A street greening quality detection method based on physiological arousal recognition

Abstract

The present invention discloses a street greening quality detection method based on physiological arousal recognition, comprising: establishing a greening quality factor index system according to high-frequency street landscape characteristics, acquiring and uniformly processing street greening images to conduct greening stimulation physiological experiments; collecting original data, reclassifying and performing difference wave processing on the original data according to the greening quality factor index, and obtaining effective physiological data that can be used for greening quality factor arousal feature identification; calculating physiological arousal feature parameters based on the obtained effective physiological data, using transfer learning to fuse and train the physiological arousal feature parameters to achieve physiological arousal feature importance judgment, and identifying the weighted average greening arousal index of the greening quality factor; analyzing the weighted average greening arousal index data of the greening quality factor to form a street greening quality detection model, inputting the labeled street samples to be analyzed into the street greening quality detection model, and obtaining the labeling results of the street greening quality grading detection target data.

Description

Technical Field

[0001] The present invention relates to the field of built environment quality measurement, and in particular to a street greening quality detection method based on physiological arousal recognition. Background Art

[0002] Street greening is an integral part of the urban landscape system, closely related to people's lives, and plays an indispensable role in the development of the built environment. As my country's urbanization enters the era of "refined operation," the connotation of building environmental stock renewal is becoming increasingly rich. Street greening, as an important part of human settlement construction, is also the most basic landscape component for urban repair, ecological empowerment, and environmental quality improvement. It is also an appropriate construction behavior that meets the people's demand for a better life and highlights the quality of the city. For example, when planning and designing urban street greening, it is important to measure and improve the greening structure and form, sight lines, and other components that directly affect the level of greening construction, thereby affecting the street space form, environmental quality, and public vitality.

[0003] In recent years, common quality detection methods for built environmental greening have mostly established an evaluation index system based on the visual characteristics of greening images, analyzed the changing trends of greening elements, their combined impacts and their construction needs for environmental sites, and produced a series of research results through comparative evaluation of urban environmental samples; related intellectual property achievements include: an automatic identification method for plant growth quality through static plant image acquisition, extraction of high-quality plant features, structured storage of time series, and calculation of morphological feature entropy (application number: 201911040562.8); a comprehensive evaluation method for ornamental plant landscape performance through the establishment of a multi-level performance evaluation system, information multivariate database call, weight judgment matrix calculation, and stacked bar chart generation (application number: 201910407314.6); an analysis method for the visual comfort of plant landscapes through the acquisition of plant image evaluation data, landscape effect threshold scores, green view rate difference analysis, and logistic stepwise regression model calculation (application number: 202210304173.7), etc. Although the current inventions on greening quality have made certain progress, there are still certain limitations: the acquisition of raw data and the final presentation of results are still restricted by technical conditions and subjective will, the cost of manual labeling is high, and the data update speed is difficult to cope with the changes and needs of urban high-speed construction; the greening quality indicator system is mostly based on existing models and frameworks, focusing only on a single or a few greening components, affecting the accuracy and globality of the overall environmental quality analysis system; in specific plans, the acquisition and analysis of greening quality factor parameters mostly rely on experience or evaluation scoring, and its operating efficiency, scientificity and universality need to be improved. Summary of the Invention

[0004] In order to address the deficiencies mentioned in the above background technology, the present invention aims to provide a method for detecting the quality of street greening based on physiological arousal recognition.

[0005] The purpose of the present invention can be achieved by the following technical solution: a method for detecting the quality of street greening based on physiological arousal recognition, the method comprising the following steps:

[0006] Establish a greening quality factor index system based on high-frequency streetscape characteristics, and obtain and uniformly process street greening images to conduct greening stimulation physiological experiments;

[0007] The original EEG, ECG, electrodermal, and electromyographic data were collected in response to street greening image stimulation. After reclassification and difference wave processing based on the greening quality factor index, effective physiological data that can be used to extract the arousal features of the greening quality factor was obtained.

[0008] Based on the valid physiological data obtained, the EEG, ECG, skin conduction, and myoelectricity physiological arousal feature parameters of the greening quality factor are calculated. Transfer learning is used to integrate and train the physiological arousal feature parameters to determine the importance of physiological arousal features and identify the weighted average greening arousal index of the greening quality factor.

[0009] Analyze the weighted average greening awakening index data of greening quality factors to form a street greening quality detection model for comparative detection of street greening quality;

[0010] The labeled street samples to be analyzed are input into the street greening quality detection model to obtain the labeling results of the street greening quality grading detection target data.

[0011] Preferably, the process of establishing the greening quality index system includes the following steps:

[0012] Frequency statistics were conducted on the research of street greening structure, plant attributes, and visual landscape, and high-frequency street greening quality factors were selected. At the same time, theoretical analysis methods were used to sort out the constituent elements, typical characteristics, and environmental connotations of street greening, and an indicator system of street greening quality factors was established. The indicator system of street greening quality factors includes primary factor dimensions, secondary variable factors, and tertiary factor change form indicators.

[0013] Obtain built environment street scene image data, use location scene recognition technology to identify street greening scene category images in street scene images, use image element semantic segmentation technology to perform feature sampling on the single greening quality variable factor of the greening scene category images, and use the square gradient function to determine a clear street greening target image;

[0014] M street greening images were randomly selected from the street greening target images, and phase randomization was performed three times on the m street greening images to obtain 3*m phase randomized images to form the experimental stimulus image library. All experimental stimulus images in the experimental stimulus image library were divided into n groups through an inter-group experiment, and played in a laboratory environment with random groups and the same frequency. The EEG, ECG, EDA, EMG and oscilloscope trigger signals of the corresponding data segments of the images were obtained in real time.

[0015] Preferably, the first-level element dimension indicators of the greening quality factor index system include greening structure, plant texture, sight relationship and landscape characteristics; the second-level variable factor indicators are an extension of the first-level greening quality factors; the third-level factor change form indicators are the expression form of variable factors, and the factor characteristics of greening quality are sampled through built environment street view pictures.

[0016] Preferably, the process of collecting the original EEG, EKG, GEM and EMG data stimulated by street greening images and reclassifying and performing difference wave processing on the original data according to the greening quality factor index comprises the following steps:

[0017] The original data of each street greening target image when it is stimulated is intercepted. According to the markers recorded by the trigger signal, the original data segments representing the same greening quality variable factor are divided into one category. Each category of data segments reflects the changes in the EEG, EKG, GEM and EMG of the subjects under the influence of a certain variable factor. The original data segments are baseline corrected, bandpass filtered, averaged, referenced, processed, noise reduced and artifact removed, and empirical mode decomposition (EMD) is used to correct the signal offset, so as to obtain the average amplitude and difference wave of the electrical signal of the greening quality factor stimulated state and the non-stimulated state.

[0018] The amplitude and phase images of the difference wave waveform were analyzed using Hanning window processing, fast Fourier transform and wavelet transform. The β and α frequency bands of the five leads PZ, P4, P5, O1, OZ and O2 of the EEG difference wave within the time window of the street greening target image presentation, the low-frequency and high-frequency bands of the RR interval of the ECG difference wave, the high-frequency band of the EMG difference wave after full-wave rectification and the normalized conductivity GSR of the skin electrode difference wave were extracted. The power spectral density of the EEG, ECG and EMG frequency bands, as well as the first-order difference of the skin electrode conductivity were calculated to obtain effective physiological data for identifying the physiological arousal characteristics of the greening quality factor.

[0019] Preferably, the process of calculating the EEG, EKG, EGG, and EMG physiological arousal characteristic parameters of the greening quality factor based on the obtained valid physiological data, using transfer learning to fuse and train the physiological arousal characteristic parameters, realizing the importance determination of the physiological arousal characteristics, and identifying the weighted average greening arousal index of the greening quality factor includes the following steps:

[0020] Based on the valid physiological data obtained, the physiological data of each type of green quality factor are superimposed and averaged, and the EEG, EKG, GEM and EMG physiological arousal characteristic parameters of the green quality factor are calculated respectively. The calculated EEG, EKG, GEM and EMG physiological arousal characteristic parameters of the green quality factor are then standardized.

[0021] The physiological arousal feature vector of the green quality factor is obtained based on the EEG, ECG, skin conduction and myoelectricity physiological arousal feature parameters of the green quality factor after standardization. Where, Represents the mth physiological arousal feature of the xth type of greening quality factor object, and constructs the physiological arousal feature importance judgment matrix B = {b ij}, where b ij It represents the importance ratio between the i-th awakening feature dimension and the j-th awakening feature dimension, thereby obtaining the weight vector of each feature w*=[w1,w2,…,w j ];

[0022] Transfer learning TLDA is used to perform physiological arousal feature fusion. 70% of the samples are used as the source domain dataset and the remaining 30% as the target domain dataset. The labeled arousal values ​​of the source domain street greening target images are obtained and the physiological arousal feature vector A of the greening quality factor is calculated. (m) and the marked arousal Y to perform sparse autoencoding processing, determine the number of neurons in the autoencoder to be q, q<m, and A (m) By bringing it into the neural network and assigning weights to physiological arousal features through neural network training, we can obtain the integrated vector E of the fusion features and the weighted average green arousal index O corresponding to E, as shown in the following formula:

[0023]

[0024] Where, ω i Represents the weight of each source domain, that is, the modulus of the similarity vector in the integrated vector E, o k It represents the predicted arousal level of the k-th category greening quality factor.

[0025] Preferably, the process of calculating the EEG, EKG, EGG and EMG physiological arousal characteristic parameters of the greening quality factor comprises the following steps:

[0026] Calculate the EEG wake-up characteristic parameter A of the greening factor object EEG, the calculation formula is as follows:

[0027]

[0028] Where, P β,x and P α,x Indicates the relative average power of the β and α bands of the five leads PZ, P4, P5, O1, OZ, and O2 for the current greening quality factor object x;

[0029] Calculate the ECG wake-up characteristic parameter A of the greening factor object ECG , the calculation formula is as follows:

[0030]

[0031] Where, P LF,x The power value of the low-frequency component of the electrocardiogram of the xth greening quality factor, P HF,x The power value of the high-frequency component of the electrocardiogram representing the xth greening quality factor;

[0032] Calculate the skin electrical arousal characteristic parameter A of the greening factor object EDA , the calculation formula is as follows:

[0033]

[0034] Where, t peak and t onset Indicates the peak point and start of the ΔGSR rise time during stress response, s peak and s onset Indicates the peak point and beginning of the ΔGSR amplitude value during stress response; A EDA It represents arousal 10% above baseline during GSR stress response;

[0035] Calculate the electromyographic awakening characteristic parameter A of the greening factor object EMG , the calculation formula is as follows:

[0036]

[0037] Where, P EMG,x Represents the power spectral density function of the electromyographic signal of the current greening quality factor object x, and f represents the frequency of the electromyographic signal;

[0038] The physiological arousal characteristic parameters of each greening quality factor are standardized and the following calculation formula is introduced:

[0039]

[0040] Where A(i) represents the i-th wake-up feature parameter, A min and A maxIndicates the minimum and maximum values ​​of the wake-up characteristic parameters, A Normalized Represents the normalized wake-up feature parameters.

[0041] Preferably, the process of analyzing the weighted average greening awakening index data of the greening quality factor to form a street greening quality detection model includes the following steps:

[0042] Obtain the weighted average green awakening index of each type of green quality factor, and use the sampling suitability KMO test and Bartlett sphericity test to test the weighted average green awakening index data of green quality factors. When the KMO value is greater than 0.6 and the sphericity test accompanying probability P value is ≤ 0.01, it is considered that the correlation between factor variables is strong and it is suitable for further analysis of the green factor object.

[0043] Calculate the initial greening quality variable matrix potential X = {x ij}, (i=1,2,3,...,m;j=1,2,3,...,n) the cumulative variance contribution rate M of the principal component K , select M K The potential main components of greening quality ≥80% are:

[0044]

[0045] Where x ij represents the jth greening quality variable factor of the i-th sample; represents the kth potential principal component of greening quality, ε ij represents the factor loading number of the jth potential principal component of the i-th variable factor, η ij represents the characteristic root of the jth potential principal component;

[0046] Extract the first k potential principal components to detect the quality of street greening, and calculate the weight w of the single greening quality variable factor based on the correlation coefficient matrix and variance contribution rate. i ',for:

[0047]

[0048] Where, γ j represents the variance contribution rate corresponding to the jth potential principal component of greening quality, w i 'The larger the value, the greater the importance of the greening quality variable factor;

[0049] The street greening quality detection model is formed according to the weight coefficient of the greening quality variable factor, which is used for comparative detection of street greening quality:

[0050] G=λ1x1+λ2x2+λ3x3+…+λ j xj

[0051] In the formula, λ1 represents the influence coefficient of the i-th factor, x j represents the j-th re-extracted factor greening arousal index data.

[0052] Preferably, the process of inputting the labeled street samples to be analyzed into the street greening quality detection model to obtain the labeled results of the street greening quality grading detection target data includes the following steps:

[0053] The physiological data of J subjects for I street greening images of N street samples are obtained, and the initial greening quality variable matrix Z of M greening quality variable factors of N street samples is obtained. ij}, (i=1,2,3,...,M; j=1,2,3,...,N), classify and label the EEG, EKG, EGG and EMG physiological arousal feature parameters of N street samples according to the green quality variable factor index category;

[0054] Establish a green arousal relationship fusion model between EEG, ECG, electrodermal, and electromyographic arousal characteristic parameters, generate a fusion green arousal index for J subjects for street greening quality variable factors, set the confidence level of the green arousal index data within the interval [0,1], and label the variable factors of the street samples with arousal levels;

[0055] The greening quality testing conditions are preset, and the street greening quality is divided into four levels: G1, G2, G3, and G4. The element dimensions of the street samples are graded and assigned values ​​from high to low, and the greening quality of the street samples is ranked accordingly.

[0056]

[0057] Where x ij Represents the green awakening degree labeling data of the jth element dimension at the i-th location, represents the average value of green arousal of all samples, σ represents the standard deviation of green arousal of all samples, C i It represents the grade of greening quality after the i-th element dimension is assigned;

[0058] The labeled street samples are input into the street greening quality detection model. The awakening degree of each element dimension of the street sample is obtained one by one through the calculated greening variable factors and the weight values ​​of the element dimensions. On this basis, the greening quality of each element dimension is re-assigned according to the greening quality grading detection conditions, and weighted superposition is performed according to the weights of the greening quality element dimensions to form and label the dimensionless greening quality value of the street sample.

[0059]

[0060] Where Y represents the greening quality value of the street sample, represents the weight of the t-th element dimension, C t 'Indicates the grade assignment of greening quality of the tth element dimension.

[0061] Preferably, a device comprises:

[0062] one or more processors;

[0063] a memory for storing one or more programs;

[0064] When one or more of the programs are executed by one or more of the processors, the one or more processors implement the above-mentioned method for detecting the quality of street greening based on physiological arousal recognition.

[0065] Preferably, a storage medium containing computer executable instructions is provided, wherein the computer executable instructions, when executed by a computer processor, are used to perform the above-mentioned street greening quality detection method based on physiological arousal recognition.

[0066] Beneficial effects of the present invention:

[0067] In response to the problem of insufficient overall analysis of characteristic data of greening in complex built environments, the present invention adopts a theoretical analysis method to sort out the constituent elements, typical characteristics and environmental connotations of street greening, establishes a greening quality factor index system that includes element dimensions, variable factors and specific change forms of factors, focuses on the correlation data between physiological awakening of street greening and street greening quality, realizes the overall analysis of multi-dimensional data of greening factor objects, and improves the speed and efficiency of street greening quality detection.

[0068] 2 In order to address the defects of complicated environmental arousal data collection process, significant fluctuation amplitude, and long processing cycle, the present invention identifies multi-source physiological signals as physiological arousal features with higher precision, uses transfer learning TLDA to perform multiple physiological arousal feature fusion training, and obtains the greening arousal index of the built environment greening image that can be quantified. It improves the greening arousal analysis data collection efficiency and the time and spatial accuracy of feature fusion analysis, overcomes the interference of social desirability bias, and promotes the scientific, objective and standardized development of the acquisition of basic data on street greening in the built environment.

[0069] 3. In view of the fact that the labeling of detection results is highly subjective and costly, and focuses on early measurement methods without exploring later applications, the present invention integrates the physiological arousal relationship between built environment location samples into the hierarchical greening quality detection results, takes greening quality factor objects as the data set, and presets greening quality grading detection conditions. This can improve the generalization ability of the street greening quality detection model, thereby providing an application approach for improving the greening quality of the built environment and evidence-based design, and promoting the refined development of the built environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, those skilled in the art can derive other drawings based on these drawings without inventive effort.

[0071] Figure 1 This is a flow chart of the street greening quality detection method based on physiological arousal recognition of the present invention;

[0072] Figure 2 is a graph of superimposed average physiological arousal characteristic feedback of subjects according to an embodiment of the present invention;

[0073] Figure 3 is a diagram of a physiological arousal feature fusion structure according to an embodiment of the present invention;

[0074] Figure 4 is a potential principal component relationship diagram of greening quality according to an embodiment of the present invention;

[0075] Figure 5 3 is a distribution diagram of the influence weights and comprehensive score coefficients of greening quality factors according to an embodiment of the present invention. DETAILED DESCRIPTION

[0076] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0077] like Figure 1 As shown, a method for detecting the quality of street greening based on physiological arousal recognition includes the following steps:

[0078] (1) Establish a greening quality factor index system based on high-frequency street landscape characteristics, obtain and uniformly process street greening images to conduct greening stimulation physiological experiments:

[0079] (1-1) Conduct frequency statistics on studies on street greening structure, plant attributes, visual plant landscape, etc. in existing literature, select high-frequency street greening quality influencing factors, and use theoretical analysis methods to sort out the constituent elements, typical characteristics and environmental connotations of street greening, and establish a street greening quality factor index system, including first-level factor dimensions, second-level variable factors and third-level factor specific change form indicators, as shown in Table 1. The first-level factor dimensions include greening structure, plant texture, sight relationship and landscape characteristics. The second-level indicators are an extension of the first-level greening quality factors, including 18 street greening variable factors such as greening form, greening composition, greening composition, and green view rate. The third-level indicators are the specific manifestations of variable factors, and the factor characteristics of greening quality are sampled through streetscape pictures of the built environment;

[0080] Table 1 Street greening quality factor index system

[0081]

[0082]

[0083] (1-2) Obtaining built environment street scene image data, using location scene recognition technology to determine street greening scene category images in the street scene images, performing feature sampling on a single greening quality variable factor of the greening scene category images using image element semantic segmentation technology, and using a square gradient function to determine a clear street greening target image;

[0084] (1-3) Randomly select m street greening target images from the experimental materials and perform three-fold phase randomization on these m street greening target images to obtain a library of 3*m phase randomized images to form the experimental stimulus image library. All experimental stimulus images are divided into n groups through an intergroup experiment and played back in a laboratory environment at the same frequency in randomized groups. EEG, ECG, EDA, EMG, and oscilloscope trigger signals are acquired in real time for the corresponding data segments of the images.

[0085] In this example, the Places365-CNN model dataset and the ResNet152-Hybrid1365 scene classification model were selected as tools for street greening scene detection. Images with the top three detection labels and semantically related to green elements were identified as street greening scene images. The ADE20K-CNN dataset and the Cascade-DilatedNet semantic segmentation model were used to perform feature sampling on the variable factors of street greening scene images. Images with a green view rate greater than 5% were designated as street greening target images for further analysis and processing.

[0086] For each greening quality factor, 80 target images of typical street greening features with 3 to 5 different variations were selected. This yielded a library of 3*80=240 phase-randomized street greening images, each subjected to a three-fold phase randomization process, to form the experimental stimulus image library. The stimulus images were divided into two groups for inter-group experiments. Each image was looped three times and flashed at a 10 Hz frequency for 3000 ms, with a resting state interval of 3000 ms between images. While the stimulus images were playing, markers were sent to a physiological oscilloscope, which recorded and collected raw data in real time: EEG, electrodermal, electromyographic, electrocardiographic, and oscilloscope trigger signal changes. A total of 320 raw data segments were captured.

[0087] (2) Collecting raw EEG, ECG, electrodermal, and electromyographic data stimulated by street greening images, reclassifying and performing difference wave processing on the raw data according to the greening quality factor index, and obtaining effective physiological data that can be used to extract the awakening features of the greening quality factor;

[0088] (2-1) The raw data of each street greening target image when stimulated is intercepted. Based on the markers recorded by the trigger signal, the raw data segments representing the same greening quality variable factor are classified into one category. Each category of data segments reflects the changes in the EEG, EKG, GEM, and EMG of the subject under the influence of a certain variable factor. The raw data segments are baseline corrected, bandpass filtered, averaged, referenced, subjected to independent component analysis (ICA), noise reduction, and artifact removal. Empirical mode decomposition (EMD) is used to correct signal offsets, thereby obtaining the average amplitude and difference wave of the electrical signals in the greening quality factor stimulated state and the non-stimulated state.

[0089] (2-2) The amplitude and phase images of the difference wave waveforms were analyzed using Hanning window processing, fast Fourier transform and wavelet transform, and the (8-12Hz) β and α (14-30Hz) frequency bands of the five leads PZ, P4, P5, O1, OZ, and O2 of the EEG difference wave within the time window of the street greening target image presentation were extracted, the low-frequency (LF: 0.04-0.15Hz) and high-frequency (HF: 0.15-0.4Hz) bands of the RR interval of the ECG difference wave, the high-frequency band (MF: 50-150Hz) of the full-wave rectified EMG difference wave, and the normalized conductivity GSR of the skin electrode difference wave were calculated. The power spectral density of the EEG, ECG, and EMG frequency bands, as well as the first-order difference of the skin electrode conductivity, were used to obtain effective physiological data for identifying the physiological arousal characteristics of the greening quality factor.

[0090] In this example, experimental data from 65 subjects were collected, with 61 sets of valid data. The signal sampling frequency was 500 Hz, and the baseline correction, noise reduction, artifact removal, filtering, independent component analysis (ICA), and signal offset correction preprocessing of the raw data were completed using software packages such as ECGLab, LedaLab, and HRVAS on the Matlab platform. The resistance value of each lead electrode of the EEG was below 10 kΩ. The Hanning window size was set to 25 ms, and the wavelet transform was Daubechiesdb2. The EEG, ECG, electrodermal, and electromyographic signal change data from 2000 ms before to 5000 ms after the street greening target image stimulus were intercepted, and then the signal change data reclassified according to the 18 categories of greening quality factors were subjected to difference wave analysis.

[0091] (3) Based on the obtained effective physiological data, the EEG, ECG, skin conduction and myoelectricity physiological arousal feature parameters of the greening quality factor are calculated, and the physiological arousal feature parameters are fused and trained using transfer learning to achieve the importance judgment of the physiological arousal feature and identify the weighted average greening arousal index of the greening quality factor;

[0092] (3-1) Obtain effective physiological data after street greening factor processing according to step (2-2), superimpose and average the physiological data of each type of greening quality factor, and calculate the EEG, EKG, GEM and EMG physiological arousal characteristic parameters of the greening quality factor (such as Figure 2 ), the specific steps are as follows:

[0093] (3-1-1) Calculate the EEG wake-up characteristic parameter A of the greening factor object EEG , the calculation formula is as follows:

[0094]

[0095] Where, P β,x and P α,x Indicates the relative average power of the β and α bands of the five leads PZ, P4, P5, O1, OZ, and O2 for the current calculation of the greening quality factor object x, A EEG The larger the value, the higher the degree of arousal of the visual area of ​​the brain caused by street greening;

[0096] (3-1-2) Calculate the ECG wake-up characteristic parameter A of the greening factor object ECG , the calculation formula is as follows:

[0097]

[0098] Where, P LF,x The power value of the low-frequency component of the electrocardiogram of the xth greening quality factor, P HF,x A represents the power value of the high-frequency component of the electrocardiogram of the xth greening quality factor.ECG The larger the value, the more active the sympathetic nerves are, that is, the higher the arousal level of street greening;

[0099] (3-1-3) Calculate the skin electrical arousal characteristic parameter A of the greening factor object EDA , the calculation formula is as follows:

[0100]

[0101] Where, t peak and t onset Indicates the peak point and start of the ΔGSR rise time during stress response, s peak and s onset Indicates the peak point and beginning of the ΔGSR amplitude value during stress response; A EDA GSR stress response represents arousal 10% above baseline, Ar EDA The larger the value, the greater the awakening energy of street greening;

[0102] (3-1-4) Calculate the electromyographic awakening characteristic parameter A of the greening factor object EMG , the calculation formula is as follows:

[0103]

[0104] Where, P EMG,x Represents the power spectral density function of the electromyographic signal of the current greening quality factor object x, f represents the frequency of the electromyographic signal, A EMG The larger the value, the greater the awakening energy of street greening;

[0105] In order to eliminate individual differences among subjects, the physiological arousal characteristics of each greening quality factor were standardized and the following calculation formula was introduced:

[0106]

[0107] Where A(i) represents the i-th wake-up feature parameter, A min and A max Indicates the minimum and maximum values ​​of the wake-up characteristic parameters, A Normalized represents the normalized physiological arousal characteristic parameter;

[0108] (3-2) Based on the normalized EEG, ECG, skin conduction and myoelectric physiological arousal characteristic parameters of the green quality factor, the physiological arousal characteristic vector of the green quality factor is obtained. Where, Represents the mth physiological feature of the xth type of green quality factor object. Construct the physiological arousal feature importance judgment matrix B = {b ij}, where b ijIt represents the importance ratio between the i-th awakening feature dimension and the j-th awakening feature dimension, thereby obtaining the weight vector of each feature w*=[w1,w2,…,w j ];

[0109] (3-3) Using transfer learning TLDA to perform physiological arousal feature fusion (such as Figure 3 ), 70% of the samples are used as the source domain dataset, and the remaining 30% are used as the target domain dataset. The labeled arousal value of the source domain street greening target image is obtained, and the physiological arousal feature vector A of the greening quality factor is calculated. (m) and the marked arousal Y to perform sparse autoencoding processing, determine the number of neurons in the autoencoder to be q, q<m, and A (m) Bring it into the neural network, and after neural network training, give the green awakening feature weights, and obtain the integrated vector E of the fusion feature and the weighted average green awakening index O corresponding to E, as shown in the following formula:

[0110]

[0111] Where, ω i Represents the weight of each source domain, that is, the modulus of the similarity vector in the integrated vector E, o k It represents the predicted arousal level of the k-th category greening quality factor.

[0112] In this example, EEG signals from five leads (PZ, P4, P5, O1, OZ, and O2) were collected, resulting in five EEG features. Together with two electrodermal features, one electrocardiogram (ECG) feature, and one electromyography (EMG) feature, there are nine physiological arousal features. The source domain arousal level, Y, is set using the ambient arousal level from the SAM scale as the criterion. Four transfer learning models were trained based on the EEG, ECG, EEG, and EMG physiological arousal features. These models employed a sigmoid activation function and optimized their weight parameters using gradient descent (SGD). The fusion models were evaluated using the balanced F1-score and accuracy (see Table 2). After training, the EEG + ECG + EEG + EMG model achieved the highest accuracy, with weights for each physiological arousal feature being 44.2%, 35.47%, 12.16%, and 8.17%, respectively.

[0113] Table 2 Comparison of physiological arousal feature fusion results

[0114]

[0115] (4) Analyze the weighted average greening awakening index data of greening quality factors to form a street greening quality detection model for street greening quality comparison detection

[0116] (4-1) Obtain the weighted average green awakening index of each type of green quality factor, and use the sampling suitability KMO test and Bartlett's sphericity test to test the weighted average green awakening index data of the green quality factor. When the KMO value is greater than 0.6 and the sphericity test accompanying probability P value is ≤ 0.01, it is considered that there is a strong correlation between the factor variables and it is suitable for further analysis of the green factor object;

[0117] (4-2) Calculate the initial greening quality variable matrix X = {x ij}, (i=1,2,3,...,m; j=1,2,3,...,n) the cumulative variance contribution rate M of the potential principal component K , select M K ≥80% of the potential main components of greening quality are:

[0118]

[0119] Where x ij represents the jth greening product variable factor of the i-th sample; represents the kth potential principal component of greening quality, ε ij represents the factor loading number of the jth potential principal component of the i-th variable factor, η ij represents the characteristic root of the jth potential principal component;

[0120] (4-3) Extract the first k potential principal components to test the quality of street greening, and calculate the weight w of the single greening quality variable factor based on the correlation coefficient matrix and variance contribution rate. i ', specifically:

[0121]

[0122] Where, γ j represents the variance contribution rate corresponding to the jth potential principal component of greening quality, w i 'The larger the value, the greater the importance of the greening quality variable factor;

[0123] (4-4) A street greening quality detection model is formed based on the factor weight coefficients for comparative detection of street greening quality. The model reflects the main factors affecting greening quality and their contribution to greening quality, specifically:

[0124] G=λ1x1+λ2x2+λ3x3+…+λ j x j

[0125] In the formula, λ1 represents the influence coefficient of the i-th factor variable, x j represents the j-th re-extracted factor fusion greening awakening index data.

[0126] In this example, the overall arousal of the street greening environment is used as the dependent variable, and the greening quality factor greening arousal data is used as the independent variable to construct the initial greening quality variable matrix. After KMO and Bartlett sphericity tests, the KMO sampling suitability of the variable matrix is ​​0.610>0.6, and the accompanying probability P value of the Bartlett test is equal to 0.000<0.01. Both conditions are met for further analysis. The potential principal components of the greening quality factors are extracted to obtain 5 new variables that are independent of each other and contain the initial factor information. The total explained variance is 80.782%, which exceeds 80% (as shown in Table 3), thus obtaining the potential principal component relationship diagram of greening quality (as shown in Table 3). Figure 4 According to step (4-3), the potential component factor coefficient (as shown in Table 4) and the influence weight (as shown in Table 4) of each initial greening quality variable factor are obtained. Figure 5 ), the influence weights of all greening quality factors are distributed between 5% and 7%, among which viewing mode (GM) has the greatest impact on greening quality.

[0127] Table 3 Total variance explained by potential principal components of greening quality

[0128]

[0129]

[0130] Table 4. Greening quality potential principal component score coefficient matrix

[0131]

[0132] (5) Input the labeled street samples to be analyzed into the street greening quality detection model to obtain the labeling results of the street greening quality grading detection target data.

[0133] (5-1) Obtain the physiological data of J subjects for I street greening images of N street samples, and obtain the initial greening quality variable matrix Z = {z ij}, (i=1,2,3,...,M; j=1,2,3,...,N), classify and label the EEG, EKG, EGG and EMG physiological arousal feature parameters of N street samples according to the green quality variable factor index category;

[0134] (5-2) Establishing a green arousal relationship fusion model between physiological arousal characteristic indicators through (3-3) generates a fusion green arousal index for J subjects for street greening quality variable factors, which is conducive to accurately identifying the subjects' arousal level for different street greening at the sampling time point. Set the arousal confidence level in the range of [0,1] and label the variable factors of the street samples with arousal levels;

[0135] (5-3) Preset greening quality testing conditions, divide the street greening quality into four levels G1, G2, G3, and G4, and assign values ​​to the element dimensions of street samples from high to low, thereby ranking the greening quality of street samples;

[0136]

[0137] Where x ij Represents the green awakening degree labeling data of the jth element dimension at the i-th location, represents the average value of green awakening degree of all samples, σ represents the standard deviation of green awakening data of all samples, C i It represents the grade of greening quality after the i-th element dimension is assigned;

[0138] (5-4) The labeled street samples are input into the street greening quality detection model. The awakening degree of the street sample element dimension is obtained one by one by calculating the greening variable factor and the weight value of the element dimension. On this basis, the greening quality of each element dimension is re-assigned according to the greening quality grading detection conditions, and weighted superposition is performed according to the weight of the greening quality element dimension to form and label the dimensionless greening quality value of the street sample.

[0139]

[0140] Where Y represents the greening quality value of the street sample, represents the weight of the t-th element dimension, C t 'Indicates the grade assignment of greening quality of the tth element dimension.

[0141] In this example, 60 subjects collected EEG, ECG, electrodermal, and electromyographic physiological data for greenery images at four street sample locations. Based on the (1-1) greenery quality factor index system, multiple greenery awakening characteristic parameters were integrated to obtain greenery awakening index data for 18 variable factors and four element dimensions for the street samples. After potential principal component analysis, weighted superposition, and greenery quality grading detection, the greenery quality values ​​of the element dimension indicators of the statistically annotated samples were compared. For areas with abnormal indicators, feasible improvement measures were proposed, and the greenery quality status of the obtained street samples was ranked and compared (as shown in Table 5). The annotated results of this example were compared with the results obtained from the previous greenery quality detection model (4-4) test, and the matching rate was found to reach 86%.

[0142] Table 5 Arousal and environmental quality of some street samples

[0143]

[0144] Based on the same inventive concept, the present invention also provides a computer device, which includes: one or more processors and a memory for storing one or more computer programs; the program includes program instructions, and the processor is used to execute the program instructions stored in the memory. The processor 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. It is the computing core and control core of the terminal, which is used to implement one or more instructions, specifically for loading and executing one or more instructions in a computer storage medium to implement the above method.

[0145] It should be further explained that, based on the same inventive concept, the present invention also provides a computer storage medium having a computer program stored thereon, which executes the above method when executed by a processor. The storage medium can be any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electrical, magnetic, infrared, or semiconductor system, device or component, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, device or component.

[0146] Throughout this specification, references to terms such as "one embodiment," "example," or "specific example" indicate that a specific feature, structure, material, or characteristic described in conjunction with that embodiment or example is included in at least one embodiment or example of the present disclosure. In this specification, schematic representations of these terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0147] The above shows and describes the basic principles, main features and advantages of the present disclosure. Those skilled in the art should understand that the present disclosure is not limited to the above embodiments. The above embodiments and descriptions are merely illustrative of the principles of the present disclosure. Various changes and improvements may be made to the present disclosure without departing from the spirit and scope of the present disclosure, and such changes and improvements shall fall within the scope of the present disclosure.

Claims

1. A method for detecting the quality of street greening based on physiological arousal recognition, characterized in that: The method comprises the following steps: Establish a greening quality factor index system based on high-frequency streetscape characteristics, obtain and uniformly process street greening images to conduct greening stimulation physiological experiments; The original EEG, ECG, electrodermal, and electromyographic data were collected in response to street greening image stimulation. After reclassification and difference wave processing based on the greening quality factor index, effective physiological data that can be used to extract the arousal features of the greening quality factor was obtained. Based on the valid physiological data obtained, the EEG, ECG, skin conduction, and myoelectricity physiological arousal feature parameters of the greening quality factor are calculated. Transfer learning is used to integrate and train the physiological arousal feature parameters to determine the importance of physiological arousal features and identify the weighted average greening arousal index of the greening quality factor. Based on the valid physiological data obtained, the EEG, ECG, skin conduction, and myoelectricity physiological arousal feature parameters of the greening quality factor are calculated. Transfer learning is used to integrate and train the physiological arousal feature parameters, realize the importance determination of physiological arousal features, and identify the weighted average greening arousal index of the greening quality factor. The process includes the following steps: Based on the valid physiological data obtained, the physiological data of each type of green quality factor are superimposed and averaged, and the EEG, EKG, GEM and EMG physiological arousal characteristic parameters of the green quality factor are calculated respectively. The calculated EEG, EKG, GEM and EMG physiological arousal characteristic parameters of the green quality factor are then standardized. The physiological arousal feature vector of the green quality factor is obtained based on the EEG, ECG, skin conduction and myoelectricity physiological arousal feature parameters of the green quality factor after standardization. Where, Represents the mth physiological arousal feature of the xth type of greening quality factor object, and constructs the physiological arousal feature importance judgment matrix B = {b ij }, where b ij It represents the importance ratio between the i-th awakening feature dimension and the j-th awakening feature dimension, thereby obtaining the weight vector of each feature w*=[w1,w2,...,w j ]; Transfer learning TLDA is used to perform physiological arousal feature fusion. 70% of the samples are used as the source domain dataset and the remaining 30% as the target domain dataset. The labeled arousal values ​​of the source domain street greening target images are obtained and the physiological arousal feature vector A of the greening quality factor is calculated. (m) and the marked arousal Y to perform sparse autoencoding processing, determine the number of neurons in the autoencoder to be q, q<m, and A (m) By bringing it into the neural network and assigning weights to physiological arousal features through neural network training, we can obtain the integrated vector E of the fusion features and the weighted average green arousal index O corresponding to E, as shown in the following formula: Where, ω i Represents the weight of each source domain, that is, the modulus of the similarity vector in the integrated vector E, o k represents the predicted arousal level of the k-th category greening quality factor; The process of calculating the EEG, EKG, EGG and EMG physiological arousal characteristic parameters of the greening quality factor comprises the following steps: Calculate the EEG wake-up characteristic parameter A of the greening factor object EEG , the calculation formula is as follows: Where, P β,x and P α,x Indicates the relative average power of the β and α bands of the five leads PZ, P4, P5, O1, OZ, and O2 for the current greening quality factor object x; Calculate the ECG wake-up characteristic parameter A of the greening factor object ECG , the calculation formula is as follows: Where, P LF,x The power value of the low-frequency component of the electrocardiogram of the xth greening quality factor, P HF,x The power value of the high-frequency component of the electrocardiogram representing the xth greening quality factor; Calculate the skin electrical arousal characteristic parameter A of the greening factor object EDA , the calculation formula is as follows: Where, t peak and t onset Indicates the peak point and start of the ΔGSR rise time during stress response, s peak and s onset Indicates the peak point and beginning of the ΔGSR amplitude value during stress response; A EDA It represents arousal 10% above baseline during GSR stress response; Calculate the electromyographic awakening characteristic parameter A of the greening factor object EMG , the calculation formula is as follows: Where, P EMG,x Represents the power spectral density function of the electromyographic signal of the current greening quality factor object x, and f represents the frequency of the electromyographic signal; The physiological arousal characteristic parameters of each greening quality factor are standardized and the following calculation formula is introduced: Where A(i) represents the i-th wake-up feature parameter, A min and A max Indicates the minimum and maximum values ​​of the wake-up characteristic parameters, A Normalized represents the normalized wake-up feature parameter; Analyze the weighted average greening awakening index data of greening quality factors to form a street greening quality detection model for comparative detection of street greening quality; The process of analyzing the weighted average greening awakening index data of the greening quality factor to form a street greening quality detection model includes the following steps: Obtain the weighted average green awakening index of each type of green quality factor, and use the sampling suitability KMO test and Bartlett sphericity test to test the weighted average green awakening index data of green quality factors. When the KMO value is greater than 0.6 and the sphericity test accompanying probability P value is ≤ 0.01, it is considered that the correlation between factor variables is strong and it is suitable for further analysis of the green factor object. Calculate the initial greening quality variable matrix potential X = {x ij }, i=1,2,3,...,m; j=1,2,3,...,n The cumulative variance contribution rate M of the principal component K , select M K The potential main components of greening quality ≥80% are: Where x ij represents the jth greening quality variable factor of the i-th sample; represents the kth potential principal component of greening quality, ε ij represents the factor loading number of the jth potential principal component of the i-th variable factor, η ij represents the characteristic root of the jth potential principal component; Extract the first k potential principal components to detect the quality of street greening, and calculate the weight w of the single greening quality variable factor based on the correlation coefficient matrix and variance contribution rate. i ',for: Where, γ j represents the variance contribution rate corresponding to the jth potential principal component of greening quality, w i 'The larger the value, the greater the importance of the greening quality variable factor; The street greening quality detection model is formed according to the weight coefficient of the greening quality variable factor, which is used for comparative detection of street greening quality: G=λ1x1+λ2x2+λ3x3+...+λ j x j In the formula, λ1 represents the influence coefficient of the i-th factor, x j represents the jth re-extracted factor greening arousal index data; Input the labeled street samples to be analyzed into the street greening quality detection model to obtain the labeling results of the street greening quality grading detection target data; The process of inputting the labeled street samples to be analyzed into the street greening quality detection model to obtain the labeled results of the street greening quality grading detection target data includes the following steps: The physiological data of J subjects for I street greening images of N street samples are obtained, and the initial greening quality variable matrix Z of M greening quality variable factors of N street samples is obtained. ij }, i = 1, 2, 3, ..., M; j = 1, 2, 3, ..., N, classify and label the EEG, EKG, EGG and EMG physiological arousal feature parameters of N street samples according to the green quality variable factor index category; Establish a green arousal relationship fusion model between EEG, ECG, electrodermal, and electromyographic arousal characteristic parameters, generate a fusion green arousal index for J subjects for street greening quality variable factors, set the confidence level of the green arousal index data within the interval [0,1], and label the variable factors of the street samples with arousal levels; The greening quality testing conditions are preset, and the street greening quality is divided into four levels: G1, G2, G3, and G4. The element dimensions of the street samples are graded and assigned values ​​from high to low, and the greening quality of the street samples is ranked accordingly. Where x ij Represents the green awakening degree labeling data of the jth element dimension at the i-th location, represents the average value of green arousal of all samples, σ represents the standard deviation of green arousal of all samples, C i It represents the grade of greening quality after the i-th element dimension is assigned; The labeled street samples are input into the street greening quality detection model. The awakening degree of each element dimension of the street sample is obtained one by one through the calculated greening variable factors and the weight values ​​of the element dimensions. On this basis, the greening quality of each element dimension is re-assigned according to the greening quality grading detection conditions, and weighted superposition is performed according to the weights of the greening quality element dimensions to form and label the dimensionless greening quality value of the street sample. Where Y represents the greening quality value of the street sample, represents the weight of the t-th element dimension, C t 'Indicates the grade assignment of greening quality of the tth element dimension.

2. The method for detecting street greening quality based on physiological arousal recognition according to claim 1, characterized in that: The process of establishing the greening quality factor index system includes the following steps: Frequency statistics were conducted on street greening structure, plant attributes, and visual landscape research, and high-frequency street greening quality factors were selected. At the same time, theoretical analysis methods were used to sort out the constituent elements, typical characteristics, and environmental connotations of street greening, and an indicator system of street greening quality factors was established. The indicator system includes primary factor dimensions, secondary variable factors, and tertiary factor change form indicators. Obtain built environment street scene image data, use location scene recognition technology to identify street greening scene category images in street scene images, use image element semantic segmentation technology to perform feature sampling on the single greening quality variable factor of the greening scene category images, and use the square gradient function to determine a clear street greening target image; M street greening images were randomly selected from the street greening target images, and phase randomization was performed three times on the m street greening images to obtain 3*m phase randomized images to form the experimental stimulus image library. All experimental stimulus images in the experimental stimulus image library were divided into n groups through an inter-group experiment, and played in a laboratory environment with random groups and the same frequency. The EEG, ECG, EDA, EMG and oscilloscope trigger signals of the corresponding data segments of the images were obtained in real time.

3. The method for detecting street greening quality based on physiological arousal recognition according to claim 2, characterized in that: The first-level element dimension indicators of the greening quality factor index system include greening structure, plant texture, sight relationship and landscape characteristics; the second-level variable factor indicators are an extension of the first-level greening quality factors; the third-level factor change form indicators are the expression form of variable factors, and the factor characteristics of greening quality are sampled through built environment street scene pictures.

4. The method for detecting street greening quality based on physiological arousal recognition according to claim 1, characterized in that: The process of collecting the original EEG, ECG, GEM and EMG data stimulated by street greening images and reclassifying and performing difference wave processing on the original data according to the greening quality factor index comprises the following steps: The original data of each street greening target image when it is stimulated is intercepted. According to the markers recorded by the trigger signal, the original data segments representing the same greening quality variable factor are divided into one category. Each category of data segments reflects the changes in the EEG, EKG, GEM and EMG of the subjects under the influence of a certain variable factor. The original data segments are baseline corrected, bandpass filtered, averaged, referenced, processed, noise reduced and artifact removed, and empirical mode decomposition (EMD) is used to correct the signal offset, so as to obtain the average amplitude and difference wave of the electrical signal of the greening quality factor stimulated state and the non-stimulated state. The amplitude and phase images of the difference wave waveform were analyzed using Hanning window processing, fast Fourier transform and wavelet transform. The β and α frequency bands of the five leads PZ, P4, P5, O1, OZ and O2 of the EEG difference wave within the time window of the street greening target image presentation, the low-frequency and high-frequency bands of the RR interval of the ECG difference wave, the high-frequency band of the EMG difference wave after full-wave rectification and the normalized conductivity GSR of the skin electrode difference wave were extracted. The power spectral density of the EEG, ECG and EMG frequency bands, as well as the first-order difference of the skin electrode conductivity were calculated to obtain effective physiological data for identifying the physiological arousal characteristics of the greening quality factor.

5. An electronic device, characterized in that: include: one or more processors; a memory for storing one or more programs; When one or more of the programs are executed by one or more of the processors, the one or more processors implement the street greening quality detection method based on physiological arousal recognition as described in any one of claims 1 to 4.

6. A storage medium containing computer-executable instructions, characterized in that: When executed by a computer processor, the computer executable instructions are used to execute a street greening quality detection method based on physiological arousal recognition as described in any one of claims 1 to 4.

Images