A landscape sequence design method based on human mental health benefits

Abstract

The application discloses a landscape sequence design method based on human psychological health benefits, and relates to the technical field of landscape planning; the method comprises the following steps: S1, obtaining landscape sequence data, wherein the landscape sequence data comprises visual data, auditory data and sequence data of each scenic spot and a psychological health benefit score; S2, performing standardization processing on the visual data, the auditory data and the sequence data, screening important features by using a recursive feature elimination (RFE) method, taking the important features as independent variables, taking the psychological health benefit score as a dependent variable, setting hyperparameters, and obtaining a trained random forest model; S3, recording positions of all scenic spots and road positions, and calculating N shortest landscape sequence paths without repetition based on a Dijkstra algorithm; and S4, inputting the landscape sequence paths into the trained random forest model to obtain corresponding psychological health benefit scores, and taking a landscape sequence path with the highest score as the best one; the method is high in efficiency and good in effect.

Description

Technical Field

[0001] This invention relates to the field of landscape planning technology, and in particular to a landscape sequence design method based on the psychological health benefits of humans. Background Technology

[0002] With the acceleration of urbanization, mental health issues among urban residents are becoming increasingly prominent. Studies have shown that visual and auditory experiences in the natural environment have a significant positive impact on mental health. However, current research on methods for systematically organizing landscape sequences in landscape planning to maximize mental health benefits remains limited. Existing techniques mostly rely on expert experience or simple optimization of single variables, lacking methods that comprehensively consider visual, auditory, and path characteristics. Summary of the Invention

[0003] This invention provides a landscape sequence design method based on the psychological health benefits of humans, solving the technical problem of low efficiency in obtaining the optimal landscape sequence path.

[0004] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0005] A landscape sequence design method based on the benefits of human mental health includes the following steps:

[0006] Step S1: Obtain landscape sequence data, which includes visual data, auditory data, sequence data, and mental health benefit score for each scenic spot. The visual data includes panoramic images, green view rate, canopy density, and sky visibility factor. The auditory data includes sound data, sound type, sound pressure level, and sound spectrum. The sequence data includes the location of each scenic spot, road location, distance between scenic spots, and the order of each path. The mental health benefit score is the pleasure score of all scenic spots under each path.

[0007] Step S2: Standardize the visual data, auditory data, and sequence data to obtain standardized visual data, auditory data, and sequence data; use the RFE method based on recursive feature elimination to select important features from the standardized visual data, auditory data, and sequence data; use the selected important features as independent variables, the mental health benefit score as the dependent variable, set hyperparameters, train the random forest model, and obtain the trained random forest model.

[0008] Step S3: Record the location information of all scenic spots and roads, including longitude and latitude coordinates. Calculate the N shortest landscape sequence paths that connect all scenic spots without repetition based on the Dijkstra algorithm.

[0009] Step S4: Input the data of the N landscape sequence paths into the trained random forest model to obtain the mental health benefit scores of all landscape sequence paths. Compare the mental health benefit scores of each landscape sequence path and select the landscape sequence path with the highest score as the best sequence.

[0010] A further technical solution is as follows: in step S1, the scoring range is 0 to 10 points; in step S3, GIS software is used to record the location information of all scenic spots and roads, N=5, to obtain the five shortest landscape sequence paths.

[0011] A further technical solution is that step S1 specifically includes the following steps:

[0012] Step S101: Obtain visual data for each attraction;

[0013] Step S102: Obtain auditory data for each attraction;

[0014] Step S103: Obtain sequence data;

[0015] Step S104: Obtain the pleasure rating of all attractions along each route by having volunteers fill out questionnaires, and use the pleasure rating as a psychological health benefit rating.

[0016] A further technical solution is that step S101 specifically includes the following steps:

[0017] Step S1011: Use a 360-degree panoramic camera to take pictures of each scenic spot and generate a high-resolution panoramic image of each scenic spot;

[0018] Step S1012: Using the U-Net image segmentation algorithm based on deep learning, extract the proportion of green pixels from the panoramic image of each scenic spot and calculate the green view rate;

[0019] Step S1013: By analyzing the vegetation coverage area in the panoramic image of each scenic spot, the vegetation coverage area ratio is calculated using the Mask R-CNN object detection algorithm to obtain the canopy closure.

[0020] Step S1014: Using a sky region segmentation algorithm, the sky visibility factor is calculated by extracting the proportion of the sky region in the panoramic image of each scenic spot.

[0021] A further technical solution is that step S102 specifically includes the following steps:

[0022] Step S1021: Set up a high-sensitivity microphone at each scenic spot to collect sound data for each scenic spot;

[0023] Step S1022: Input the sound data of each scenic spot obtained in step S1021 into the audio classification model based on convolutional neural network to identify different sound types;

[0024] Step S1023: Measure the sound pressure level of each scenic spot in real time using a sound level meter, in dB;

[0025] Step S1024: Perform Fourier transform (FFT) on the sound data of each scenic spot obtained in step S1021 to extract the sound spectrum data and analyze the sound frequency distribution characteristics.

[0026] A further technical solution is that step S103 specifically includes the following steps:

[0027] Step S1031: Use a GPS device to record and obtain the location of each attraction;

[0028] Step S1032: Mark the roads between attractions in the GIS software, obtain the road locations, calculate the distances between attractions, and record the order in which each path is obtained.

[0029] A further technical solution is that, in step S2, hyperparameters are set as follows:

[0030] The number of trees, n_estimators, can be selected from 100 to 500.

[0031] Maximum depth (max_depth): Automatically adjusted to optimize model performance;

[0032] Minimum number of split samples (min_samples_split): Set to 2-5 depending on the amount of data.

[0033] The beneficial effects of adopting the above technical solution are as follows:

[0034] A landscape sequence design method based on the psychological health benefits of human beings includes the following steps: Step S1: Obtain landscape sequence data, which includes visual data, auditory data, sequence data, and psychological health benefit scores for each scenic spot; Step S2: Standardize the visual data, auditory data, and sequence data, filter important features using the Recursive Feature Emission (RFE) method and use them as independent variables, with the psychological health benefit score as the dependent variable, set hyperparameters, and obtain a trained random forest model; Step S3: Record the location information of all scenic spots and roads, and calculate the N shortest landscape sequence paths that connect all scenic spots without repetition based on the Dijkstra algorithm; Step S4: Input the data of the N landscape sequence paths into the trained random forest model respectively to obtain the psychological health benefit scores of all corresponding landscape sequence paths, and select the landscape sequence path with the highest score as the optimal sequence. This method is highly efficient and effective in obtaining the optimal landscape sequence path.

[0035] See the detailed implementation section for further description. Attached Figure Description

[0036] Figure 1 This is a flowchart of the present invention;

[0037] Figure 2 yes Figure 1 Flowchart of step S1;

[0038] Figure 3 yes Figure 2 Flowchart of step S101;

[0039] Figure 4 yes Figure 2 The flowchart of step S102. Detailed Implementation

[0040] The purpose of this application is:

[0041] The aim is to provide a scientific and easy-to-use method to optimize the organization of landscape sequences based on visual, auditory, and path data, thereby maximizing the psychological benefits for human beings. By combining modeling and algorithms, the scientific rigor and practicality of landscape planning are improved.

[0042] This involves the intersection of mental health and landscape planning, aiming to optimize the landscape experience to enhance mental health benefits.

[0043] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit this application or its application or use. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0044] Many specific details are set forth in the following description in order to provide a full understanding of this application. However, this application may also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the spirit of this application. Therefore, this application is not limited to the specific embodiments disclosed below.

[0045] like Figure 1 As shown, this invention discloses a landscape sequence design method based on the benefits of human mental health, comprising the following steps:

[0046] Step S1: Obtain landscape sequence data.

[0047] Landscape sequence data was obtained, including visual, auditory, and sequence data for each scenic spot, as well as a mental health benefit score. Visual data included panoramic images, green view ratio, canopy density, and sky visibility factor. Auditory data included sound data, sound type, sound pressure level, and sound spectrum. Sequence data included the location of each scenic spot, road location, distance between scenic spots, and the order of each path. The mental health benefit score was the pleasure score of all scenic spots under each path, with a score range of 0 to 10.

[0048] Details are as follows:

[0049] like Figure 2 , Figure 3 and Figure 4 As shown, step S1 is specifically divided into the following steps:

[0050] Step S101: Obtain visual data.

[0051] Obtain visual data for each attraction, including panoramic images, green view rate, canopy density, and sky visibility factor.

[0052] Step S1011: Panoramic shooting.

[0053] Each scenic spot is photographed using a 360-degree panoramic camera, generating a high-resolution panoramic image of each spot.

[0054] Step S1012: Calculate the green view rate.

[0055] The green view percentage is calculated by extracting the proportion of green pixels from the panoramic image of each scenic spot using the U-Net image segmentation algorithm based on deep learning.

[0056] Step S1013: Calculation of canopy closure.

[0057] By analyzing the vegetation cover area in the panoramic image of each scenic spot, the proportion of vegetation cover area is calculated using the Mask R-CNN object detection algorithm to obtain the canopy closure.

[0058] Step S1014: Sky visibility factor.

[0059] The sky region segmentation algorithm is used to calculate the sky visibility factor by extracting the proportion of the sky region in the panoramic image of each scenic spot.

[0060] Step S102: Obtain auditory data.

[0061] Acoustic data is obtained for each attraction, including sound data, sound type, sound pressure level, and sound spectrum.

[0062] Step S1021: Sound acquisition.

[0063] High-sensitivity microphones were set up at each attraction to collect sound data for each location.

[0064] Step S1022: Sound type.

[0065] The sound data of each scenic spot obtained in step S1021 is input into an audio classification model based on a convolutional neural network to identify different sound types, including birdsong, flowing water, and mechanical noise.

[0066] Step S1023: Sound pressure level.

[0067] The sound pressure level of each scenic spot is measured in real time using a sound level meter, and the unit is dB.

[0068] Step S1024: Sound spectrum.

[0069] Perform Fourier transform (FFT) on the sound data of each scenic spot obtained in step S1021 to extract the sound spectrum data and analyze the sound frequency distribution characteristics.

[0070] Step S103: Obtain sequence data.

[0071] Obtain sequence data, which includes the location of each attraction, road location, distance between attractions, and the order of each path.

[0072] Step S1031: Location of the attraction.

[0073] The location of each attraction was recorded using a GPS device.

[0074] Step S1032: Path sequence.

[0075] In GIS software, mark the roads between attractions, obtain the road locations, calculate the distances between attractions, and record the order in which each path is obtained.

[0076] Step S104: Obtain a mental health benefit score.

[0077] The pleasure rating of all attractions along each route was obtained by volunteers filling out questionnaires. The rating ranged from 0 to 10 points, and the pleasure rating was used as a psychological health benefit rating.

[0078] Step S2: Construct and obtain the random forest model.

[0079] The visual, auditory, and sequence data obtained in step S1 are standardized to obtain standardized visual, auditory, and sequence data. Important features are selected from the standardized visual, auditory, and sequence data using the RFE method based on recursive feature elimination. The selected important features are used as independent variables, and the mental health benefit score obtained in step S1 is used as the dependent variable. Hyperparameters are set, and a random forest model is trained to obtain a well-trained random forest model.

[0080] Details are as follows:

[0081] Step S2 specifically includes the following steps:

[0082] Step S201: Data preprocessing.

[0083] Visual, auditory, and sequence data are standardized to obtain standardized visual, auditory, and sequence data, thereby eliminating dimensional differences.

[0084] Important features are obtained from standardized visual, auditory, and sequence data using the RFE method based on recursive feature elimination.

[0085] Step S202: Random Forest Model Construction.

[0086] Using the key features obtained in step S201 as independent variables and the mental health benefit score obtained in step S104 as the dependent variable, hyperparameters are set to obtain a random forest model.

[0087] Set hyperparameters:

[0088] The number of trees, n_estimators, can be selected from 100 to 500.

[0089] Maximum depth (max_depth): Automatically adjusted to optimize model performance.

[0090] Minimum number of split samples (min_samples_split): Set to 2-5 depending on the amount of data.

[0091] Use cross-validation to evaluate model performance and ensure the model's generalization ability.

[0092] Step S3: Generate landscape sequence paths.

[0093] Using GIS software, the location information of all scenic spots and roads in the case study was recorded, including longitude and latitude coordinates. Based on Dijkstra's algorithm, the five shortest landscape sequence paths that can connect all scenic spots without repetition were calculated.

[0094] Step S4: Benefit scoring and optimal sequence selection.

[0095] The data of the five generated landscape sequence paths are input into the trained random forest model to obtain the mental health benefit scores of all landscape sequence paths. The mental health benefit scores of each landscape sequence path are compared, and the landscape sequence path with the highest score is taken as the best sequence.

[0096] In addition to the above embodiments, the optimal sequence can also be obtained by comparing three landscape sequence paths.

[0097] In addition to the above embodiments, it is also possible to obtain the optimal sequence by comparing seven landscape sequence paths.

[0098] You can choose how many landscape sequence paths to compare as needed.

[0099] advantage:

[0100] 1. Highly scientific: It utilizes machine learning methods to comprehensively consider multidimensional data and predict mental health benefits.

[0101] 2. High practicality: The trained model is a general-purpose model, applicable to different types of scenic spots or parks. Subsequent simple landscape data collection processes can be used to calculate the health benefits of the scenic area, eliminating the need to recruit a large number of volunteers to fill out questionnaires each time.

[0102] 3. Enhance health benefits: Provide quantifiable scientific evidence for urban planning and landscape design, and help improve public mental health.

[0103] Specific implementation examples:

[0104] Take a city park as an example:

[0105] 1. Data Acquisition:

[0106] Visual and auditory data were recorded for 10 major attractions within the park, as follows:

[0107] Visual data: green visibility rate 0.3-0.8, canopy closure 0.2-0.7, sky visibility factor 0.4-0.9.

[0108] Auditory data: sound types include birdsong, flowing water and mechanical noise, sound pressure level 50-70 dB, sound spectrum distribution.

[0109] The survey involved 20 volunteers who rated their level of enjoyment, recording the ratings at 10 points along each route.

[0110] 2. Feature Engineering and Model Building:

[0111] The above data was used to train a random forest model, and the model's prediction accuracy was evaluated through cross-validation.

[0112] 3. Landscape sequence path generation:

[0113] Five landscape sequences containing the aforementioned 10 attractions were generated using Dijkstra's algorithm.

[0114] 4. Benefit scoring and optimal sequence selection:

[0115] Inputting the random forest model yielded mental health benefit scores of 8.5, 7.2, 9.1, 8.0, and 8.7.

[0116] Results: The best landscape sequence with a score of 9.1 was selected for recommendation.

Claims

1. A landscape sequence design method based on the benefits of human mental health, characterized in that: Includes the following steps, Step S1: Obtain landscape sequence data, which includes visual data, auditory data, sequence data, and mental health benefit score for each scenic spot. The visual data includes panoramic images, green view rate, canopy density, and sky visibility factor. The auditory data includes sound data, sound type, sound pressure level, and sound spectrum. The sequence data includes the location of each scenic spot, road location, distance between scenic spots, and the order of each path. The mental health benefit score is the pleasure score of all scenic spots under each path. Step S2: Standardize the visual data, auditory data, and sequence data to obtain standardized visual data, auditory data, and sequence data; use the RFE method based on recursive feature elimination to filter out important features from the standardized visual data, auditory data, and sequence data; The key features obtained through screening are used as independent variables, and the mental health benefit score is used as the dependent variable. Hyperparameters are set, and a random forest model is trained to obtain a well-trained random forest model. Step S3: Record the location information of all scenic spots and roads, including longitude and latitude coordinates. Calculate the N shortest landscape sequence paths that connect all scenic spots without repetition based on the Dijkstra algorithm. Step S4: Input the data of the N landscape sequence paths into the trained random forest model to obtain the mental health benefit scores of all landscape sequence paths, and take the landscape sequence path with the highest score as the best sequence.

2. The landscape sequence design method based on the psychological health benefits of humans, as described in claim 1, is characterized in that: In step S1, the scoring range is 0 to 10 points; in step S3, GIS software is used to record the location information of all scenic spots and roads, N=5, to obtain the five shortest landscape sequence paths.

3. The landscape sequence design method based on the psychological health benefits of humans, as described in claim 1, is characterized in that: Step S1 specifically includes the following steps: Step S101: Obtain visual data for each attraction; Step S102: Obtain auditory data for each attraction; Step S103: Obtain sequence data; Step S104: Obtain the pleasure rating of all attractions along each route by having volunteers fill out questionnaires, and use the pleasure rating as a psychological health benefit rating.

4. The landscape sequence design method based on the psychological health benefits of humans, as described in claim 3, is characterized in that: The specific steps of step S101 include the following: Step S1011: Use a 360-degree panoramic camera to take pictures of each scenic spot and generate a high-resolution panoramic image of each scenic spot; Step S1012: Using the U-Net image segmentation algorithm based on deep learning, extract the proportion of green pixels from the panoramic image of each scenic spot and calculate the green view rate; Step S1013: By analyzing the vegetation coverage area in the panoramic image of each scenic spot, the vegetation coverage area ratio is calculated using the Mask R-CNN object detection algorithm to obtain the canopy closure. Step S1014: Using a sky region segmentation algorithm, the sky visibility factor is calculated by extracting the proportion of the sky region in the panoramic image of each scenic spot.

5. The landscape sequence design method based on the psychological health benefits of humans, as described in claim 4, is characterized in that: The specific steps of step S102 include the following: Step S1021: Set up a high-sensitivity microphone at each scenic spot to collect sound data for each scenic spot; Step S1022: Input the sound data of each scenic spot obtained in step S1021 into the audio classification model based on convolutional neural network to identify different sound types; Step S1023: Measure the sound pressure level of each scenic spot in real time using a sound level meter, in dB; Step S1024: Perform Fourier transform (FFT) on the sound data of each scenic spot obtained in step S1021 to extract the sound spectrum data and analyze the sound frequency distribution characteristics.

6. The landscape sequence design method based on the psychological health benefits of humans according to claim 3, characterized in that: The specific steps of step S103 include the following: Step S1031: Use a GPS device to record and obtain the location of each attraction; Step S1032: Mark the roads between attractions in the GIS software, obtain the road locations, calculate the distances between attractions, and record the order in which each path is obtained.

7. The landscape sequence design method based on the psychological health benefits of humans according to claim 1, characterized in that: In step S2, the hyperparameters are set: The number of trees, n_estimators, can be selected from 100 to 500. Maximum depth (max_depth): Automatically adjusted to optimize model performance; Minimum number of split samples (min_samples_split): Set to 2-5 depending on the amount of data.

Images