Historical mountain block ownership perception evaluation method based on multi-modal data

Abstract

The invention belongs to the crossing field of urban design and multidisciplinary, and particularly relates to a mountain historical block ownership perception evaluation method based on multi-modal data, public spaces of mountain historical blocks are divided into multiple types, an experimental path is selected from each public space, and multiple experimental parking points are selected from each experimental path; the evaluation index system comprises physiological data, psychological data and spatial data, and the spatial data comprises feature data and acoustic environment data; for each experiment path, obtaining feature data, recruiting a plurality of experimenters, collecting acoustic environment data and physiological data of each experimenter in real time, and collecting psychological data from the experimenters in real time in a questionnaire mode at an experiment stop point; independent variables are defined as spatial data, dependent variables are defined as psychological data and physiological data, based on experimental data, the relation between the independent variables and the dependent variables is learned through a random forest model, and bar charts of the influence degrees of the independent variables and the dependent variables are output; and the evaluation result is more accurate and persuasive.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of architecture, urban design, environmental behavior and environmental psychology, and particularly relates to a body perception evaluation method for a mountainous historical block based on multi-modal data. BACKGROUND

[0002] With the acceleration of urbanization, China's urban construction has entered the stock updating stage. The International Council on Monuments and Sites points out that the protection of historical blocks not only needs to focus on the material environment, but also needs to pay attention to the perceptual experience and cultural identity of block users. At present, the field of historical block protection evaluation relies on single-dimensional subjective questionnaires or pure engineering technical index analysis, which is difficult to truly reflect the comprehensive perceptual experience of people in complex mountainous environments. Although the traditional method such as AHP hierarchical analysis method can construct an evaluation system, it lacks the fusion and verification of multi-modal physiological data. Some patents propose a street image emotion analysis model, but they do not solve the problem of subjective and objective and spatial data time and space synchronization, and they do not involve the superimposed influence of mountainous terrain and historical environment on perception.

[0003] At present, the coupling analysis of sound environment parameters and visual perception mostly uses linear regression, which cannot construct an analysis model that adapts to the nonlinear spatial characteristics of mountainous streets. The existing demand response model relies too much on statistical mean, ignoring the perceptual difference characteristics of different groups of people, resulting in insufficient universality of optimization strategies.

[0004] With the improvement of the accuracy of wearable devices (such as Tobii Pro Glasses 3 eye tracker, electroencephalograph, and physiological electrical monitor), it is possible to establish a multi-modal data collection standard in outdoor complex environments. The development of deep learning technology provides a new idea for the fusion of nonlinear correlation and spatial multi-dimensional data. The industry urgently needs to construct an evaluation method that takes into account subjective and objective evaluation, adapts to mountainous terrain characteristics, and can generate differentiated improvement strategies. SUMMARY

[0005] The technical problem to be solved by the present application is to make up for the shortcomings of the prior art and provide a body perception evaluation method for a mountainous historical block based on multi-modal data.

[0006] To solve the above technical problems, the technical scheme of the present application is as follows: A body perception evaluation method for a mountainous historical block based on multi-modal data, comprising the following steps: S1: Select a mountainous historical block as an evaluation object, divide the public space of the mountainous historical block into multiple types, and select one experimental path in each type of public space, and select NT experimental stop points in each experimental path, 4≤NT≤6; S2: Constructing an evaluation index system, the evaluation index system including multiple physiological data, multiple psychological data and multiple spatial data, wherein the spatial data includes multiple characteristic data and multiple sound environment data; S3: For each experimental path, acquiring the characteristic data by combining field collection with paper statistics; S4: Recruiting a number of experimenters, asking the experimenters to walk on the experimental path and stop at the experimental stop points, collecting the sound environment data and the physiological data of each experimenter in real time on the experimental path and the experimental stop points, collecting the psychological data of the experimenters in real time at the experimental stop points through questionnaires; S5: For the data obtained on each experimental path, the following processing is performed: S51: First, synchronizing the sound environment data, the physiological data and the psychological data in time on the ErgoLAB DataLOG platform, then pre-processing the psychological data, the physiological data and the spatial data through statistical software to obtain effective psychological data, effective physiological data and effective spatial data: S52: Taking the effective spatial data as grouping variables, taking the effective psychological data and the effective physiological data as testing variables, performing K-W test one by one to obtain a significant variable relationship table; S53: Correspondingly arranging the psychological data, the physiological data and the spatial data in the significant variable relationship table to obtain an experimental path data set; S54: Defining the spatial data as independent variables and the psychological data and the physiological data as dependent variables, learning the relationship between the independent variables and the dependent variables based on the experimental path data set through a random forest model, and outputting a bar chart of the influence degree of each independent variable on each dependent variable.

[0007] Further, in step S1, the public space of the mountainous historical block is divided into four types, which are: square type public space, street type public space, courtyard type public space and terrace type public space.

[0008] Further, in step S1, the type classification method of the public space of the mountainous historical block is: first, pre-research the space characteristics and crowd usage habits of each public space in the mountainous historical block, the space characteristics include the parking function and the terrain characteristics, the crowd usage habits include the high-frequency access space and the high-flow road section, and the parking function, the terrain characteristics, the high-frequency access space and the high-flow road section of each public space are respectively rated according to high, medium and low three levels; define the square type public space: the public space with medium parking function, low terrain characteristics, high-frequency access space and high-flow road section; define the street type public space: the public space with low parking function, medium terrain characteristics, medium high-frequency access space and high-flow road section; define the courtyard type public space: the public space with high parking function, low terrain characteristics, medium high-frequency access space and high-flow road section; define the ladder type public space: the public space with low parking function, high terrain characteristics, medium high-frequency access space and high-flow road section.

[0009] Further, in step S1, the selection method of the experimental path is: randomly selecting 3 paths in each type of public space as the candidate path, the length of each candidate path is 40-70 meters, recording the video at a uniform speed on each candidate path, and then asking relevant professional experts to score the parking function, terrain characteristics and sound environment related features of the candidate path according to the content of the recorded video through a questionnaire, and score is summarized; in each type of public space, select the candidate path with the best score as the experimental path.

[0010] Further, in step S1, the sound environment related features include the visual features and the sound features; the visual features include the water surface, the temple, the characteristic old shop for more than 50 years, the historical landscape and the traditional culture signboard, and the sound features include the artificial sound, the natural sound, the intangible cultural heritage sound and the traffic sound; the selection method of the experimental parking point is: for a single experimental path, find all the space feature nodes on the experimental path, the space feature nodes include the branch intersection, the turning point and the terrain change position, wherein the terrain change refers to the position of the flat land changing into the sloping land, mark the number of visual features and sound features on each space feature node; first, select the space feature nodes with the number of visual features and sound features not less than 3 as the candidate nodes, and randomly select NT candidate nodes with adjacent distance greater than 6 meters as the experimental parking points in the multiple candidate nodes.

[0011] Further, the physiological data includes physiological electrical data, electroencephalogram data, and eye movement data; the physiological electrical data includes heart rate, heart rate variability, respiratory rate, and skin conductivity, which are collected by a wireless physiological recorder; the electroencephalogram data includes a-EEG, b-EEG, and b / a, which are collected by an electroencephalograph; the eye movement data includes fixation times, fixation time, and average blink times, which are collected by an eye tracker; the psychological data includes scores of audio-visual perception indicators and scores of comprehensive satisfaction evaluation obtained by a spatial perception and satisfaction questionnaire; the audio-visual perception indicators include visual perception indicators and auditory perception indicators; the visual perception indicators include diversity, orderliness, beauty, accessibility, comfort, fluctuation, crowding, recognizability, and openness; the auditory perception indicators include pleasantness, annoyance, quietness, disorder, liveliness, monotony, eventfulness, non-eventfulness, and diversity; the feature data includes morphological data, historical feature data, and path feature data; the morphological data includes slope, openness, and green view rate; the historical feature data includes the number of business types, the number of color types, and the number of material types; and the path feature data includes path length, platform area, and elevation difference; and the sound environment data includes group activity sound, conversation sound, bird chirping sound, insect chirping sound, water sound, traffic sound, broadcast sound, construction sound, and intangible cultural heritage sound, and the sound environment data is obtained by first recording and then performing semi-automatic extraction of sound source features on the recorded file by using Adobe Audition software.

[0012] Further, in step S4, the environmental conditions during the collection of the physiological data and the psychological data need to meet the following requirements: temperature of 18-25°C, humidity of 60-80% RH, and illumination of 5000-30000 lux.

[0013] Further, in step S51, the sound environment data, the physiological data, and the psychological data are first synchronized in time on an ErgoLAB DataLOG platform, and then the synchronized sound environment data, the synchronized physiological data, and the synchronized psychological data are exported. The synchronized psychological data is first preprocessed as follows: abnormal values are removed, missing values are interpolated, and Z-score standardization is performed, and then validity test and reliability test are performed on the standardized synchronized psychological data; if the standardized synchronized psychological data meets the test requirements, it is used as the valid psychological data; if not, step S4 needs to be re-executed. The feature data, the synchronized sound environment data, and the synchronized physiological data are further preprocessed as follows: abnormal values are removed, missing values are interpolated, automatic filtering and denoising are performed, and Min-Max normalization is performed, to obtain valid spatial data and valid physiological data.

[0014] Furthermore, in step S53, the experimental path dataset includes a physiological data variable pair dataset and a psychological data variable pair dataset. The physiological data variable pair dataset includes physiological data and spatial data compiled from the significant variable relationship table, and the psychological data variable pair dataset includes psychological data and spatial data compiled from the significant variable relationship table. In step S54, the physiological data variables are randomly divided into training set A and test set A according to a certain proportion, and the psychological data variables are randomly divided into training set B and test set B according to a certain proportion. The relationship between spatial data and physiological data is learned through random forest model A, and the relationship between spatial data and psychological data is learned through random forest model B. Then, random forest models A and B are tested using test sets A and B respectively. When the test results of random forest models A and B both satisfy R... 2 When the value is greater than 0.5, both models are considered to have reasonable predictive ability. Then, Random Forest Model A and Random Forest Model B output bar charts showing the influence of each variable on each dependent variable.

[0015] The beneficial effects that this invention can achieve are as follows: Multimodal data fusion: Integrating multi-dimensional data such as physiological, psychological, spatial and behavioral trajectories to comprehensively and objectively reflect the user's embodied perception of place.

[0016] Adapting to the nonlinear characteristics of mountainous terrain: By employing a random forest model, we can effectively analyze the nonlinear correlation between mountainous terrain and embodied perception, as well as the complex interactions between variables, revealing the unique influence mechanism of mountainous space.

[0017] Scientifically guided design updates: Optimization strategies generated based on multimodal data analysis and demand response models are more scientific and targeted, effectively improving the spatial quality, cultural identity, and user satisfaction of mountainous historical districts.

[0018] Improve the efficiency and accuracy of assessments: A standardized and operable evaluation process and data processing workflow are provided. Compared with traditional methods, the assessment results are more accurate and convincing, providing a reliable basis for the protection and renewal of mountainous historic districts. Attached Figure Description

[0019] Figure 1 This is a planar schematic diagram of the experimental path in an embodiment of the present invention.

[0020] Figure 2 This is a framework diagram of physiological data in an embodiment of the present invention.

[0021] Figure 3 This is a framework diagram of psychological data in an embodiment of the present invention.

[0022] Figure 4This is a framework diagram of spatial data in an embodiment of the present invention.

[0023] Figure 5 This is a spatial perception and satisfaction questionnaire in an embodiment of the present invention.

[0024] Figure 6 This is a bar graph showing the influence of individual variables on heart rate in street-type public spaces, as output by an embodiment of the present invention.

[0025] Figure 7 This is a bar graph showing the influence of individual variables on diversity in street-type public spaces, as output by an embodiment of the present invention. Detailed Implementation

[0026] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0027] This embodiment takes the Ciqikou neighborhood as an example to describe in detail a method for embodied perception evaluation of mountainous historical neighborhoods based on multimodal data, including the following steps: S1: Select the Ciqikou Mountain Historical District as the evaluation object, divide the public space of the mountain historical district into multiple types, select one experimental path in each type of public space, and select NT experimental stops in each experimental path. In this embodiment, NT is selected as 5.

[0028] Specifically, the public spaces in the Ciqikou Mountain Historical District are divided into four types: plaza-type public spaces, street-type public spaces, courtyard-type public spaces, and terrace-type public spaces.

[0029] Specifically, the method for classifying the public spaces in the Ciqikou Mountain Historical District is as follows: First, a preliminary survey is conducted on the spatial characteristics and usage habits of each public space in the Ciqikou Mountain Historical District.

[0030] Spatial features include parking functions and topographical characteristics.

[0031] Stopping function: refers to whether people can easily stay and gather in the space to enjoy the view or have conversations.

[0032] Topographic features: Observe whether the space has mountainous environmental features such as steps, retaining walls, or slopes.

[0033] User habits include frequently accessing spaces and high-traffic areas.

[0034] High-frequency access space: derived from semi-structured interview data.

[0035] High-traffic sections: determined through behavioral annotation.

[0036] The public spaces are rated according to three levels: high, medium, and low, based on their parking function, terrain features, high-frequency access areas, and high-traffic sections. Then, according to Table 1 below, they are defined as plaza-type public spaces, street-type public spaces, courtyard-type public spaces, and stepped public spaces.

[0037] Table 1: Classification of Public Spaces Specifically, the experimental paths were selected as follows: three paths were randomly chosen from each type of public space as candidate paths, each with a length of 40–70 meters. Videos were recorded at a constant speed along each candidate path. The visual and acoustic environmental characteristics were defined as visual and acoustic features. Visual features included water surfaces, temples, shops with over 50 years of history, historical landscapes, and traditional cultural road signs. Acoustic features included artificial sounds, natural sounds, sounds from intangible cultural heritage sites, and traffic sounds. Then, relevant experts used a questionnaire to score the candidate paths based on the recorded videos, evaluating three aspects: stopping function, terrain features, and visual and acoustic environmental characteristics. Each aspect was scored out of 10, for a total of 30 points. The scores were then tallied. In each type of public space, the candidate path with the highest total score was selected as the experimental path, ultimately determining four experimental paths.

[0038] Specifically, the method for selecting experimental stopping points is as follows: For a single experimental path, all spatial feature nodes on the path are identified. Spatial feature nodes include intersections, turning points, and locations of terrain changes, where terrain changes refer to locations where flat land changes into slopes. The number of visual and acoustic features at each spatial feature node is marked. First, spatial feature nodes with at least 3 visual and acoustic features are selected as candidate nodes. From these candidate nodes, 5 nodes with an adjacent distance greater than 6 meters are randomly selected as the experimental stopping points.

[0039] Figure 1 The diagram shows the experimental path selected in the street-type public space, namely Path A, which includes a1-a5, a total of 5 experimental stopping points. Figure 1 The diagram shows the experimental path selected in the stepped public space, namely Path B, which includes five experimental stopping points, b1-b5.

[0040] S2: Construct an evaluation index system, which includes various physiological data, various psychological data, and various spatial data. The spatial data includes various characteristic data and various acoustic environment data.

[0041] Specifically, physiological data includes physiological electrical data, electroencephalogram (EEG) data, and eye movement data.

[0042] Physiological electrical data include heart rate (HR), heart rate variability (LFa / HFa), respiratory rate (RESP), and skin conductivity (SCL).

[0043] EEG data: including α-EEG, β-EEG, and β / α.

[0044] Eye movement data include fixation counts (FC), fixation duration (TFD), and average blink counts (n / s).

[0045] Specifically, the psychological data includes audiovisual perception scores and overall satisfaction scores obtained through spatial perception and satisfaction questionnaires.

[0046] The audiovisual perception indicators include visual perception indicators and auditory perception indicators, as shown in Table 2.

[0047] Visual perception indicators include: diversity, orderliness, aesthetics, accessibility, comfort, undulation, crowding, recognizability, and openness.

[0048] Auditory perception indicators include: pleasantness, annoyance, tranquility, disorder, vitality, monotony, eventfulness, absence of events, and diversity.

[0049] Table 2: Audiovisual Perception Indicators Specifically, the feature data includes morphological data, historical feature data, and path feature data.

[0050] Morphological data include slope, openness, and green view rate.

[0051] Historical characteristic data includes the number of business types, the number of colors, and the number of materials.

[0052] The path feature data includes path length, area of ​​stopping platforms, and elevation difference.

[0053] Specifically, the sound environment data includes group activity sounds, conversation sounds, birdsong sounds, insect chirping sounds, water sounds, traffic sounds, broadcast sounds, construction sounds, and sounds from intangible cultural heritage.

[0054] S3: For each experimental path, feature data is obtained by combining on-site collection with statistical analysis of drawings.

[0055] In this embodiment, steps S4-S5 are demonstrated by describing an experimental path in a street-type public space.

[0056] S4: Recruit several participants and have them walk along the experimental path and stop at experimental rest points. Collect real-time acoustic environment data and physiological data of each participant along the experimental path and at the experimental rest points. Collect real-time psychological data from the participants at the experimental rest points through questionnaires.

[0057] Specifically, the environmental conditions during the collection of physiological and psychological data must meet the following requirements: temperature 18–25°C, humidity 60–80% RH, and illuminance 5000–30000 lux.

[0058] The acoustic environment data is obtained by first recording audio and then semi-automatically extracting sound source features from the audio files using Adobe Audition software. The quantitative data includes: LAeq (equivalent sound level) and the percentage of sound source duration (such as the percentage of insect chirping ICS).

[0059] Physiological data acquisition: Eye movement data, including fixation counts (FC), fixation duration (TFD), and average blink counts (n / s), were captured using a Tobii Pro eye tracker. Skin conductance (SCL), heart rate variability (LFa / HFa), heart rate (HR), and respiratory rate (RESP) were acquired using an ErgoLAB wireless physiological recorder. α-EEG, β-EEG, and β / α brainwave signals were simultaneously acquired using an ErgoLAB EEG electroencephalogram (EEG) device. The observer's field of view was maintained between 120° and 180°, and the dwell time at each node was no less than 15 seconds.

[0060] Psychological data collection: At each experimental stop, a spatial perception and satisfaction questionnaire was used for immediate evaluation, which included 18 audiovisual perception indicators (using a 7-point Likert scale), such as... Figure 5 As shown, a comprehensive satisfaction evaluation score is given at the end of the experimental path.

[0061] S5: The data obtained from each experimental path is processed as follows.

[0062] S51: First, synchronize the acoustic environment data, physiological data, and psychological data on the ErgoLAB DataLOG platform. Then, use statistical software to preprocess the psychological data, physiological data, and spatial data to obtain effective psychological data, effective physiological data, and effective spatial data.

[0063] Specifically, the acoustic environment data, physiological data, and psychological data are first synchronized in time on the ErgoLAB DataLOG platform, and then the synchronized acoustic environment data, synchronized physiological data, and synchronized psychological data are exported.

[0064] The statistical software used can be SPSS, Stata, Python, Jamovi, JASP, Minitab, PSPP, etc.

[0065] The synchronous psychological data is first preprocessed as follows: outlier removal, missing value imputation, and Z-score standardization are performed sequentially. Then, the standardized synchronous psychological data is tested for validity and reliability. If the standardized synchronous psychological data meets the test requirements, it is considered as valid psychological data. If it does not meet the requirements, step S4 needs to be repeated.

[0066] The validity test passed if the KMO value was greater than 0.6 and the Bartlett p-value was less than 0.05. This indicates that the data satisfies the homogeneity of variance assumption, which means the scale data is valid and can be further tested for reliability.

[0067] The reliability test passed with a Cronbach's α ≥ 0.7. This indicates that the spatial perception and satisfaction questionnaire has high internal consistency and the measurement results are reliable.

[0068] The feature data, synchronous acoustic environment data, and synchronous physiological data are then preprocessed as follows: outlier removal, missing value imputation, automatic filtering and noise reduction, and Min-Max normalization are performed sequentially to obtain effective spatial data and effective physiological data.

[0069] S52: Using statistical software, take the effective spatial data as the grouping variable and the effective psychological data and effective physiological data as the test variables, and perform KW tests one by one to obtain a table of significant variable relationships.

[0070] Based on the results of the Kruskal-Wallis test, pairs of variables whose significance level (Kruskal-Wallis p-value) satisfies the condition Kruskal-Wallis p-value < 0.05 are selected. This condition means that the spatial data grouping variable has a statistically significant impact on the median of the corresponding physiological or psychological test variable. A table of significant variable relationships is then output, as shown in Table 3. Table 3: Significance of Relationships Between Significant Variables In Table 3, * indicates Kruskal-Wallis p-value < 0.05, and ** indicates Kruskal-Wallis p-value < 0.01.

[0071] S53: Organize the psychological, physiological and spatial data in the significant variable relationship table to obtain the experimental path dataset.

[0072] Specifically, the experimental path dataset includes a dataset of physiological data variable pairs and a dataset of psychological data variable pairs.

[0073] The physiological data variable pair dataset includes physiological and spatial data compiled from the significant variable relationship table, as shown in Tables 4-1 to 4-3.

[0074] Table 4-1: Illustration of physiological data variables on dataset (Part 1) Table 4-2: Illustration of physiological data variables on datasets (Part 2) Table 4-3: Illustration of physiological data variables on datasets (Part 3) The psychological data variable pair dataset includes psychological and spatial data compiled from the significant variable relationship table, as shown in Tables 5-1 to 5-3.

[0075] Table 5-1: Schematic diagram of psychological data variables on dataset (Part 1) Table 5-2: Schematic diagram of psychological data variables on datasets (Part 2) Table 5-3: Schematic diagram of psychological data variables on datasets (Part 3) S54: Define the independent variables as spatial data and the dependent variables as psychological and physiological data. Based on the experimental path dataset, learn the relationship between the independent and dependent variables through a random forest model, and output a bar chart showing the degree of influence of each independent variable on each dependent variable.

[0076] Specifically, the physiological data variables are randomly divided into training set A and test set A according to a certain proportion, and the psychological data variables are randomly divided into training set B and test set B according to a certain proportion. Random forest model A learns the relationship between spatial and physiological data, and random forest model B learns the relationship between spatial and psychological data. Then, test sets A and B are used to test random forest models A and B respectively. When the test results of random forest models A and B both satisfy R² > 0.5, they are considered to have reasonable predictive ability. After ensuring the effectiveness of the models, SHAP analysis is performed on random forest models A and B respectively, and bar charts of the influence of each independent variable on each dependent variable are output, such as... Figure 6 and Figure 7 As shown.

[0077] Based on the bar chart of the influence of each independent variable on each dependent variable, the influence of spatial data on physiological or psychological data is ranked. The greater the influence, the more important it is. Positive values ​​indicate a positive influence, and negative values ​​indicate a negative influence. This can then be used to guide the renovation of the mountainous historical district.

[0078] According to Figure 6 It can be seen that the key positive factors for improving tourists' enjoyment and physiological relaxation (such as reducing HR) are, in order: color richness, green view rate, and business diversity; the main negative factors are: traffic and broadcast noise, as well as excessive road elevation differences.

[0079] Based on this, subsequent renovations should revolve around three main strategies: "enhancing color schemes, gentler slopes, and quieter alleyways." Color Refinement and Activation: The positive impact of color richness is applied to design. By extracting and utilizing the historical color spectrum of the neighborhood, a unified tone is achieved for building facades, with accent colors added locally. This enhances visual appeal while maintaining overall harmony, thereby increasing environmental attractiveness.

[0080] Path smoothing design: To address the negative experience caused by excessive elevation differences, the steep sections of the path are being significantly improved. By adding gentle ramps, rest platforms, and safety handrails, the continuous elevation differences are broken down into rhythmic walking segments. This ensures both safety and comfort while transforming unfavorable terrain into scenic viewpoints where visitors can stop.

[0081] Noise control and soundscape optimization: Aiming for "quiet alleyways," noise from vehicles and construction will be limited. Simultaneously, the green space ratio at key nodes will be increased, such as by adding micro-courtyards and pocket parks, and introducing natural sounds (like flowing water) to mask residual noise. In terms of business formats, a distinctive, rich, yet not overly noisy commercial atmosphere will be fostered, comprehensively creating a pleasant neighborhood environment.

[0082] The above description is only one embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the spirit of the present invention are within the scope of protection of the present invention.

Claims

1. A method for embodied perception evaluation of mountainous historic blocks based on multi-modal data, characterized in that: The method comprises the following steps: S1: selecting a mountainous historical block as an evaluation object, dividing the public space of the mountainous historical block into multiple types, selecting one experimental path in each type of public space, and selecting NT experimental stop points in each experimental path, 4≤NT≤6; S2: constructing an evaluation index system, the evaluation index system comprising multiple physiological data, multiple psychological data, and multiple spatial data, wherein the spatial data comprises multiple characteristic data and multiple sound environment data; S3: for each experimental path, acquiring the characteristic data by combining field collection with paper statistics; S4: recruiting a plurality of experimenters, asking the experimenters to walk on the experimental path and stop at the experimental stop points, collecting sound environment data and physiological data of each experimenter in real time on the experimental path and the experimental stop points, and collecting psychological data of the experimenters in real time at the experimental stop points through questionnaires; S5: processing the data obtained in each experimental path as follows: S51: first, synchronizing the sound environment data, physiological data, and psychological data on the ErgoLAB DataLOG platform, and then preprocessing the psychological data, physiological data, and spatial data through statistical software to obtain effective psychological data, effective physiological data, and effective spatial data: S52: taking the effective spatial data as a grouping variable, taking the effective psychological data and effective physiological data as test variables, and performing K-W test one by one to obtain a significant variable relationship table; S53: corresponding to the psychological data, physiological data, and spatial data in the significant variable relationship table, obtaining an experimental path data set; S54: defining the independent variables as spatial data, the dependent variables as psychological data and physiological data, learning the relationship between the independent variables and the dependent variables based on the experimental path data set through a random forest model, and outputting a bar chart of the influence of each independent variable on each dependent variable.

2. The method of claim 1, wherein the method further comprises: In step S1, the public space of the mountainous historical block is divided into four types, namely, square-type public space, street-type public space, courtyard-type public space, and terrace-type public space.

3. The method of claim 2, wherein the method is a method of embodied perception evaluation of a mountainous historical block based on multi-modal data, The characteristic is that: in step S1, the type division method of the public space of the mountainous historical block is: first, pre-researching the spatial characteristics and crowd usage habits of each public space in the mountainous historical block, the spatial characteristics including stop function and terrain characteristics, and the crowd usage habits including high-frequency access space and high-flow road section, respectively rating the stop function, terrain characteristics, high-frequency access space, and high-flow road section of each public space according to high, medium, and low levels; defining the square-type public space: the public space with medium stop function, low terrain characteristics, high high-frequency access space, and medium high-flow road section; defining the street-type public space: the public space with low stop function, medium terrain characteristics, medium high-frequency access space, and high high-flow road section; defining the courtyard-type public space: the public space with high stop function, low terrain characteristics, medium high-frequency access space, and medium high-flow road section; and defining the terrace-type public space: the public space with low stop function, high terrain characteristics, medium high-frequency access space, and medium high-flow road section.

4. The method of claim 1, wherein the method further comprises: In step S1, the selection method of the experimental path is: randomly selecting 3 paths in each type of public space as candidate paths, the length of each candidate path is 40-70 meters, recording video at a constant speed on each candidate path, then asking relevant professional experts to score the stop function, terrain characteristics, and visual and acoustic environment related features of the candidate paths according to the content of the recorded video through a questionnaire, and to aggregate the scores; in each type of public space, the candidate path with the best aggregated score is selected as the experimental path.

5. The embodied perception evaluation method for mountainous historical districts based on multimodal data according to claim 4, characterized in that: In step S1, the visual and acoustic environment related features include visual features and acoustic features; the visual features include water surface, temple, characteristic old shop for more than 50 years, historical landscape and traditional culture signboard, and the acoustic features include artificial sound, natural sound, intangible cultural heritage sound and traffic sound; the selection method of the experimental stop point is: for a single experimental path, finding all the spatial feature nodes on the experimental path, the spatial feature nodes include branch intersection, turning point and terrain change position, wherein the terrain change refers to the position where the flat ground changes into the sloping ground, marking the number of visual features and acoustic features on each spatial feature node; first selecting the spatial feature nodes with the number of visual features and acoustic features both not less than 3 as candidate nodes, and randomly selecting NT candidate nodes with adjacent distance greater than 6 meters as the experimental stop points.

6. The method of claim 1, wherein the method further comprises: The physiological data includes physiological electrical data, electroencephalogram data and eye movement data; the physiological electrical data includes heart rate, heart rate variability, respiratory rate and skin conductivity, which are collected by a wireless physiological recorder; the electroencephalogram data includes α-EEG, β-EEG and β / α, which are collected by an electroencephalograph; the eye movement data includes fixation times, fixation time and average blink times, which are collected by an eye tracker; the psychological data includes the scoring of audiovisual perception indicators and the comprehensive evaluation of satisfaction obtained through a spatial perception and satisfaction questionnaire; the audiovisual perception indicators include visual perception indicators and auditory perception indicators; the visual perception indicators include diversity, orderliness, beauty, accessibility, comfort, undulation, crowding, recognizability and openness; the auditory perception indicators include pleasantness, annoyance, tranquility, disorder, vitality, monotony, event, no event and diversity; the feature data includes morphological data, historical feature data and path feature data; the morphological data includes slope, openness and greenness rate; the historical feature data includes the number of business types, the number of color types and the number of material types; the path feature data includes path length, platform area and elevation difference; the acoustic environment data includes group activity sound, conversation sound, bird chirping sound, insect chirping sound, water sound, traffic sound, broadcast sound, construction sound and intangible cultural heritage sound, which are obtained by first recording and then semi-automatically extracting the sound source features of the recorded files through Adobe Audition software.

7. The method of claim 1, wherein the method further comprises: In step S4, the environmental conditions during the collection of physiological data and psychological data need to meet the following requirements: temperature 18-25°C, humidity 60-80% RH, and illuminance 5000-30000 lux.

8. The method of claim 1, wherein the method further comprises: In step S51, the sound environment data, physiological data and psychological data are first time-synchronized on the ErgoLAB DataLOG platform, and then the synchronized sound environment data, synchronized physiological data and synchronized psychological data are exported; The synchronized psychological data is first pre-processed as follows: the abnormal values are first removed, the missing values are then interpolated, and the Z-score standardization processing is performed, and then the standardized synchronized psychological data is subjected to validity test and reliability test; if the standardized synchronized psychological data meets the test requirements, it is taken as the effective psychological data; if not, step S4 needs to be re-executed; The feature data, synchronized sound environment data and synchronized physiological data are further pre-processed as follows: the abnormal values are first removed, the missing values are then interpolated, the automatic filtering denoising is performed, and the Min-Max normalization is performed, to obtain the effective spatial data and effective physiological data.

9. The method of claim 1, wherein the method further comprises: In step S53, the experimental path data set includes a physiological data variable pair data set and a psychological data variable pair data set, the physiological data variable pair data set includes physiological data and spatial data sorted out from the significant variable relationship table, and the psychological data variable pair data set includes psychological data and spatial data sorted out from the significant variable relationship table; In step S54, the physiological data variable pair dataset is randomly divided into a training set A and a test set A according to a certain proportion, and the psychological data variable pair dataset is randomly divided into a training set B and a test set B according to a certain proportion; the relationship between the spatial data and the physiological data is learned through a random forest model A, and the relationship between the spatial data and the psychological data is learned through a random forest model B; then the test set A and the test set B are respectively used to test the random forest model A and the random forest model B, and when the test results of the random forest model A and the random forest model B both satisfy R 2 > 0.5, it is considered that both have reasonable prediction ability, and then the random forest model A and the random forest model B respectively output the influence degree bar chart of each variable pair on each dependent variable.

Images