KR101025510B1 - Personal optimization system of emotional recognition device and its optimization method - Google Patents

Abstract

The present invention relates to an individual optimization system and method of the emotion recognition device using a bio-signal, and to be optimized for the individual user to exhibit more accurate and sophisticated emotion recognition performance. To this end, the present invention, the measurement unit for measuring the pulse wave (PPG) and skin conductivity (GSR); Emotion recognition unit for extracting a plurality of emotional feature variables for predicting heart rate variability (HRV), respiration, pulse velocity (PWV), tension from the bio-signals, and by performing individual optimization; It provides a personal optimization system and method comprising a; biometric information storage unit and display unit and error removal unit for collecting, storing and displaying the optimized emotion characteristic variable and emotion recognition result optimized for each individual, and remove the error. By using this system to accurately measure the emotional state of users, medical field (depression, mood swings, chronic stress prevention), industrial field (work fatigue, risk accident prevention, drowsiness driving prevention), education field (improving concentration of learning), entertainment field In interactive games, virtual reality, etc., the utility of products and services can be improved.

Vital signs, emotional recognition, heart rate variability (HRV), respiration, pulse wave (PPG), skin conductivity (GSR)

Description

Individual optimization system of emotion recognition device and its optimization method {Individual optimization system of recognizing emotion apparatus, method

The present invention relates to an individual optimization system and method of the emotion recognition device, and more particularly, by selectively applying only the characteristic signals suitable for a user through an individual optimization process from among a plurality of emotion recognition feature signals extracted from the biosignal. The present invention relates to a personal optimization system and a method of an emotion recognition device capable of exhibiting accurate and sophisticated emotion recognition performance for an individual user.

Emotional engineering aims to design products that appeal to emotion with comfort, stability, and high-grade feelings by specifically designing human emotions into physical design elements.

The most basic in emotional engineering is the technology to measure human emotion. Currently, methods such as facial expression analysis, eye movement analysis, emotional language analysis and human biosignal analysis are used.

Among them, the method of performing emotional recognition based on bio-signals is difficult to generalize due to inaccurate judgment performance, and the physical technology for solving this is advanced, but the quality service technology for humans is still Insufficient

Such emotion recognition technology is expected to be used as a core technology for equipment developed around humans in the ubiquitous era that is developing recently. In response, researches on emotion recognition technology are actively being conducted in domestic and overseas research institutes or universities, and most of these emotion recognition technologies exhibit low accuracy.

On the other hand, the physiological response signal according to the same emotional stimulus also because everyone can not represent the same, for a more accurate emotional recognition is a situation that requires an emotion recognition algorithm that can be optimized for individual users.

The technical problem to be achieved by the present invention is to measure only bio-signals including pulse wave (PPG) and skin conductivity (GSR), extract a plurality of emotional feature variables required for emotion recognition, and optimize each individual among a plurality of emotional feature variables By selectively applying only the appropriate emotional feature variables for use, it can be used as a characteristic variable for individual user's emotional recognition, so that it can show the optimized emotional recognition performance for individual users even using a small amount of bio feature data. To provide a personal optimization system and method of the emotion recognition device using a bio-signal.

Another technical problem to be achieved by the present invention is to continuously modify or delete the emotional feature variables reduced and selected for the individual optimization through an error elimination process by external input, thereby creating a continuously evolving emotional recognition system It is an object of the present invention to provide an individual optimization system and method for sensibility recognition apparatus using bio-signals that can be implemented.

One embodiment of the present invention for achieving the above object is a measuring unit for measuring a bio-signal including a pulse wave (PPG) and skin conductivity (GSR) of the individual user; Extracting a plurality of emotional feature variables for heart rate variability (HRV), respiration, pulse wave velocity (PWV), and tonicity from the bio signals measured by the measurement unit, and evaluates the performance of the extracted plurality of emotional feature variables Emotion recognition unit for recognizing the individual emotional state by the optimized emotional feature variables by performing individual optimization by reducing the individual feature dimension; A biometric information storage unit which is accessed by the emotion recognition unit and collects and stores the personalized emotion characteristic variables and emotion recognition results which are processed and recognized by the emotion recognition unit; And a display unit for displaying the personalized emotion characteristic variables and the emotion recognition result recognized by the emotion recognition unit.

Another embodiment of the present invention for achieving the above object is, in addition to the configuration of the system of the present invention, connected to the emotion recognition unit, the modification or deletion of the emotion characteristic parameters and emotion recognition results optimized by the emotion recognition unit Error correction unit for providing an external input signal for the; further comprising, personal optimization system of the emotion recognition device using a biological signal.

Another embodiment of the present invention for achieving the above object is a bio-signal measuring step of measuring a bio-signal including a pulse wave (PPG) and skin conductivity (GSR) from the individual user; From the measured pulse wave (PPG), skin conductance (GSR) signal, a plurality of emotional feature variables for heart rate variability (HRV), respiration, pulse wave rate (PWV), tension prediction are extracted, and the extracted plurality of emotional feature variables An emotional recognition step of recognizing the emotional state of the individual by the optimized emotional feature variables by performing individual optimization through performance evaluation and individual feature dimension reduction for each field; An individual optimization method of an individual using a bio-signal, comprising: an individual's emotional feature storage and display step of collecting, storing and displaying an individual's optimized emotional feature variable and an emotional recognition result processed and recognized through the emotional recognition step to be.

Another embodiment of the present invention for achieving the above object is, in addition to the configuration of the method of the present invention, the external for modifying or deleting individual emotional specific variables and emotional recognition results processed and recognized through the emotional recognition step Error correction step of providing an input signal; Personal optimization method of the emotional recognition device using a biological signal.

According to the individual optimization system and method of the emotional recognition device using the bio-signal according to the present invention, by using the emotional recognition information to accurately measure the emotional state of the user in the medical field (depression, mood swings, chronic stress prevention), industrial field (work Fatigue, dangerous accident prevention, drowsiness driving prevention), education (improving concentration of learning), entertainment (interactive games, virtual reality), etc. can increase the effectiveness of products and services.

Hereinafter, the configuration and operation according to an embodiment of the present invention will be described in detail with reference to the accompanying drawings. The terms or words used in this specification and claims are not to be construed as limiting in their usual or dictionary meanings, and the inventors may properly define the concept of terms in order to best explain their invention in the best way possible. It should be interpreted as meaning and concept corresponding to the technical idea of the present invention. Therefore, since the embodiments shown in the specification and the configuration shown in the drawings is only one of the most preferred embodiment of the present invention, it is understood that there may be various equivalents and modifications that can replace them at the time of the present application shall.

1 is a block diagram showing an embodiment of an individual optimization system of an emotion recognition apparatus using a biosignal according to the present invention. As shown in the drawing, the present invention includes a measurement unit 10 and an emotion recognition unit 20. The personal information optimization system of the emotion recognition device including the biometric information storage unit 30 and the display unit 40, or the personal optimization system of the emotion recognition device further including an error correction unit 50.

The measurement unit 10 measures a user's bio-signal, that is, pulse wave (PPG) and skin conductivity (GSR), and includes a sensor unit 11 and a preprocessor 12.

The sensor unit 11 includes one or more sensors to measure bio signals such as pulse wave (PPG) and skin conductivity (GSR). The sensor unit 11 provides audiovisual materials for emotional stimulation to the experimenter, and the sensor unit 11 preferably uses a sensor of a commercial product and attaches each sensor to a palm or a finger to measure the sensor.

The preprocessor 12 amplifies the measured biosignal and extracts only the pure biosignal by performing filtering to remove noise.

The emotion recognition unit 20 includes an emotion feature extraction unit 21, a feature dimension reduction unit 22, and an emotion determination recognition unit 23. The heart rate variability is measured from the bio signals measured by the measurement unit 10. (HRV), respiration, pulse wave velocity (PWV), and a variety of emotional feature variables for the prediction of tension (arousal), and the performance evaluation of the extracted emotional feature variables and individual feature dimension reduction By performing the optimization, the individual emotional state is recognized by the optimized emotional feature variables.

Emotional feature extraction unit 21 from the pulse wave (PPG) signal and skin conductance (GSR) signal measured by the measuring unit 10, heart rate variability (HRV), respiration, pulse wave rate (PWV), tension (the degree of awakening for each individual) ) Extract many emotional feature variables for prediction.

The emotional feature extractor 21 filters down-sampling of the pulse wave (PPG) signal and filters and Fourier transforms within a predetermined band (BPF, for example, a frequency range of 0.13 to 0.48 Hz). ), A number of emotional feature variables, dominant frequency estimation (hereinafter, referred to as 'DFE') and entropy, are extracted.

In addition, the emotional feature extractor 21 detects peak values of the pulse wave (PPG) signal and calculates a plurality of peaks for HRV, blood vessel contraction and blood pressure, and pulse wave velocity (PWV) from the peak value detection results. Emotional feature variables are extracted.

Emotional feature variables for HRV predictions include HRV mean, standard deviation of the normal-to-normal intervals (SDNN), and root mean square of the successive differences between consecutive NN intervals (HRV time-domain). index quantifying the parasympathetic tone), NN50 (number of interval differences of successive NN intervals greater than 50ms), total power, nHF (high frequency band), nLF (low frequency band), LF / HF (radio high / low frequency band These feature variables are frequency interpolation (e.g., interpolation of 4 Hz), time domain HRV, and frequency domain analysis of the RR interval between peaks of a pulse wave (PPG) signal. HRV).

Emotional feature variables for the prediction of blood vessel contraction, blood pressure, and pulse wave velocity (PWV) are mean values and standard deviations, and these feature variables are the predictions of the peak amplitude and pulse wave velocity (PWV) detected from pulse wave (PPG) signals. Is extracted through.

In addition, the emotional feature extracting unit 21 extracts a plurality of emotional feature variables that can predict the arousal level of each individual using the gradient technique and the magnitude of the Y-axis variation from the skin conductivity (GSR) signal. These feature variables include down-sampling, noise reduction filtering, phasic detection, mean and standard deviation of skin conductance (GSR) signals, frequency detection through yonic detection, and derivatives. Count value.

The feature dimension reduction unit 22 performs a performance evaluation on the plurality of emotional feature variables to reduce the feature dimension to an area including variables having excellent evaluation performance. In this case, performance evaluation of the plurality of emotional feature variables can be performed through a sequential floating forward search algorithm, and as a result, an area containing variables having excellent evaluation performance is reduced, and the feature dimension for individual optimization is reduced. Run

The emotional determination recognizer 23 processes the emotional feature variables in the feature dimension selected by the feature dimension reduction unit 22 according to a predetermined emotion pattern recognition rule to determine and recognize the current emotional state of each individual. Here, the emotional feature variables in the selected feature dimension are used as input signals to the neural network algorithm learned according to the emotion pattern recognition rule, and generate an output of the emotional state judgment result of each individual user through one or more hidden layers. The emotional state for each individual may be distinguished, for example, by fear, sadness, joy, or the like.

The biometric information storage unit 30 is accessed by the emotion recognition unit 20, and collects and stores the personalized emotion characteristic variables and emotion recognition results optimized and extracted through the emotion recognition unit 20.

The display unit 40 displays the optimized emotion feature variable and the emotion recognition result for each person extracted and recognized by the emotion recognition unit 20. In this display unit, a change in a biosignal based on the recognized plurality of emotional states can be displayed and displayed as a graph, a numerical value, and an image.

The error correction unit 50 is connected to the emotion recognition unit 20 and used as an interface means between the external user and the system for error removal, and the emotion characteristic variable and the emotion recognition result optimized by the emotion recognition unit 20. The external input signal for the correction or deletion is provided to the emotion recognition unit. As a result, the emotional feature variables reduced and selected for the individual optimization can be selectively modified or deleted through an error elimination process by external input, thereby enabling the implementation of a continuously evolving emotional recognition system.

FIG. 2 is a flowchart illustrating an individual optimization method of an emotion recognition device using a biosignal according to the present invention, and includes a biosignal measurement step (S101 and S103) and an emotion recognition step (S105 to S109). The personal optimization method of the emotion recognition device including a step (S111), or the individual optimization method of the emotion recognition device further includes an error correction step (S113-S117).

In the biosignal measurement step, a biosignal including pulse wave (PPG) and skin conductivity (GSR) is measured from an individual user (S101), amplifying the measured biosignal, and performing filtering to remove noise. It comprises a step (S103) of pre-processing the biometric information.

In the emotion recognition step, extracting a plurality of emotional feature variables for predicting heart rate variability (HRV), respiration, pulse wave velocity (PWV), and tension from the measured and preprocessed pulse wave (PPG) and skin conductivity (GSR) signals ( S105), performing personal optimization by performing performance evaluation on the plurality of extracted emotional feature variables and reducing individual feature dimensions (S107), and recognizing the individual emotional state using the optimized emotional feature variables. (S109).

The emotional feature variable extraction is performed through sampling and peak value detection of a pulse wave (PPG) signal and a skin conductivity (GSR) signal. That is, DFE and entropy for respiratory prediction are extracted through down sampling of the pulse wave (PPG) signal and filtering and Fourier transform in a predetermined band (for example, BPF in the range of 0.13 to 0.48 Hz). In addition, after detecting peak values of the pulse wave (PPG) signal, HRV mean, SDNN, RMSSD, and NN50 are extracted by interpolating and time-domain analysis of a specific frequency (4 Hz) for the interval (RR interval) between the peak values. In addition, total power, nHF, nLF, and LF / HF are extracted through frequency domain analysis of the interval (RR interval) between the peak values subjected to the specific frequency interpolation. The standard deviation of the peak amplitude of the pulse wave (PPG) signal is extracted to extract variables (mean and standard deviation) for predicting blood vessel contraction and blood pressure, and between the maximum and minimum peak values of the pulse wave (PPG) signal. The intervals are used to extract variables (mean and standard deviation) for the prediction of pulse wave velocity (PWV).

In addition, the skin conductance (GSR) signal extracts variables (average value, standard deviation, frequency, etc.) for predicting each individual's arousal degree using the gradient technique and the magnitude of the Y-axis variation.

On the other hand, in the individual optimization process, the performance evaluation of the plurality of emotional feature variables is performed through a sequential floating forward search algorithm to reduce the feature dimension to an area including variables having excellent evaluation performance. Emotional feature variables in the selected feature dimension are inputted to a neural network algorithm learned according to a predetermined emotion pattern recognition rule to determine the current emotional state of each individual.

The step of storing and displaying an individual emotional feature is a step of collecting, storing and displaying an individual optimized emotional feature variable and an emotional recognition result processed and recognized through the emotional recognition step (S111). The display operation may be performed by outputting a change in a biosignal based on the recognized emotional state as a graph, a numerical value, and an image.

The error correction step may include checking whether an external signal input by the user is received (S113), and when the external signal is received as a result of the checking, correcting or deleting the personal emotional characteristic variable and the recognition result according to the input signal. And a step (S117) of storing and displaying a result of the modification or deletion of the individual emotion-specific variable and the emotion recognition result.

Referring to the operation and effect of the individual optimization system and method of the present invention configured as described above are as follows.

First, the measurement unit 10 measures a biosignal including pulse wave PPG and skin conductivity GSR from the user through the sensor unit 11 (S101), and measures the biosignal measured by the preprocessor 12. The signal is amplified by a predetermined level and filtered to remove noise (S103) to detect only a pure biosignal.

Next, the emotion recognition unit 20 extracts a plurality of emotion feature variables through the emotion feature extraction unit 21 (S105). That is, DFE and entropy for respiratory prediction are extracted through down sampling of the pulse wave (PPG) signal measured and preprocessed by the measurement unit 10, band filtering in the range of 0.13 to 0.48 Hz, and Fourier transform. The HRV mean, SDNN, RMSSD, and NN50 are extracted through specific frequency (4Hz) interpolation and time domain analysis of the interval between peaks of the (PPG) signal. In addition, total power, nHF, nLF, and LF / HF are extracted through frequency domain analysis of the interval (RR interval) between peak values subjected to 4 Hz frequency interpolation. In addition, the standard deviation of the peak amplitude of the pulse wave (PPG) signal is extracted to extract variables (mean and standard deviation) for predicting blood vessel contraction and blood pressure, and between the maximum and minimum peak values of the pulse wave (PPG) signal. The intervals are used to extract variables (mean and standard deviation) for the prediction of pulse wave velocity (PWV). In addition, the emotional feature extractor 21 of the emotion recognizer 20 may predict variables of the individual's arousal degree using the gradient technique and the magnitude of the Y-axis variation from the skin conductance (GSR) signal (average value and standard deviation and Frequency, etc.).

Next, the emotion recognition unit 20 performs an individual optimization process through the feature dimension reduction unit 22. In other words, by performing a performance evaluation on the plurality of emotional feature variables through a sequential floating forward search algorithm (S107) by reducing the feature dimension to the area containing the variables having excellent evaluation performance, The optimization is performed only with the emotional feature variables within the feature dimension selected to be suitable for the application. The emotion characteristic variables optimized as described above are input to the neural network algorithm mounted in the emotion determination recognition unit 23, and thus, the emotion determination recognition unit 23 uses the neural network algorithm learned according to a predetermined emotion pattern recognition rule. It is possible to output the current emotional state judgment and recognition results of each individual.

The emotion recognition result and the emotional feature variables in the feature dimension reduced through the individual optimization process are collected and stored in the biometric information storage unit 30, and the change of the biosignal based on the plurality of recognized emotional states. Is output as a graph, a numerical value, and an image through the display unit 40.

As described above, the emotion recognition results stored and displayed in the biometric information storage unit 30 and the display unit 40 and the emotion feature variables selected through the individual optimization process are externally input by the user through the error correction unit 50. It can be modified or deleted by the signal, so that the emotional feature variables selected through the optimization process for each individual can be selectively modified and corrected as necessary or even when an error occurs with variables that are not suitable for each individual. The error elimination process allows for continuous evolution.

In addition, the result modified or deleted by the user may also be displayed (S117) through the biometric information storage unit 30 and the display unit 40.

As described above, the present invention has been described by way of limited embodiments and drawings, but the present invention is not limited to the above-described embodiments, which can be variously modified and modified by those skilled in the art to which the present invention pertains. Modifications are possible. Accordingly, the spirit of the present invention should be understood only by the claims set forth below, and all equivalent or equivalent modifications thereof will belong to the scope of the present invention.

1 is a block diagram showing a configuration example of an individual optimization system of the emotion recognition device using a bio-signal according to the present invention.

2 is an operation flowchart showing an example of the configuration of the individual optimization method of the emotion recognition device using a bio-signal according to the present invention.

<Description of the symbols for the main parts of the drawings>

10: measuring unit 11: sensor unit

12: preprocessing unit 20: emotion recognition unit

21: emotional feature extraction section 22: feature dimension reduction section

23: emotional judgment recognition unit 30: biometric information storage unit

40: display unit 50: error correction

Claims (23)

A measuring unit measuring a biosignal including a pulse wave (PPG) and a skin conductivity (GSR) of an individual user; Extracting a plurality of emotional feature variables for heart rate variability (HRV), respiration, pulse wave velocity (PWV), and tonicity from the bio signals measured by the measurement unit, and evaluates the performance of the extracted plurality of emotional feature variables Emotion recognition unit for recognizing the individual emotional state by the optimized emotional feature variables by performing individual optimization by reducing the individual feature dimension; A biometric information storage unit which is accessed by the emotion recognition unit and collects and stores the personalized emotion characteristic variables and emotion recognition results which are processed and recognized by the emotion recognition unit; And a display unit for displaying the optimized emotion characteristic variable and the emotion recognition result for each person recognized by the emotion recognition unit. A measuring unit measuring a biosignal including a pulse wave (PPG) and a skin conductivity (GSR) of an individual user; From the bio signals measured by the measurement unit extracts a plurality of emotional feature variables for predicting heart rate variability (HRV), respiration, pulse wave velocity (PWV), tension, performance evaluation of the extracted plurality of emotional feature variables and Emotion recognition unit for recognizing the individual emotional state by the optimized emotional feature variables by performing individual optimization by reducing the individual feature dimension; A biometric information storage unit which is accessed by the emotion recognition unit and collects and stores the optimized emotion feature variables and emotion recognition results for each individual processed and recognized through the emotion recognition unit; A display unit for displaying an optimized emotion characteristic variable and an emotion recognition result for each person recognized through the emotion recognition unit; An error correction unit connected to the emotion recognizing unit and providing an external input signal for correcting or deleting the emotion characteristic variable and the emotion recognition result optimized by the emotion recognizing unit; Personal optimization system of recognition device. The measuring unit according to any one of claims 1 and 2, wherein At least one sensor unit configured to measure a biosignal including pulse wave (PPG) and skin conductivity (GSR); And a pre-processing unit for amplifying the measured bio-signals and performing filtering to remove noise, thereby extracting only the pure bio-signals. The emotion recognition unit according to claim 1 or 2, An emotional feature extraction unit for extracting a plurality of emotional feature variables for predicting heart rate variability (HRV), respiration, pulse wave velocity (PWV), and tension from the PPG signal and the GSR signal measured by the measurement unit; A feature dimension reduction unit configured to reduce a feature dimension to an area including variables having excellent evaluation performance by performing performance evaluation on the plurality of emotional feature variables; And an emotional judgment recognizing unit configured to determine and recognize a current emotional state of each individual by signal processing the emotional feature variables selected in the feature dimension selected by the feature dimension reduction unit according to a preset pattern recognition rule. Personal optimization system of emotion recognition device using bio signals. The method of claim 4, wherein the emotional feature extraction unit, And a plurality of emotional feature variables for respiratory prediction through down sampling of the pulse wave (PPG) signal, filtering in a predetermined band, and Fourier transform. The method of claim 5, wherein the emotional feature extraction unit, Detecting peak values of the pulse wave (PPG) signal, and extracting a plurality of emotional feature variables for HRV prediction through time domain analysis and frequency domain analysis of the RR intervals between the peak values. Personal optimization system of emotion recognition device using a biological signal characterized in that. The method of claim 6, wherein the emotional feature extraction unit, Emotion using a biosignal, characterized in that it detects a peak value of the pulse wave (PPG) signal and extracts a plurality of emotional feature variables for predicting blood vessel contraction and blood pressure value by predicting a standard deviation of the amplitude of the pulse wave (PPG) signal. Personal optimization system of recognition device. The method of claim 7, wherein the emotional feature extraction unit, Detecting a peak value of the pulse wave (PPG) signal, and extracting a plurality of emotional feature variables for predicting the pulse wave velocity (PWV) using the interval between the maximum value and the minimum value of the detected peak value. Personal optimization system of emotion recognition device using. The method of claim 8, wherein the emotional feature extraction unit, Personal optimization of the emotion recognition device using biosignals, which extracts a plurality of emotional feature variables that can predict the degree of arousal of each individual using the gradient technique and the magnitude of the Y-axis variation from the skin conductance (GSR) signal. system. The method of claim 4, wherein the feature dimension reduction unit, The performance evaluation of the plurality of emotional feature variables is performed through a sequential floating forward search algorithm, which includes variables having excellent evaluation performance, and performs feature dimension reduction for individual optimization. Personal optimization system of emotion recognition device using bio signals. The method of claim 4, wherein the emotion recognition determination unit, Applying the emotional feature variables in the selected feature dimension to the neural network algorithm learned by the emotion pattern recognition rule, to determine and recognize a plurality of emotional state for each individual, each individual in the emotional recognition device using a bio-signal Optimization system. The method of claim 11, wherein the display unit, Personal optimization system of the emotion recognition device using a bio-signal, characterized in that for displaying the change in the bio-signal based on the recognized plurality of emotion state in a graph, a numerical value and an image. A biosignal measuring step of measuring a biosignal including a pulse wave (PPG) and a skin conductivity (GSR) from an individual user; From the measured pulse wave (PPG) signal and skin conductivity (GSR) signal, a plurality of emotional feature variables for heart rate variability (HRV), respiration, pulse wave rate (PWV), and tension prediction are extracted, and the extracted plurality of emotional features An emotional recognition step of recognizing the emotional state of the individual by the optimized emotional feature variables by performing individual optimization through performance evaluation of the variables and reduction of the individual feature dimension; An emotional recognition device using a bio-signal, comprising: an individual emotional feature storage and display step of collecting, storing, and displaying an optimized emotional feature variable and an emotion recognition result for each individual processed and recognized through the emotion recognition step Personal optimization method. The method of claim 13, An error correction step of providing an external input signal for the correction or deletion of the individual emotional feature variable and the emotional recognition result processed and recognized through the emotional recognition step; of the emotional recognition device using a biological signal further comprising Personal optimization method. The method of claim 13 or 14, wherein the emotion recognition step, Emotional feature extraction step of extracting a plurality of emotional feature variables for predicting heart rate variability (HRV), respiration, pulse wave velocity (PWV), tension through sampling and peak value detection of the pulse wave (PPG) signal and skin conductance (GSR) signal ; A feature dimension reduction step of performing a performance evaluation on the plurality of emotional feature variables to reduce a feature dimension to an area including variables having excellent evaluation performance; And an emotional judgment recognition step of determining and recognizing the current emotional state of each individual by processing the emotional feature variables selected in the feature dimension selected through the feature dimension reduction step according to a preset emotion pattern recognition rule. Personal optimization method of the emotion recognition device using a bio-signal. The method of claim 15, wherein extracting the emotional feature, And optimizing a plurality of emotional feature variables for respiratory prediction through down sampling of the pulse wave (PPG) signal, filtering in a predetermined band, and Fourier transform. The method of claim 16, wherein extracting the emotional feature, Detecting peak values of the pulse wave (PPG) signal, and extracting a plurality of emotional feature variables for HRV prediction through time domain analysis and frequency domain analysis of the RR intervals between the peak values. Personal optimization method of the emotion recognition device using a biological signal characterized in that. 18. The method of claim 17, wherein extracting the emotional features, Emotion using a biosignal, characterized in that it detects a peak value of the pulse wave (PPG) signal and extracts a plurality of emotional feature variables for predicting blood vessel contraction and blood pressure value by predicting a standard deviation of the amplitude of the pulse wave (PPG) signal. Personal optimization method of recognition device. 19. The method of claim 18, wherein the emotional feature extraction step, Detecting a peak value of the pulse wave (PPG) signal, and extracting a plurality of emotional characteristic variables for predicting the pulse wave velocity (PWV) using a distance between the maximum value and the minimum value of the detected peak value. Individual optimization method of emotion recognition device using. The method of claim 19, wherein the emotional feature extraction step, Personal optimization of the emotion recognition device using biosignals, which extracts a plurality of emotional feature variables that can predict the degree of arousal of each individual using the gradient technique and the magnitude of the Y-axis variation from the skin conductance (GSR) signal. Way. The method of claim 16, wherein the reducing the feature dimension, The performance evaluation of the plurality of emotional feature variables is performed through a sequential floating forward search algorithm, which includes variables having excellent evaluation performance, and performs feature dimension reduction for individual optimization. Personal optimization method of the emotion recognition device using a bio-signal. The method of claim 16, wherein the emotion recognition step, By applying the emotional feature variables in the selected feature dimension to the neural network algorithm learned by the emotion pattern recognition rule, a plurality of emotional states for each individual are determined and recognized. Optimization method. The method of claim 22, wherein the storing and displaying step, Personal optimization method of the emotion recognition device using a bio-signal, characterized in that for outputting the change in the bio-signal based on the recognized plurality of emotion state as a graph, a numerical value and an image.

Images