Street space pleasure evaluation and optimization method based on multi-source data

Abstract

The invention discloses a street space pleasure evaluation and optimization method based on multi-source data. The method comprises the following steps: data acquisition: acquiring street scene images, three-dimensional building data and POI data in a research area; subjective perception data acquisition: acquiring a subjective perception score of a user for street pleasure by using a street view image and an online data collection platform; objective index extraction: extracting objective indexes such as physical characteristics, interface characteristics, urban functions and macroscopic forms of streets from the multi-source data; correlation analysis: carrying out correlation analysis on objective indexes and subjective perception scores, and determining key factors influencing street pleasure; establishing an evaluation model: establishing a street space pleasure evaluation model based on the key factors; and optimization strategy making: proposing a targeted street design optimization strategy according to an evaluation result. According to the method, multi-source data and individual perception are comprehensively considered, the street space pleasure can be evaluated more accurately, and a scientific basis is provided for street design and planning.

Description

Technical Field

[0001] The present invention relates to the technical field of urban planning and design, and particularly relates to a method for evaluating and optimizing the amenity of street space based on multi-source data. Background Art

[0002] As an important part of urban public space, the amenity of streets directly affects the quality of life of residents and the overall image of the city. Traditional street evaluation methods mainly rely on on-site investigations and expert scoring, which have problems such as cumbersome procedures, strong subjectivity, small data volume, and low accuracy. With the development of information technology, various new data and new technologies have provided new possibilities for the research of street space. However, most existing studies start from a single perspective such as street vitality, quality, and walkability, lacking quantitative analysis of the relationship between individual psychological perception and street space elements, and mostly using a single data source, unable to comprehensively and accurately evaluate the amenity of street space. Summary of the Invention

[0003] Aiming at the deficiencies in the prior art, the purpose of the present invention is to provide a method for evaluating and optimizing the amenity of street space based on multi-source data. This method comprehensively considers multi-source data and individual perception, can more accurately evaluate the amenity of street space, and provides a scientific basis for street design and planning. To achieve the above object and other advantages of the present invention, a method for evaluating and optimizing the amenity of street space based on multi-source data is provided, including the following steps:

[0004] S1. Obtain street view images, 3D building data, and POI data within the research area;

[0005] S2. Obtain the subjective perception scores of users on the amenity of the street through street view images and an online data collection platform;

[0006] S3. Extract objective indicators of the physical characteristics, interface characteristics, urban functions, and macroscopic morphology of the street from the data in step S1;

[0007] S4. Conduct a correlation analysis between the objective indicators and the subjective perception scores to determine the key factors affecting the amenity of the street;

[0008] S5. Establish an evaluation model for the amenity of street space based on the key factors, and conduct an evaluation of the amenity of street space to obtain an evaluation result;

[0009] S6. Propose targeted street design optimization strategies according to the evaluation results.

[0010] First, obtain street view images through the Baidu Map API, acquire 3D building data from the OSM website, obtain POI data using web crawlers, and collect subjective perception scores of users for street view images through the online platform of the Massachusetts Institute of Technology. Then, extract the physical features of the street from multi-source data, namely objective indicators such as street width, aspect ratio, green view rate, interface features such as building line ratio, interface permeability, number of store signs on the interface, urban functions such as function mix, function density, number of commercial facilities, and macroscopic forms such as building density, floor area ratio, and open space ratio. Next, conduct a correlation analysis, normalize the objective indicators and perform a regression analysis with the subjective perception scores to determine the key factors and establish a multiple regression evaluation model. Finally, propose optimization strategies for the key factors based on the evaluation results, such as adjusting the green view rate, building density, and aspect ratio, etc., apply them to the actual street design and planning and continuously improve.

[0011] Compared with the prior art, the beneficial effects of the present invention are as follows: It comprehensively considers multi-source data and individual perceptions, can more accurately evaluate the comfort of street space, and provides a scientific basis for street design and planning. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 FIG. for determining the near-person space from the human visual field range in the method for evaluating and optimizing the comfort of street space based on multi-source data according to the present invention;

[0013] Figure 2 FIG. for determining the near-person space from the degree of closeness of the relationship between people and street space in the method for evaluating and optimizing the comfort of street space based on multi-source data according to the present invention;

[0014] Figure 3 FIG. for selecting an example of the comfort level of a living street in the method for evaluating and optimizing the comfort of street space based on multi-source data according to the present invention;

[0015] Figure 4 FIG. for the schematic diagram of the street comfort score in the method for evaluating and optimizing the comfort of street space based on multi-source data according to the present invention;

[0016] Figure 5 FIG. for the visualization diagram of the street comfort score based on multi-source data in the method for evaluating and optimizing the comfort of street space according to the present invention;

[0017] Figure 6 FIG. for the actual street view corresponding to the aspect ratio after automatic classification of street views in the method for evaluating and optimizing the comfort of street space based on multi-source data according to the present invention;

[0018] Figure 7 FIG. for identifying store signs using a deep learning model in the method for evaluating and optimizing the comfort of street space based on multi-source data according to the present invention;

[0019] Figure 8 Schematic diagram of using a deep learning model to identify arbors and shrubs for the method of evaluating and optimizing the street space comfort based on multi-source data according to the present invention;

[0020] Figure 9 Flowchart of the method of evaluating and optimizing the street space comfort based on multi-source data according to the present invention;

[0021] Figure 10 Flowchart of calculating the interface feature index for the method of evaluating and optimizing the street space comfort based on multi-source data according to the present invention;

[0022] Figure 11 Schematic diagram of obtaining street view images for the method of evaluating and optimizing the street space comfort based on multi-source data according to the present invention. Detailed implementation manners

[0023] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0024] Referring to Figure 1 , a method for evaluating and optimizing the street space comfort based on multi-source data includes:

[0025] I. Data acquisition stage

[0026] 1. Obtaining street view images, as Figure 11 shown:

[0027] According to the scope of the research area, use the Baidu Map API to access and retrieve publicly available Google Street View (GSV) images at each sampling point location, and obtain a large number of street view images according to the set sampling point spacing, image size, and selection rules to ensure coverage of the main streets within the research area.

[0028] 2. Obtaining 3D building data

[0029] Download the 3D building data of the research area from the OSM website, and organize and preprocess the data to meet the requirements of subsequent analysis.

[0030] 3. Obtaining POI data

[0031] By writing a web crawler program, crawl the POI data of large map websites, screen the POI points related to vitality according to the range on both sides of the street, and perform data cleaning and classification.

[0032] II. Subjective Perception Data Collection Phase

[0033] (1) Street view images were published on the MIT Spatial Pulse online data collection platform to attract a large number of users to participate in the evaluation. The image dataset contains 123,988 street view images taken between 2007 and 2015, spanning 56 cities in 28 countries across six continents.

[0034] (2) Based on two randomly selected street view images shown on the platform, users selected the more pleasant street or the "equal" option. The platform recorded the users' selections in real time. After multiple comparisons and calculations, the positive rate, negative rate, and score of each image were obtained.

[0035] Each image sample i was compared with other images i'. The calculation of the positive rate (P) of image i along a certain perception index is shown in the formula:

[0036]

[0037] The calculation of the negative rate (N) is shown in the formula:

[0038]

[0039] Where Pi and Ni represent the number of times image i was selected or not selected in the comparison, and ei is the number of times image i was considered equal to another image. The score of picture i is:

[0040]

[0041] (3) A model was built between street view pictures and individual perception based on the classical deep convolutional neural network (ResNet). The model took street view pictures as input, used the deep convolutional neural network to perform binary classification prediction on the pictures, and then performed interval mapping on the predicted confidence (probability score) to restore the continuous value of the pleasantness perception score. Overall, the experiment achieved 72% accuracy in the prediction of four aspects related to pleasantness: spatial convenience, comfort, safety, and aesthetics. During the calculation, each street view picture was automatically scored by the model, and the average score of each street view picture was statistically calculated by section.

[0042] III. Objective Index Provision Phase

[0043] As shown in the following table:

[0044]

[0045] 1. Calculation of Physical Feature Indexes

[0046] (1) Street width: According to Harvey's method, for each section within the range of 1 - 40 m, a buffer zone is drawn every 1 m with the distance (d) from the interface line to the center line of the road as the width, and the ratio of the area of the unilateral buffer zone and the area after removing buildings on one side is calculated. The mutation point with the largest ratio difference between d and d + 1 is found; subsequently, the distances from the bilateral street boundary lines of each section to the center line are added together. Among them, sections without buildings on both sides have been deleted, and sections without buildings on one side are calculated according to the maximum value of 40 m set by Harvey.

[0047] (2) Aspect ratio: Use a deep learning model to process street view images, identify the aspect ratio types of each street view image on each section, and classify them into 4 categories: 0 < H / W < 1, 1 < H / W < 2, 2 < H / W < 4, H / W > 4; finally, count the number of street view images of the 4 aspect ratio types on the street by section, and take the category with the largest number as the aspect ratio type of this section.

[0048] (3) Green view rate: Based on a deep convolutional neural network, the scene analysis model (PSPNet) can assign a category label to each pixel in the image, achieving a pixelization accuracy of 79.70%. Instead of looking for "green" pixels with large errors and difficult to define in photos, this method directly learns the visual characteristics of green vegetation and identifies various vegetation types such as trees and shrubs. Refer to this deep learning model of PSPNet to identify and locate the vegetation in the street view image, and count the proportion of the pixels of the vegetation in the image.

[0049] 2. Calculation of interface feature indicators, as Figure 10 shown:

[0050] (1) Building alignment rate: Calculate the proportion of the intersection length according to the relationship between the street boundary line and the building location.

[0051] (2) Interface permeability: Analyze the ratio of the transparent interface to the constructed interface in the street view image.

[0052] (3) Number of store signs on the interface: Count the number of store signs at all sampling points within the street section and calculate the average value.

[0053] 3. Calculation of urban function indicators

[0054] (1) Functional mixing degree: Assume that the total number of POIs around a street is A, which is divided into N types in total, including residential, commercial, transportation, catering and entertainment, education and training, medical services, enterprise factory offices, government agencies and social organizations, green spaces, etc. (N ≤ 9).

[0055] If the numbers of each type are A1, A2,..., An respectively. Then Define the probability as Obviously, there is The formula for the street function mixing degree is as follows:

[0056] (2) Function density: Calculate the function density based on the total number of POIs and the street length. The calculation formula is as follows: total number of POIs / street length × 100, that is, the density of POI points in every 100m street segment.

[0057] (3) Number of businesses: Directly count the number of stores in the area.

[0058] 4. Calculation of macroscopic morphological indicators

[0059] (1) Building density: Calculate the ratio of the total building area to the area of the square area based on the three-dimensional building data.

[0060] (2) Floor area ratio: Obtain relevant information from OSM data for calculation.

[0061] (3) Proportion of open space: Calculate the ratio of the area of park green space and water surface to the area of the sampling area.

[0062] IV. Correlation analysis stage

[0063] Normalize the 12 extracted objective indicators, use the amenity index as the dependent variable, and analyze it using regression analysis tools (such as SPSS, R language, etc.). Calculate the Pearson correlation coefficient and variance inflation coefficient between the independent variables to judge the correlation and collinearity. If the Pearson correlation coefficients between the independent variables are all less than 0.7, it indicates that the correlation between the independent variables is weak and there is no collinearity between the independent variables. The calculation formula is as follows:

[0064] AMENITY (amenity index) = a0 + a1 × H / W (aspect ratio) + a2 × W (street width) + a3 × GREEN (green view rate) + a4 × RATE (interface line attachment rate) + a5 × PERMEABILITY (interface permeability) + a6 × SHOP SIGN (number of interface shop signs) + a7 × FUNCTION DENSITY (function density) + a8 × ∑NI (function mixing degree) + a9 × BUSINESSES (number of businesses) + a10 × DENSITY (building density) + a11 × FAR (floor area ratio) + a12 × OPEN (proportion of open space) + ε

[0065] Among them, a1 - a12 are the weight coefficients to be identified in the regression analysis, a0 is the constant term, and ε is the random error. If the variance inflation coefficient VIF value after regression is less than 5, it indicates that there is no collinearity between the independent variables. Determine the significant features affecting the amenity index according to the regression results to obtain the regression model and the correlation coefficient.

[0066] V. Stage of establishing an evaluation model

[0067] Based on the key factors determined by the correlation analysis, substitute them into the multiple regression model formula to determine the weight coefficients and the constant term, and establish an evaluation model for the street space amenity. Use this model to predict the amenity scores of the streets within the research area.

[0068] VI. Stage of formulating optimization strategies

[0069] According to the results of the evaluation model, analyze the influence degree of the key factors on the amenity. For key factors such as the green view rate, building density, and aspect ratio, propose specific optimization measures. For example, for streets with a low green view rate, street greening can be increased and appropriate vegetation species can be selected; for areas with too high building density, the building layout can be reasonably planned to increase open spaces; for streets with an unsatisfactory aspect ratio, the near-human space can be processed, such as setting up arcade buildings along the street, adding storefront signs and facade decorations, etc., to improve the amenity of the street space. Apply these optimization strategies in the actual street design and planning, continuously monitor and evaluate their effects, and further adjust and improve the optimization plan according to the feedback.

[0070] The equipment quantities and treatment scales described here are used to simplify the description of the present invention, and the application, modification, and variation of the present invention are obvious to those skilled in the art. Although the embodiments of the present invention have been disclosed as above, it is not limited to the applications listed in the specification and the embodiments. It can be fully applied to various fields suitable for the present invention. For those familiar with the field, additional modifications can be easily achieved. Therefore, without departing from the general concept defined by the claims and the equivalent scope, the present invention is not limited to the specific details and the illustrated and described examples here.

Claims

1. A method for evaluating and optimizing the comfort of street space based on multi-source data, characterized in that, It includes the following steps: S1. Obtain street view images, 3D building data, and POI data within the research area; S2. Obtain the subjective perception scores of the street amenity by the users through the street view images and the online data collection platform; S3. Extract the objective indicators of the physical features, interface features, urban functions, and macroscopic forms of the street from the data in step S1; S4. Conduct a correlation analysis between the objective indicators and the subjective perception scores to determine the key factors affecting the street amenity; S5. Establish an evaluation model for the street space amenity based on the key factors, and conduct an evaluation of the street space amenity to obtain the evaluation results; S6. Propose targeted street design optimization strategies according to the evaluation results.

2. The method for evaluating and optimizing the street space amenity based on multi-source data according to claim 1, characterized in that, The data of the street view images in step S1 is an intelligent evaluation based on street view pictures. Specifically, according to the scope of the research area, the Baidu Map API is used to access and retrieve GSV images at each sampling point location, and a large number of street view images are obtained according to the set sampling point spacing, image size, and selection rules to ensure that the main streets within the research area are covered.

3. The method for evaluating and optimizing the street space amenity based on multi-source data according to claim 2, characterized in that The 3D building data in step S1 is a street perception and form analysis based on GIS and 3D building data. Specifically, the 3D building data of the research area is downloaded from the OSM website, the data is sorted and preprocessed, and according to the levels divided by the OSM website, the highway, main road, under the viaduct, bridge, and the road sections in the residential area are deleted to meet the requirements of subsequent analysis.

4. The method for evaluating and optimizing the street space amenity based on multi-source data according to claim 3, characterized in that, The POI data in step S1 is an exploration of the street vitality based on the POI data. Specifically, by writing a web crawler program, the POI data of a large map website is crawled, the POI points related to vitality are screened according to the range on both sides of the street, and the data is cleaned and classified.

5. The method for evaluating and optimizing the amenity of street space based on multi-source data according to claim 1, wherein Step S2 specifically includes the following steps: S21. Complete the evaluation of the street view images by a large number of users through the MIT Spatial Pulse online data collection platform to form a machine learning data set; on the website, the participants are required to select one from two randomly selected pictures, and the options are the left image, the right image, or "equal" to indicate their perceptual judgment; S22. Compare each image sample i with other images i', and calculate the positive rate and negative rate of image i along a certain perception index; S23. Build a model between the street view pictures and the individual perception based on the classical deep convolutional neural network (ResNet). During the calculation, the model automatically scores each street view picture, and the average score of each street view picture is statistically calculated by section.

6. The method for evaluating and optimizing the street space amenity based on multi-source data according to claim 1, wherein, The objective indicators in step S3 include physical feature indicators, interface feature indicators, urban function indicators, and macroscopic form indicators.

7. The method for evaluating and optimizing the street space amenity based on multi-source data according to claim 6, characterized in that, The physical feature indicators include street width, height-width ratio, and green view rate; The interface feature indicators include building line ratio, interface permeability, and the number of interface shop signs; The urban function indicators include function mix degree, function density, and the number of commercial establishments; The macroscopic form indicators include building density, floor area ratio, and open space ratio.

Images