcting the probability model again as in the second embodiment It is possible to infer the motion by comparing the currently detected feature factor value F2 with the corrected probability model.
Hereinafter, an operation reasoning device capable of performing operation reasoning according to the above-described method will be described.
The
The plurality of
The signal processing means 110 performs signal processing such as filtering and amplification on the signals detected by the
The probability model generating means 120 generates a probability model for the feature parameter of the signal detected by each
The noise detection means 130 detects a first pause signal strength S1 corresponding to noise before or after a user's operation for training is input in the probability model generation mode and stores the detected first pause signal strength S1 in the storage means 190. [ Also, the second idle period signal S2 corresponding to the noise is detected and stored in the operation monitoring mode.
The noise ratio calculating means 132 calculates the ratio r of the second pause signal intensity S2 and the first pause signal intensity S1.
The calibration means 140 may include feature value calibration means and / or probability model calibration means. The feature value correcting means calculates the calibrated feature factor value F2 'by multiplying the sensor feature factor value F2 detected by the user operation in the operation monitoring mode by the
The
When the
The input unit 160 inputs a command of a user, and the type of the input unit 160 is not particularly limited. Therefore, a button, a keypad, a touch pad, a touch screen, a joystick, and the like can be used without limitation. It may also include speech recognition means including a microphone, voice analysis means, and the like.
The output means 170 provides the user with certain information visually or audibly, and may include at least one of a display, a light emitting LED, and a speaker.
The communication unit 180 performs wired or wireless communication between the
The implementation methods of the signal processing means 110, the probability model generating means 120, the noise ratio calculating means 130, the calibrating means 140, and the operation reasoning means 150 are not particularly limited, Or may be implemented by a combination of software and hardware (circuit, chip, module, etc.), or may be implemented only by hardware.
While the present invention has been described in connection with what is presently considered to be practical exemplary embodiments, it is to be understood that the invention is not limited to the disclosed embodiments, but, on the contrary,
For example, the type of the bio-signal used to deduce the operation of the user according to the embodiment of the present invention is not particularly limited. Therefore, the present invention can be applied to any type of detection apparatus or operation inference apparatus that infer an operation of a user using an EMG signal, an electroencephalogram signal, an electrocardiogram signal, a safety signal, or other types of bioelectric / magnetic signals.
In addition, the types of probability models that are the basis of the operation reasoning are not limited to the normal distribution probability models, and the correction method according to the present invention is applicable to the Poisson distribution probability model, the gamma distribution probability model, Can be applied.
It is to be understood that the present invention may be embodied in many other specific forms without departing from the spirit or essential characteristics thereof, and it is to be understood that the invention is not limited to the specific embodiments thereof except as defined in the appended claims.
10: sensor 100: motion inference device
110: Signal processing means 120: Probability model generating means
130: Noise detecting means 132: Noise ratio calculating means
140: calibration means 150: operation reasoning means
160: input means 170: output means
180: communication means 190: storage means
192: control means
Claims (9)
A probability model generation step of generating a probability model for each operation corresponding to the user operation command, detecting and storing the first pause signal strength (S1);
Detecting a second pause signal strength (S2) in an operation monitoring mode;
(R) of the first pause signal intensity (S1) and the second pause signal strength (S2), and corrects the correction value using the feature factor value (F2) detected by the user operation and the ratio ≪ / RTI >
Estimating the corrected feature value (F2 ') against the stored probability model
A method of calibrating a bio-signal including a signal to infer an action
Wherein the ratio r is S2 / S1, and the calibrated characteristic factor value F2 'is calculated by multiplying the characteristic factor value F2 by an inverse number of the ratio r. How to infer an action
Wherein the calibrated characteristic factor value (F2 ') is calculated by multiplying the characteristic factor value (F2) by the reciprocal of the ratio (r) and the weighting factor.
Detects and stores a feature parameter value (F1) for each sensor corresponding to each user action command, generates a probability model parameter using the feature parameter value (F1), and generates a first pause signal strength (S1) A probability model generation step of detecting and storing the probability model;
Detecting a second pause signal strength (S2) in an operation monitoring mode;
(R) of the first pause signal intensity (S1) and the second paused signal intensity (S2), and calculates a corrected characteristic factor value (r) by using the characteristic factor value (F1) F1 ');
A probability model calibration step of correcting the stored probability model using the corrected feature value F1 ';
Inferring an operation by comparing the feature value (F2) detected by the user operation with the corrected probability model
A method of calibrating a bio-signal including a signal to infer an action
Wherein the ratio r is S2 / S1, and the calibrated characteristic factor value F1 'is calculated by multiplying the characteristic factor value F1 by the ratio r. How to
Wherein the calibrated characteristic factor value (F1 ') is calculated by multiplying the characteristic factor value (F1) by the ratio (r) and the weighting factor.
Wherein the bio-signal sensor includes a plurality of sensors, and the ratio r is calculated on a sensor-by-sensor basis.
Probability model generating means for generating and storing a probability model for each sensor for each operation corresponding to a user operation command;
Noise detection means for detecting a first pause signal strength (S1) in a probability model generation mode and for detecting a second pause signal strength (S2) in a motion monitoring mode;
Noise ratio calculating means for calculating a ratio (r) of the first pause signal intensity (S1) and the second pause signal strength (S2);
Calibration means for calculating a feature factor value F2 'calibrated using the feature factor value F2 detected by the user's operation and the ratio r;
An operation inference means for inferring an operation by comparing the corrected feature value (F2 ') with a stored probability model
Lt; RTI ID = 0.0 >
A probability model generating means for detecting and storing a feature value F1 for each sensor for each operation corresponding to a user operation command and generating and storing a probability model parameter using the feature value F1;
Noise detection means for detecting a first pause signal strength (S1) in a probability model generation mode and for detecting a second pause signal strength (S2) in a motion monitoring mode;
Noise ratio calculating means for calculating a ratio (r) of the first pause signal intensity (S1) and the second pause signal strength (S2);
The feature parameter value F1 'corrected using the feature parameter value F1 and the ratio r is calculated and the probability model parameter is updated using the corrected feature parameter value F1' Correcting means for correcting the model;
An operation inference means for inferring an operation by comparing the feature value (F2) detected by the user operation with the corrected probability model
Lt; RTI ID = 0.0 >
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
KR1020150092159A KR101613055B1 (en) | 2015-06-29 | 2015-06-29 | Method of motion reasoning by adjusting biological signal using noise and motion reasoning device for the same |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
KR1020150092159A KR101613055B1 (en) | 2015-06-29 | 2015-06-29 | Method of motion reasoning by adjusting biological signal using noise and motion reasoning device for the same |
Publications (1)
Publication Number | Publication Date |
---|---|
KR101613055B1 true KR101613055B1 (en) | 2016-04-19 |
Family
ID=55917117
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
KR1020150092159A KR101613055B1 (en) | 2015-06-29 | 2015-06-29 | Method of motion reasoning by adjusting biological signal using noise and motion reasoning device for the same |
Country Status (1)
Country | Link |
---|---|
KR (1) | KR101613055B1 (en) |
-
2015
- 2015-06-29 KR KR1020150092159A patent/KR101613055B1/en active IP Right Grant
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP5432989B2 (en) | How to find the desired state in a subject | |
WO2018214529A1 (en) | Electromyographic signal acquisition method and apparatus | |
US10842402B2 (en) | Determination system, control signal output system, rehabilitation system, determination method, control signal output method, and recording medium | |
JP6190466B2 (en) | Biological signal measuring instrument and contact state estimation method | |
CN111603673A (en) | Adjusting method of neck massager and neck massager | |
KR20050117253A (en) | Electric stimulus generating device and method, and health training/monitoring apparatus and method using it | |
KR20150060304A (en) | Wearable mobile devices, and method for using selective biological signals by wearable mobile devices | |
US20180333056A1 (en) | Apparatus for monitoring cardiovascular health | |
US20160199203A1 (en) | Determination system, control signal output system, rehabilitation system, determination method, control signal output method, recording medium, and electroencephalogram signal acquisition system | |
EP2294977B1 (en) | Biological Signal Processing Apparatus and Medical Apparatus Controlling Method | |
WO2018105447A1 (en) | Contact state estimating device, and biological signal measuring device | |
AU2015340372A1 (en) | Method and apparatus for monitoring the pelvic floor muscles | |
JP2009297059A (en) | Brain training support apparatus | |
CN109890276B (en) | Blood pressure monitoring method, device and equipment | |
JP2009285294A (en) | Brain training support apparatus | |
JPWO2019078328A1 (en) | EEG measurement systems, EEG methods, programs, and non-temporary recording media | |
KR101613055B1 (en) | Method of motion reasoning by adjusting biological signal using noise and motion reasoning device for the same | |
KR20050063618A (en) | Method and appratus for recognizing positive/negative intention using fine change of gamma wave of brain wave | |
JP6960619B2 (en) | EEG judgment system | |
CN111973183A (en) | Joint measurement device and method for muscle fatigue and artificial limb | |
CN102283638A (en) | Blood pressure measuring method adopting intelligent coefficient matching | |
US20220280132A1 (en) | Medical device and program | |
CN105326482A (en) | Device and method for recording physiological signal | |
JP2002125945A (en) | Method and equipment for using brain wave by biological feedback | |
JP6888789B2 (en) | EEG measurement system, EEG measurement method, program, and non-temporary recording medium |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
GRNT | Written decision to grant | ||
FPAY | Annual fee payment |
Payment date: 20190418 Year of fee payment: 4 |