Odor perception and emotion recognition fusion analysis method and system, terminal and medium

Abstract

The invention discloses a fusion analysis method and system for smell perception and emotion recognition, a terminal and a medium, and relates to the technical field of sensory analys.The method comprises the steps that when smell output equipment outputs gas with smell to the nasal cavity of a user, a physiological signal is obtained, the physiological signals comprise any one or more of electroencephalogram signals, skin electric signals and electrocardiosignals; determining odor type information and emotional state information based on the physiological signal, and establishing an odor perception dimension space model to graphically present the relationship among the odor type information, the emotional state information and the physiological signal; and determining a smell recognition result based on the smell type information, and outputting an emotion recognition result based on the emotion state information. According to the method, the accuracy and the stability of odor-induced emotional state recognition are improved, joint decoding of odor and emotional response can be realized, and meanwhile, comprehensive and reliable odor sensory evaluation technical support can be provided for the food industry.

Description

Technical Field

[0001] The present invention relates to the field of sensory analysis technology, and in particular to a method, system, terminal and medium for integrating odor perception and emotion recognition. Background Art

[0002] Food odor is a key indicator for food sensory evaluation, quality control, and authenticity verification. Existing food odor assessment primarily relies on manual sensory evaluation and biomimetic sensing technologies such as electronic noses. Manual sensory evaluation relies on human perception and verbal expression, which is subject to high subjectivity and difficulty in quantification, making it difficult to meet the needs of large-scale industrial testing.

[0003] While existing machine perception technologies like electronic noses can quickly detect odor components, they are limited by the type, quantity, and sensitivity of sensors, resulting in insufficient ability to identify complex, weak, or mixed odors. Furthermore, they fail to reflect human emotional responses to odors. Consequently, these existing technologies struggle to provide high-precision, comprehensive, and standardized technical support for food odor identification, sensory quality evaluation, and personalized product development.

[0004] Therefore, the prior art still has defects. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to address the above-mentioned shortcomings of the prior art and provide a method, system, terminal and medium for the fusion analysis of taste perception and emotion recognition. The technical solution adopted by the present invention is as follows: In a first aspect, the present invention provides a method for integrating odor perception and emotion recognition, wherein the method comprises: When the odor output device outputs the odor-bearing gas to the user's nasal cavity, a physiological signal is obtained, wherein the physiological signal includes any one or more of an electroencephalogram signal, a skin electrical signal, and an electrocardiogram signal; Determining odor type information and emotional state information based on the physiological signal, and establishing an odor perception dimensional space model to graphically present the relationship between the odor type information, the emotional state information, and the physiological signal; An odor recognition result is determined based on the odor type information, and an emotion recognition result is output based on the emotional state information.

[0006] In one implementation, determining the odor type information and the emotional state information based on the physiological signal includes: Preprocessing the physiological signal, and performing feature extraction based on the preprocessed physiological signal to obtain target features; Based on the target features, odor type information and emotional state information are determined.

[0007] In one implementation, the preprocessing includes: bandpass filtering, power frequency interference removal, physiological artifact removal, and bad channel repair.

[0008] In one implementation, when the physiological signal is an electrocardiogram signal, extracting features based on the preprocessed physiological signal to obtain target features includes: Determining a heart rate variability characteristic value based on the preprocessed ECG signal, wherein the heart rate variability characteristic value includes: a root mean square difference, a low frequency power, and a high frequency power of the preprocessed ECG signal; The target feature is obtained based on the heart rate variability feature value.

[0009] In one implementation, determining the odor type information and the emotional state information based on the target feature includes: If the high-frequency power of the preprocessed electrocardiogram signal continues to decrease within a preset time period or if the ratio of the low-frequency power to the high-frequency power of the preprocessed electrocardiogram signal continues to decrease within a preset time period, determining that the odor type information is a sour odor or a bitter odor, and determining that the emotional state information is a high alert state; If the root mean square difference and high-frequency power of the preprocessed electrocardiogram signal continue to rise within a preset time period, the odor type information is determined to be a sweet odor or a fruity odor, and the emotional state information is determined to be a relaxed and happy state.

[0010] In one implementation, when the physiological signal includes an electroencephalogram (EEG) signal, a skin electrical signal, and an electrocardiogram (ECG) signal, extracting features based on the preprocessed physiological signal to obtain target features includes: Extracting the time domain features, frequency domain features, wavelet features and entropy features of each preprocessed physiological signal; Determine the optimal features for each pre-processed physiological signal based on their impact on odor recognition and emotion analysis; The target feature is obtained based on the optimal feature of each preprocessed physiological signal.

[0011] In one implementation, determining the optimal feature for each pre-processed physiological signal based on the degree of influence of the feature on odor recognition and emotion analysis includes: Calculate the influence scores corresponding to the time domain features, frequency domain features, wavelet features, and entropy features of each preprocessed physiological signal; Based on the influence degree scores, the feature with the highest influence degree score in each preprocessed physiological signal is determined, and the feature with the highest influence degree score in each preprocessed physiological signal is used as the optimal feature of each preprocessed physiological signal.

[0012] In one implementation, determining the odor type information and the emotional state information based on the target feature includes: Obtaining weight information of the optimal feature of each preprocessed physiological signal, and determining a target feature weight value based on the optimal feature of each preprocessed physiological signal and the corresponding weight information; Based on the weighted values ​​of the target features, the odor type information and the emotional state information are determined.

[0013] In one implementation, determining odor type information and emotional state information based on target feature weights includes: If the target feature weighted value is higher than a preset value, the odor type information is determined to be a sweet odor or a fruity odor, and the emotional state information is determined to be a relaxed and pleasant state; If the target feature weighted value is lower than a preset value, the odor type information is determined to be a sour odor or a bitter odor, and the emotional state information is determined to be a high alert state.

[0014] In one implementation, determining the odor type information and the emotional state information based on the target feature includes: Obtain a preset classification model, classify the target feature based on the classification model, and output odor type information and emotional state information corresponding to the target feature, wherein the classification model includes an integrated learning framework or a manifold classifier, and the emotional state information includes pleasure and arousal. When the pleasure is higher than the arousal, the emotional state information is a relaxed and pleasant state; when the pleasure is lower than the arousal, the emotional state information is a highly alert state.

[0015] In one implementation, establishing an odor perception dimensional space model includes: Mapping the odor type information, the emotional state information, and the target feature to establish an association relationship between the odor, emotion, and physiological signal; Based on the association relationship, an odor perception dimensional space model is established.

[0016] In one implementation, determining an odor recognition result based on the odor type information includes: matching the odor type information with identification information of the gas output by the odor output device to obtain a matching result; Based on the matching result, the odor recognition result is determined.

[0017] In one implementation, outputting an emotion recognition result based on the emotional state information includes: Obtain the user's self-evaluation of emotions regarding smells; An emotion recognition result is determined based on the emotion self-assessment information and the emotional state information.

[0018] In a second aspect, an embodiment of the present invention further provides a system for integrating odor perception and emotion recognition. The system is configured to implement the steps of the above-mentioned method for integrating odor perception and emotion recognition. The data processing and control device in the system includes: a physiological signal acquisition module, configured to acquire physiological signals when the odor output device outputs odorous gas to the user's nasal cavity, wherein the physiological signals include any one or more of an electroencephalogram signal, an electrical skin signal, and an electrocardiogram signal; an odor and emotion analysis module for determining odor type information and emotional state information based on the physiological signal, and establishing an odor perception dimensional space model to graphically present the relationship between the odor type information, the emotional state information, and the physiological signal; A result determination module is configured to determine an odor recognition result based on the odor type information, and output an emotion recognition result based on the emotional state information.

[0019] In a third aspect, an embodiment of the present invention further provides a terminal, wherein the terminal includes a memory, a processor, and a fusion analysis program of odor perception and emotion recognition stored in the memory and runnable on the processor. When the processor executes the fusion analysis program of odor perception and emotion recognition, the steps of the fusion analysis method of odor perception and emotion recognition of any one of the above-mentioned schemes are implemented.

[0020] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, wherein a fusion analysis program for odor perception and emotion recognition is stored on the computer-readable storage medium, and the fusion analysis program for odor perception and emotion recognition implements the steps of the fusion analysis method for odor perception and emotion recognition described in any one of the above-mentioned schemes on the computer-readable storage medium.

[0021] Beneficial Effects: Compared with the prior art, the present invention provides a fusion analysis method for odor perception and emotion recognition. First, when an odor output device outputs odorous gas into the user's nasal cavity, the present invention obtains physiological signals, which include any one or more of electroencephalogram (EEG) signals, skin conduction signals, and electrocardiogram (ECG) signals. Then, based on the physiological signals, odor type information and emotional state information are determined, and an odor perception dimensional space model is established to graphically present the relationship between the odor type information, the emotional state information, and the physiological signals. Finally, an odor recognition result is determined based on the odor type information, and an emotion recognition result is output based on the emotional state information. The present invention can achieve a comprehensive analysis of odor type information and emotional state information based on physiological signals, and judge the identified odor type information and emotional state information to obtain corresponding odor recognition results and emotion recognition results. This solves the problem of inaccurate results caused by subjective judgment of odor and emotion in the prior art, and improves the accuracy and stability of odor-induced emotional state recognition, which is conducive to the joint decoding of odor and emotional response. It also helps to provide comprehensive and reliable technical support for odor sensory evaluation for the food industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 A flowchart of a preferred embodiment of the fusion analysis method of odor perception and emotion recognition provided by an embodiment of the present invention.

[0023] Figure 2 This is a functional block diagram of the fusion analysis system for odor perception and emotion recognition provided by an embodiment of the present invention.

[0024] Figure 3 A schematic diagram of the structure of the EEG signal acquisition device in the fusion analysis system of odor perception and emotion recognition provided by an embodiment of the present invention.

[0025] Figure 4 This is a schematic diagram of the structure of the skin electrical signal acquisition device in the fusion analysis system of odor perception and emotion recognition provided by an embodiment of the present invention.

[0026] Figure 5 A schematic diagram of the structure of the central electrical signal acquisition device of the fusion analysis system for odor perception and emotion recognition provided by an embodiment of the present invention.

[0027] Figure 6 A schematic diagram of the structure of the odor output device in the fusion analysis system of odor perception and emotion recognition provided by an embodiment of the present invention.

[0028] Figure 7 The changing trends of the heart rate variability characteristic values ​​induced by three typical odors in the fusion analysis method of odor perception and emotion recognition provided by an embodiment of the present invention.

[0029] Figure 8 This is a principle block diagram of the data processing and control device in the fusion analysis system of odor perception and emotion recognition provided by an embodiment of the present invention.

[0030] Figure 9 This is a functional block diagram of a terminal provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0031] In order to make the purpose, technical solution and effect of the present invention clearer and more specific, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0032] The flowcharts shown in the accompanying drawings are for illustrative purposes only and do not necessarily include all contents, operations, or steps, nor must they be executed in the order described. For example, some operations or steps may be decomposed, combined, or partially merged, so the actual execution order may vary depending on the actual situation.

[0033] It should be understood that the terms used in this specification are for the purpose of describing particular embodiments only and are not intended to limit the present invention. As used in the specification and appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms unless the context clearly indicates otherwise. It should be understood that, to facilitate a clear description of the technical solutions of the embodiments of the present invention, in the embodiments of the present invention, terms such as "first" and "second" are used to distinguish between identical or similar items with substantially the same functions and effects. For example, the first control information and the second control information are merely used to distinguish different control information and do not limit their order. Those skilled in the art can understand that words such as "first" and "second" do not limit the quantity and execution order, and words such as "first" and "second" do not necessarily limit them to be different. It will also be understood that the term "and / or" used in the present description and appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0034] To address the problems of the prior art, this embodiment first provides a method for the fusion analysis of odor perception and emotion recognition. This method solves the problem of inaccurate results caused by subjective judgments of odor and emotion in the prior art, and improves the accuracy and stability of odor-induced emotional state recognition, facilitating the joint decoding of odor and emotional responses. It also helps provide comprehensive and reliable technical support for odor sensory evaluation in the food industry. In specific applications, this embodiment first acquires physiological signals when an odor output device outputs odorous gas into the user's nasal cavity. The physiological signals include any one or more of an electroencephalogram (EEG) signal, an electrical skin signal, and an electrocardiogram (ECG) signal. Then, based on the physiological signals, odor type information and emotional state information are determined, and an odor perception dimensional space model is established to graphically present the relationship between the odor type information, the emotional state information, and the physiological signals. Finally, an odor recognition result is determined based on the odor type information, and an emotion recognition result is output based on the emotional state information.

[0035] The fusion analysis method of odor perception and emotion recognition of this embodiment can be applied to a terminal that has the functions of data acquisition, transmission and analysis processing. The terminal can be an intelligent device such as a computer or a mobile phone. Figure 1 As shown in , the fusion analysis method of odor perception and emotion recognition in this embodiment includes the following steps: Step S100: When the odor output device outputs odorous gas to the user's nasal cavity, a physiological signal is obtained, where the physiological signal includes any one or more of an electroencephalogram signal, an electrical skin signal, and an electrocardiogram signal.

[0036] In specific applications, this embodiment provides a fusion analysis system for odor perception and emotion recognition, such as Figure 2As shown in , the system includes: a multimodal physiological signal acquisition device, an odor output device, and a data processing and control device. The multimodal physiological signal acquisition device is used to acquire multiple types of physiological signals. For example, the multimodal physiological signal acquisition device includes at least one or more of an EEG signal acquisition device, a skin electrical signal acquisition device, and an ECG signal acquisition device, which are respectively used to acquire EEG signals, skin electrical signals, and ECG signals. During use, any one or more of the EEG signal acquisition device, the skin electrical signal acquisition device, and the ECG signal acquisition device can be called. The odor output device is used to output odorous gas to the user's nasal cavity. The data processing and control device of this embodiment is respectively connected to the multimodal physiological signal acquisition device and the odor output device, and is used to process and analyze the acquired physiological signals, determine the odor type information of the gas output by the odor output device, and determine the emotional state information of the user. In actual applications, the data processing and control device can be an intelligent terminal that can control parameters such as the concentration and flow rate of the gas output by the odor output device. In order to analyze the correlation between the odor type and the user's emotional state, this embodiment can control the multimodal physiological signal acquisition device to collect multimodal physiological signals while the odor output device outputs odorous gas, so that the odor output device and the multimodal physiological signal acquisition device work synchronously to ensure the accuracy of the analysis.

[0037] Specifically, combined Figure 3 As shown in , the EEG signal acquisition device of this embodiment is in the form of a helmet, which includes a helmet 11, a first control module 12 provided on the helmet 11, and a first electrode 13 connected to the first control module 12. A plurality of first electrodes 13 can be provided, which are provided on the inside of the helmet for collecting EEG signals. The first control module 12 is connected to a data processing and control device. In actual application, the user can wear the EEG signal acquisition device on the user's head so that the scalp contacts the first electrode 13, thereby realizing the collection of EEG signals. The sampling rate of the EEG signal of this embodiment is 256 Hz, and the electrode impedance is ≤5 kΩ. It can collect the changes of EEG signals before, during, and after odor stimulation in real time. Combined with Figure 4 As shown in , the skin electrical signal acquisition device of this embodiment is set in the form of a wristwatch, and the skin electrical signal acquisition device includes a wristband 111, a second control module 112 provided on the wristband, and a second electrode 113 connected to the second control module 112. A plurality of second electrodes 113 can be provided. The second electrode 113 is used to collect skin electrical signals. The second control module 112 is also connected to the data processing and control device. In actual application, Figure 3As shown in , the skin electrical signal acquisition device can be worn on the user's wrist, the second electrode 113 and the second control module 112 are connected by a conductive wire, and the second electrode 113 can be clamped on the user's fingertips to collect the user's skin electrical signals. Figure 5 As shown in , the ECG signal acquisition device of this embodiment is configured in the form of a waist belt. The ECG signal acquisition device includes a waist belt 10, a third control module 20 provided on the waist belt 10, and a third electrode 30 connected to the third control module 20, and the third electrode 30 is used to acquire ECG signals. The third control module 20 is also connected to the data processing and control device. In actual application, the user can wear the ECG signal acquisition device around the waist. There are multiple third electrodes 30, and they are all connected to the third control module 20 through a conductive wire. The third electrode 30 is then attached to the user's chest near the heart to collect ECG signals. In this embodiment, the sampling frequency of the ECG signal acquisition device is generally set to ≥500Hz to ensure detection accuracy.

[0038] like Figure 6 As shown in FIG, the odor output device in this embodiment includes an odor storage device 310 and an odor induction device 320 for controlling the output of odor from the odor storage device 310. The odor storage device 310 includes several odor bottles 311 for storing gases with different scents, such as floral, herbal, woody, and citrus scents. The bottom of each odor bottle 311 is connected to the odor induction device 320 via a pipe. The odor output device also includes a flexible conduit 330, which is connected to each odor bottle 311 and is used to direct the scented gas into the user's nasal cavity. In this embodiment, the odor induction device 320 stores odorless ammonia gas. A driving device in the odor induction device 320 drives the odorless ammonia gas through the pipe and into the bottom of the odor bottle 311, creating a turbulent airflow that volatilizes the gas in the odor bottle 311. The odor-bearing gas is then directed into the user's nasal cavity via the flexible conduit 330. In one implementation, the scent output device of this embodiment further includes a heating module and a humidity control module, which can be positioned between the scent storage device and the flexible conduit. These modules are used to control the temperature and humidity of the gas entering the user's nasal cavity, ensuring that the gas maintains a constant temperature and humidity, improving the physiological adaptability of the scent stimulation and making the output gas temperature and humidity closer to natural breathing conditions. This reduces non-odor-related interference factors and further ensures the purity and comfort of the olfactory stimulation, as well as the consistency of the neural response. Furthermore, the scent bottle 311 of this embodiment can also be replaced with an electronically controlled sprayer or a microfluidic scent generator to achieve higher-precision scent stimulation.

[0039] Of course, the odor-inducing device 320 of the odor output device of this embodiment can also precisely control and release the gas output by the odor storage device 310. For example, odorless ammonia is introduced into the bottom of the odor bottle 311 at a constant flow rate (e.g., 1.5 L / min), simulating a "pseudo-boiling" state and continuously pushing the odor bottle 311 to volatilize the gas. After passing through the heating module and the humidity control module, the gas is introduced into the user's nasal cavity at a specific concentration (10%, 30%, 50%, 70%, 90%) and for a fixed duration (e.g., 4 seconds). This helps ensure the consistency and repeatability of the odor stimulation process and the validity of the experimental results. In addition, this embodiment can also provide a three-way solenoid valve at the end of the flexible catheter to precisely control the gas entering the user's nasal cavity. When the three-way solenoid valve is opened, the multimodal physiological signal acquisition device is simultaneously activated to collect multimodal physiological signals and record key time points to ensure the synchronous collection of multimodal physiological signals during the odor stimulation process. Furthermore, the EEG signal acquisition device of this embodiment features 32 channels, and the distal end of the flexible catheter is designed in tandem with the spatial structure of the device, enabling synchronous alignment of the stimulation pathway with the device's channels, enhancing the accuracy and spatial traceability of neural signal timing. The odor output device of this embodiment also features a roller at its base, allowing for movement and adaptability to a wider range of application scenarios.

[0040] In specific applications, while the odor output device outputs odor-bearing gas into the user's nasal cavity, physiological signals can be acquired. In this case, the physiological signals include any one or more of EEG signals, galvanic skin signals, and electrocardiogram signals. In other words, this embodiment can acquire only EEG signals, only ECG signals, or all three physiological signals by calling an EEG signal acquisition device, an galvanic skin signal acquisition device, or an electrocardiogram signal acquisition device. This embodiment automatically and synchronously marks the acquisition period of each physiological signal and completes the timing alignment of each physiological signal, which helps ensure the precise correspondence of each physiological signal under odor stimulation.

[0041] Step S200: Determine odor type information and emotional state information based on the physiological signal, and establish an odor perception dimensional space model to graphically present the relationship between the odor type information, the emotional state information, and the physiological signal.

[0042] After collecting physiological signals within a preset time period, this embodiment can preprocess the physiological signals. This preprocessing includes bandpass filtering, power frequency interference removal, physiological artifact removal, and bad channel repair. For example, a 1-70 Hz bandpass filter can be selected for filtering, and a 50 Hz notch filter can be selected for power frequency interference removal. Next, this embodiment performs feature extraction based on the preprocessed physiological signals to obtain target features.

[0043] Since the physiological signals in this embodiment include any one or more of EEG signals, galvanodermal signals, and ECG signals, the methods for determining target features vary depending on the physiological signals acquired. Specifically, when the physiological signal is an ECG signal, this embodiment can determine a heart rate variability characteristic value based on the preprocessed ECG signal. The heart rate variability characteristic value includes the root mean square difference (RMS) of the preprocessed ECG signal, the low-frequency power, and the high-frequency power. The target feature is then obtained based on the heart rate variability characteristic value. Therefore, the target feature in this case includes the RMS, low-frequency, and high-frequency power of the preprocessed ECG signal. The RMS of the preprocessed ECG signal can be used to reflect short-term wave activity in the nervous system, while the low-frequency and high-frequency powers of the preprocessed ECG signal can represent the sympathetic and parasympathetic nerve frequency domain components, respectively. Furthermore, this embodiment can also determine the ratio between the low-frequency power and the high-frequency power to reflect the state of autonomic neural balance.

[0044] Furthermore, heart rate variability characteristic values ​​can be used to distinguish between stimulus conditions with high arousal (e.g., acidic or bitter odors) and low arousal (e.g., sweet or fruity odors). Therefore, when determining odor type information and emotional state information based on the determined target features, if the high-frequency power of the preprocessed ECG signal continues to decrease within a preset time period, or if the ratio of the low-frequency power to the high-frequency power of the preprocessed ECG signal continues to decrease within a preset time period, this can indicate that the user is relatively averse to the smell. In this case, the odor type information is determined to be acidic or bitter, and the emotional state information is determined to be high alert. On the other hand, if the root mean square difference and high-frequency power of the preprocessed ECG signal continue to increase within a preset time period, this indicates that the user is relatively comfortable with the smell and is feeling relaxed and happy. In this case, the odor type information is determined to be sweet or fruity, and the emotional state information is determined to be relaxed and happy. The heart rate variability characteristic value of this embodiment not only participates in odor response modeling as a feedback channel, but can also be used for applications such as personalized odor threshold adjustment and human-computer interaction context perception, with high scalability and engineering practical value.

[0045] Furthermore, this embodiment also analyzes the effects of several typical smells on user emotions, such as Figure 7 As shown in Figure 7 The changing trends of several key indicators (such as root mean square deviation, low-frequency power, high-frequency power, and the ratio of low-frequency power to high-frequency power) in the heart rate variability characteristic values ​​induced by three typical odors (such as orange, rose, and bitter almond) are shown. Figure 7In the image, the yellow column on the left represents the scent of orange, which corresponds to a sweet or fruity odor type. This triggers a significant increase in high-frequency power and root mean square difference (RMSD), while both low-frequency power and the ratio of low-frequency power to high-frequency power decrease. This indicates parasympathetic nervous system activity, a favorable odor perception, and a relaxed and pleasant mood, corresponding to a relaxed and pleasant emotional state. The orange column in the middle represents the scent of rose, with stable heart rate variability eigenvalues ​​and stable arousal and comfort responses, indicating a low level of user sensitivity to this scent. The dark red column on the right represents the scent of bitter almond, which corresponds to an acidic or bitter odor type. This triggers an increase in low-frequency power and the ratio of low-frequency power to high-frequency power, while a decrease in high-frequency power, indicating sympathetic nervous system activation. This suggests that the user is averse to the scent, inducing tension or alertness, and a high level of emotional alertness.

[0046] In another embodiment, if the acquired physiological signals include EEG signals, galvanic skin signals and electrocardiogram signals, it means that multimodal physiological signals are acquired at this time. At this time, the time domain features, frequency domain features, wavelet features and entropy features of the preprocessed multimodal physiological signals can be extracted to obtain multidimensional features. That is to say, the time domain features, frequency domain features, wavelet features and entropy features are extracted from the preprocessed EEG signals, galvanic skin signals and electrocardiogram signals respectively. The time domain features in this embodiment include mean and variance, the frequency domain features include power spectral density, the wavelet features include wavelet coefficients and wavelet energy moments, and the entropy features include power spectral entropy, fuzzy entropy and singular value decomposition entropy. This embodiment effectively captures the multidimensional information of the neural response after odor stimulation by extracting multidimensional features such as time domain features, frequency domain features, wavelet features and entropy features, realizes efficient recognition of complex odors and subtle odor differences, and supports high-precision classification of different food odor types.

[0047] Specifically, the above mean is expressed as: (1) In formula (1), is the total number of sampling points of the physiological signal, For the The signal value of the sampling point, Indicates the sum of the signal values ​​of all collected physiological signals.

[0048] The above variance is expressed as: (2) In formula (2), is the mean of the signal; Indicates the The square of the deviation of a physiological signal from the mean is used to eliminate the offset of positive and negative deviations.

[0049] The above power spectral density is expressed as: (3) In formula (3), Represents the frequency of physiological signals, represents the observation length of the physiological signal, Represents the spectrum of a physiological signal.

[0050] The wavelet feature in this embodiment is the wavelet energy moment, which is expressed as: (4) In formula (4), It means that the physiological signal is decomposed by wavelet The coefficients of the subbands, is the coefficient index, is the number of layers for wavelet decomposition of the original physiological signal.

[0051] The entropy feature in this embodiment is fuzzy entropy, which is expressed as: (5) In formula (5), fuzzy entropy is used to calculate the similarity of patterns in physiological signals with pattern length. Change to The rate of change when , as well as There are three core parameters: and There are two key functions, Indicates the number of similarities of continuous physiological signals analyzed, usually an integer. To control the looseness of similarity, it is usually taken as a multiple of the standard deviation of the physiological signal, or a fixed value can be directly taken, for example, is 0.5, is the total number of original physiological signals, usually required . Used to indicate the presence of continuous The physiological signals satisfy the fuzzy similarity ratio. In order to indicate the existence of continuous The physiological signals satisfy the fuzzy similarity ratio.

[0052] Furthermore, after calculating the aforementioned multidimensional features, this embodiment can determine target features based on these multidimensional features. Specifically, the multidimensional features in this embodiment are extracted from EEG signals, galvanic skin signals, and electrocardiogram signals. Therefore, when determining the target features, this embodiment can extract the optimal features from the multidimensional features of each physiological signal and then use the optimal features of each physiological signal as the target features. Specifically, this embodiment can use the mRMR algorithm, the Relief-F algorithm, or the CD-FS-ILFS algorithm to extract the optimal features from the multidimensional features of each physiological signal. The mRMR (Minimum Redundancy Maximum Relevance) algorithm aims to select features that are highly correlated with the target variable while minimizing redundancy between features. The Relief-F algorithm calculates the weighted distance difference between features in neighboring samples and can be used to assess feature importance. The CD-FS-ILFS (Class-Discriminative Feature Selection + Infinite Latent Feature Selection) algorithm combines discriminative scoring with unsupervised graph modeling and is suitable for high-dimensional feature screening. Based on the above algorithm, the optimal features of each physiological signal can be screened out separately, thus obtaining the optimal features of the EEG signal, the optimal features of the skin electrode signal, and the optimal features of the ECG signal. Finally, the optimal features of the EEG signal, the optimal features of the skin electrode signal, and the optimal features of the ECG signal are used as target features.

[0053] In another implementation, this embodiment can also filter out target features from the above-mentioned multidimensional features based on the degree of influence of the features on odor recognition and emotion analysis. Specifically, this embodiment first calculates the influence score corresponding to the time domain features, frequency domain features, wavelet features, and entropy features of each preprocessed physiological signal, that is, calculates the influence score corresponding to the time domain features, frequency domain features, wavelet features, and entropy features of the EEG signal, calculates the influence score corresponding to the time domain features, frequency domain features, wavelet features, and entropy features of the skin electrode signal, and calculates the influence score corresponding to the time domain features, frequency domain features, wavelet features, and entropy features of the ECG signal. Then, based on the influence score, the feature with the highest influence score in each preprocessed physiological signal is determined, and the feature with the highest influence score in each preprocessed physiological signal is used as the optimal feature of each preprocessed physiological signal. Finally, based on the optimal feature of each preprocessed physiological signal, the target feature is obtained.

[0054] Furthermore, after obtaining the target feature, this embodiment can obtain a preset classification model, classify the target feature based on the classification model, and output odor type information and emotional state information corresponding to the target feature. In this embodiment, the classification model includes an ensemble learning framework or a manifold classifier. The ensemble learning framework can be a random forest network, a long short-term memory network, or a particle swarm optimization network. The manifold classifier can be an MDRM (Minimum Distance to Riemannian Mean) classifier or a GFMDRM (Geometric Functional Minimum Distance to Riemannian Mean) classifier, which classifies odor type and emotional state. In this embodiment, emotional state information includes pleasantness and arousal. Pleasure is an indicator that measures the degree of positivity or negativity of an emotional experience, reflecting whether the emotion brings about a person's subjective feeling of pleasure or unpleasantness. Arousal measures the level of physiological and psychological activation of the emotional experience, namely the degree of excitement or tension caused by the emotion, reflecting the intensity of physical and psychological activation.

[0055] The classification model of this embodiment is pre-configured and can integrate odor recognition functionality with a binary emotion classification model. This classification model can reflect the mapping relationship between physiological signal features and odor types, as well as arousal and pleasantness. That is, when the target features extracted from the physiological signals are input into the classification model, the classification model can automatically output the corresponding odor type information and pleasantness and arousal values ​​based on the input target features, thereby obtaining emotional state information. If the pleasantness value is high and the arousal value is low, the emotional state information is determined to be a relaxed and tranquil state. If the pleasantness value is low and the arousal value is high, the emotional state information is determined to be an anxious and alert state. For example, if the extracted target feature is the variance of the EEG signal, if the variance of the EEG signal is low or continuously decreasing, the classification model can automatically output the odor type information as acidic or bitter, and the pleasantness value is low and the arousal value is high, then the emotional state information can be determined to be a highly alert state. This embodiment utilizes an ensemble learning framework or manifold classifier to identify odor type and emotional state information, improving model stability and cross-subject generalization, enabling high-precision classification of different odor types and emotional states, making it suitable for large-scale industrial applications. The classification model of this embodiment utilizes an ensemble learning framework, which helps fully utilize the spatial structure of the covariance matrix, improves the accuracy of odor type and emotion classification, and enhances the adaptability of the classification model. Of course, in other application scenarios, this embodiment can also directly decode raw physiological signals in terms of data processing by introducing an end-to-end deep learning architecture (such as a convolutional neural network).

[0056] Furthermore, in other implementations, after determining the target feature, this embodiment can also obtain weight information for the optimal feature of each preprocessed physiological signal, as the target feature is the optimal feature of each preprocessed physiological signal. Then, based on the optimal feature of each preprocessed physiological signal and the corresponding weight information, a target feature weight value is determined. This target feature weight value can be used to comprehensively reflect the user's overall level of odor perception and emotional perception using their EEG, galvano-dermal, and ECG signals. Therefore, based on the target feature weight value, odor type information and emotional state information can be determined. Specifically, if the target feature weight value is higher than a preset value, it indicates that the user prefers the smell and is feeling relaxed and happy. Therefore, the odor type information is determined to be sweet or fruity, and the emotional state information is determined to be relaxed and happy. If the target feature weight value is lower than the preset value, it indicates that the user is feeling aversion to the smell and is feeling stressed. Therefore, the odor type information is determined to be sour or bitter, and the emotional state information is determined to be highly alert. It should be noted that, in this embodiment, the determination of odor type information and emotional state information based on physiological signals is based on the perception of most users on odors, excluding individual preferences for certain odors.

[0057] Of course, this embodiment can also collect skin electrical signals separately for analysis. Skin electrical signals can reflect the instantaneous arousal level of the autonomic nervous system under odor stimulation, and can also analyze the user's emotional state information after smelling the odor. At the same time, this embodiment can also expand the introduction of additional physiological data such as respiratory rate, skin temperature, and electromyography to build a multimodal deep learning model, explore the temporal and spatial correlation of multi-source signals, and improve the recognition ability of complex odors and slightly different odors. It is suitable for high-precision odor evaluation, new food research and development, and spice development. This embodiment greatly improves the accuracy and robustness of odor perception and emotion decoding by recording multi-level information of multimodal physiological information.

[0058] Furthermore, after obtaining the above-mentioned target features, this embodiment can also use the K-PCA (Kernel Principal Component Analysis) algorithm to map the target features to a high-dimensional feature space, making them linearly separable in the high-dimensional space, and then reduce the dimensionality to a three-dimensional space through principal component analysis. Next, the odor type information and the emotional state information are mapped to the target features to establish a correlation between odor, emotion, and physiological signals. Based on the correlation, a three-dimensional space model is formed. This three-dimensional space model is the odor perception dimensional space model. The odor perception dimensional space model of this embodiment can graphically present the spatial clustering of the odor type information, the emotional state information, and the target features, which is beneficial for intuitively assisting food authenticity judgment, sensory quality control, and product development decision-making, and promotes the construction of food safety and neurosensory evaluation systems. Of course, this embodiment can also use other visualization methods to establish the odor perception dimensional space, and the present invention is not limited to this.

[0059] Step S300: determining an odor recognition result based on the odor type information, and outputting an emotion recognition result based on the emotional state information.

[0060] After obtaining the odor type information, this embodiment can also match the odor type information with the identification information of the gas output by the odor output device. The gas identification information reflects the description of the gas's odor. If the odor type information and the identification information successfully match, it indicates that the analyzed odor type information is correct, and therefore the odor identification result can be determined to be correct. If the odor type information and the identification information fail to match, it indicates that the analyzed odor type information is incorrect, and therefore the odor identification result can be determined to be incorrect. Based on this, this embodiment can also be applied to scenarios where food odor authenticity is verified. Specifically, the authenticity of the odor output by the odor output device can be determined by matching the odor type information with the identification information of the gas output by the odor output device. For example, if the odor type information and the identification information successfully match, it indicates that the odor of the gas output is consistent with the analyzed odor type information, and therefore the odor is authentic. This helps provide comprehensive and reliable technical support for odor sensory evaluation in the food industry.

[0061] In another implementation, this embodiment can also obtain the user's self-assessment of their emotions regarding the odor when the odor output device outputs odorous gas into the user's nasal cavity. Specifically, the user can obtain the self-assessment of their emotions based on the pleasure and arousal they enter after smelling the odor. This self-assessment of emotions is then compared with the analyzed emotional state information to determine the emotion recognition result. If the self-assessment of emotions is identical or similar to the analyzed emotional state information, the emotion recognition result can be determined to be accurate.

[0062] This embodiment can be further expanded to interdisciplinary fields such as healthcare, psychology, and intelligent interaction. Furthermore, with the advancement of artificial intelligence, micro-nano sensing technology, and wearable devices, the present invention can be deeply integrated with portable, low-power sensing hardware and adaptive learning algorithms to enable the development of consumer-grade odor-sensing products. Furthermore, the method of the present invention can also be applied to emerging fields such as cross-species olfactory pattern research and animal behavior experiments, promoting the integrated development of neurosensory science research and industrial applications.

[0063] Based on the above embodiment, the present invention also provides a fusion analysis system for odor perception and emotion recognition, which is used in the steps of the above method embodiment. Specifically, Figure 8 As shown in , the data processing and control device in the fusion analysis system of odor perception and emotion recognition includes: a physiological signal acquisition module 10, an odor and emotion analysis module 20, and a result judgment module 30. The physiological signal acquisition module 10 is used to obtain physiological signals when the odor output device outputs odorous gas to the user's nasal cavity. The physiological signals include any one or more of electroencephalogram signals, skin electrical signals, and electrocardiogram signals. The odor and emotion analysis module 20 is used to determine odor type information and emotional state information based on the physiological signals, and establish an odor perception dimensional space model to graphically present the relationship between the odor type information, the emotional state information, and the physiological signals. The result judgment module 30 is used to determine the odor recognition result based on the odor type information, and output the emotion recognition result based on the emotional state information.

[0064] The working principles of each module in the fusion analysis system of odor perception and emotion recognition in this embodiment are the same as the principles of each step in the above method embodiment, and will not be repeated here.

[0065] Each module of each device in the aforementioned fusion analysis system for odor perception and emotion recognition can be implemented in whole or in part through software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a terminal in hardware form, or stored in a memory in the terminal in software form, so that the processor can call and execute the corresponding operations of each module.

[0066] Based on the above embodiment, the present invention further provides a terminal, the principle block diagram of the terminal can be as follows: Figure 9 The terminal may include one or more processors 100 ( Figure 9 Only one is shown in the figure), memory 101, and computer program 102 stored in memory 101 and executable on one or more processors 100. For example, a program for integrating odor perception and emotion recognition. When one or more processors 100 execute computer program 102, each step of the embodiment of the method for integrating odor perception and emotion recognition can be implemented. Alternatively, when one or more processors 100 execute computer program 102, the functions of each module / unit in the embodiment of the system for integrating odor perception and emotion recognition can be implemented, without limitation herein.

[0067] In one embodiment, the processor 100 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0068] In one embodiment, memory 101 may be an internal storage unit of an electronic device, such as a hard drive or memory. Memory 101 may also be an external storage device of the electronic device, such as a plug-in hard drive, smart media card (SMC), secure digital (SD) card, flash memory card, etc. Furthermore, memory 101 may include both an internal storage unit of the electronic device and an external storage device. Memory 101 is used to store computer programs and other programs and data required by the terminal. Memory 101 may also be used to temporarily store data that has been output or is about to be output.

[0069] Those skilled in the art will understand that Figure 9The principle block diagram shown in the figure is only a block diagram of a partial structure related to the solution of the present invention, and does not constitute a limitation on the terminal to which the solution of the present invention is applied. The specific terminal may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0070] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, operation database or other media used in the embodiments provided by the present invention may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct RAM bus dynamic RAM (DRDRAM), and RAM bus dynamic RAM (RDRAM), etc.

[0071] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A fusion analysis method for odor perception and emotion recognition, characterized in that: The method comprises: When the odor output device outputs the odor-bearing gas into the user's nasal cavity, a physiological signal is obtained, wherein the physiological signal includes any one or more of an electroencephalogram signal, a skin electrical signal, and an electrocardiogram signal; Determining odor type information and emotional state information based on the physiological signal, and establishing an odor perception dimensional space model to graphically present the relationship between the odor type information, the emotional state information, and the physiological signal; An odor recognition result is determined based on the odor type information, and an emotion recognition result is output based on the emotional state information.

2. The fusion analysis method of odor perception and emotion recognition according to claim 1 is characterized in that: Determining odor type information and emotional state information based on the physiological signal includes: Preprocessing the physiological signal, and performing feature extraction based on the preprocessed physiological signal to obtain target features; Based on the target features, odor type information and emotional state information are determined.

3. The fusion analysis method of odor perception and emotion recognition according to claim 2 is characterized in that: The preprocessing includes: band-pass filtering, power frequency interference removal, physiological artifact removal, and bad channel repair.

4. The fusion analysis method of odor perception and emotion recognition according to claim 3 is characterized in that: When the physiological signal is an electrocardiogram signal, the feature extraction based on the preprocessed physiological signal to obtain the target feature includes: Determining a heart rate variability characteristic value based on the preprocessed ECG signal, wherein the heart rate variability characteristic value includes: a root mean square difference, a low frequency power, and a high frequency power of the preprocessed ECG signal; The target feature is obtained based on the heart rate variability feature value.

5. The fusion analysis method of odor perception and emotion recognition according to claim 4 is characterized in that: Determining odor type information and emotional state information based on the target features includes: If the high-frequency power of the preprocessed electrocardiogram signal continues to decrease within a preset time period or if the ratio of the low-frequency power to the high-frequency power of the preprocessed electrocardiogram signal continues to decrease within a preset time period, determining that the odor type information is a sour odor or a bitter odor, and determining that the emotional state information is a high alert state; If the root mean square difference and high-frequency power of the preprocessed electrocardiogram signal continue to rise within a preset time period, the odor type information is determined to be a sweet odor or a fruity odor, and the emotional state information is determined to be a relaxed and happy state.

6. The fusion analysis method of odor perception and emotion recognition according to claim 3 is characterized in that: When the physiological signal includes an electroencephalogram signal, an electrical skin signal, and an electrocardiogram signal, the feature extraction based on the preprocessed physiological signal to obtain the target feature includes: Extracting the time domain features, frequency domain features, wavelet features and entropy features of each preprocessed physiological signal; Determine the optimal features for each pre-processed physiological signal based on their impact on odor recognition and emotion analysis; The target feature is obtained based on the optimal feature of each preprocessed physiological signal.

7. The fusion analysis method of odor perception and emotion recognition according to claim 6 is characterized in that: Based on the degree of influence of the features on odor recognition and emotion analysis, the optimal features of each pre-processed physiological signal are determined, including: Calculate the influence scores corresponding to the time domain features, frequency domain features, wavelet features, and entropy features of each preprocessed physiological signal; Based on the influence degree scores, the feature with the highest influence degree score in each preprocessed physiological signal is determined, and the feature with the highest influence degree score in each preprocessed physiological signal is used as the optimal feature of each preprocessed physiological signal.

8. The fusion analysis method of odor perception and emotion recognition according to claim 7 is characterized in that: Determining odor type information and emotional state information based on the target features includes: Obtaining weight information of the optimal feature of each preprocessed physiological signal, and determining a target feature weight value based on the optimal feature of each preprocessed physiological signal and the corresponding weight information; Based on the weighted values ​​of the target features, the odor type information and the emotional state information are determined.

9. The fusion analysis method of odor perception and emotion recognition according to claim 8 is characterized in that: Based on the weighted values ​​of the target features, the odor type information and emotional state information are determined, including: If the target feature weighted value is higher than a preset value, the odor type information is determined to be a sweet odor or a fruity odor, and the emotional state information is determined to be a relaxed and pleasant state; If the target feature weighted value is lower than a preset value, the odor type information is determined to be a sour odor or a bitter odor, and the emotional state information is determined to be a high alert state.

10. The fusion analysis method of odor perception and emotion recognition according to claim 7, characterized in that: Determining odor type information and emotional state information based on the target features includes: Obtain a preset classification model, classify the target feature based on the classification model, and output odor type information and emotional state information corresponding to the target feature, wherein the classification model includes an integrated learning framework or a manifold classifier, and the emotional state information includes pleasure and arousal. When the pleasure is higher than the arousal, the emotional state information is a relaxed and pleasant state; when the pleasure is lower than the arousal, the emotional state information is a highly alert state.

11. The fusion analysis method of odor perception and emotion recognition according to claim 2, characterized in that: Establish a dimensional space model of odor perception, including: Mapping the odor type information, the emotional state information, and the target feature to establish an association relationship between the odor, emotion, and physiological signal; Based on the association relationship, an odor perception dimensional space model is established.

12. The fusion analysis method of odor perception and emotion recognition according to claim 1, characterized in that: Determining an odor recognition result based on the odor type information includes: matching the odor type information with identification information of the gas output by the odor output device to obtain a matching result; Based on the matching result, the odor recognition result is determined.

13. The fusion analysis method of odor perception and emotion recognition according to claim 1, characterized in that: Outputting an emotion recognition result based on the emotional state information includes: Obtain the user's self-evaluation of emotions regarding smells; An emotion recognition result is determined based on the emotion self-assessment information and the emotional state information.

14. A fusion analysis system for odor perception and emotion recognition, characterized by: The system is used to implement the steps of the fusion analysis method of odor perception and emotion recognition described in claims 1-13, and the data processing and control device in the system includes: a physiological signal acquisition module, configured to acquire physiological signals when the odor output device outputs odorous gas to the user's nasal cavity, wherein the physiological signals include any one or more of an electroencephalogram signal, an electrical skin signal, and an electrocardiogram signal; an odor and emotion analysis module for determining odor type information and emotional state information based on the physiological signal, and establishing an odor perception dimensional space model to graphically present the relationship between the odor type information, the emotional state information, and the physiological signal; A result determination module is configured to determine an odor recognition result based on the odor type information, and output an emotion recognition result based on the emotional state information.

15. A terminal, characterized in that: The terminal includes a memory, a processor, and a fusion analysis program of odor perception and emotion recognition stored in the memory and runnable on the processor. When the processor executes the fusion analysis program of odor perception and emotion recognition, the steps of the fusion analysis method of odor perception and emotion recognition as described in any one of claims 1 to 13 are implemented.

16. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a fusion analysis program for odor perception and emotion recognition, and the fusion analysis program for odor perception and emotion recognition implements the steps of the fusion analysis method for odor perception and emotion recognition as described in any one of claims 1 to 13 on the computer-readable storage medium.

Images