Method and device for evaluating balance of material environment and economic environment and storage medium
The establishment of a correlation model between the urban material environment and the social and economic environment through neural networks and deep learning methods solves the problem that existing technology cannot fully capture complex factors, and achieves a more comprehensive and comprehensive assessment of the balanced urban development.
Patent Information
- Application Number
- CN202510356809.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-25
- Publication Date
- 2025-06-27
Smart Images
Figure CN120218672A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the fields of artificial intelligence technology and urban geography technology, and in particular, to a method, device, and storage medium for evaluating the balance between the physical environment and the economic environment. Background Art
[0002] The balanced development of the urban physical environment and the social and economic environment is an important requirement for the sustainable development goal. The balanced development of the city promotes the construction of inclusive and healthy communities, ensures long-term sustainable development, and promotes economic growth and prosperity, benefiting all residents. By coordinating the development of the urban physical environment, including urban landscapes and infrastructure, and the social and economic environment, including social, economic, and cultural atmospheres, the city can provide a higher quality of life, reduce social gaps, and mitigate adverse environmental impacts.
[0003] Currently, the evaluation of the balanced development of the urban physical environment and the social and economic environment usually adopts the method of index evaluation or analytic hierarchy process. The specific principle of the index evaluation method is to collect a series of relevant index data according to the requirements of the sustainable development goal, such as urban greening rate, per capita disposable income, public infrastructure coverage rate, etc., and then judge the balanced development status of the city by constructing an index evaluation system.
[0004] The disadvantages of the existing technical methods mainly come from the one-sidedness of data collection, the simplicity of the modeling method, and the difference in index design. First, the current method based on index evaluation or analytic hierarchy process can only capture the characteristics that are easy to quantify or observe in urban development, such as urban greening rate and gross national product (GDP), etc., while ignoring the complex but important factors hidden in the system, such as social capital, cultural influence, and residents' happiness, etc. The lack of these factors may lead to the incompleteness of the comprehensive evaluation of the city's development. Summary of the Invention
[0005] In order to make up for the above deficiencies, the present application proposes a method, device, and storage medium for evaluating the balance between the urban physical environment and the social and economic environment.
[0006] Its method technical solution includes: a method for evaluating the balance between the material environment and the economic environment, which uses a neural network to train the correlation model between the material environment of the urban agglomeration to be evaluated and the urban indicators, and evaluates the balance between the material environment and the economic environment of the urban agglomeration to be evaluated by comparing the deviations of the indicators between cities according to the model; the method steps include: obtaining the urban indicators of each city in the urban agglomeration to be evaluated and the street view images corresponding to the urban indicators; according to each urban indicator and the corresponding street view image, using a neural network to establish the correlation model between the urban material environment and the urban indicator of each city; performing migration calculation on the correlation models of each city to obtain the quantitative differences between the material environment and the urban indicators of different cities; according to the migration results, comparing the similarity of the transfer deviation trends between the material environment and the urban indicators of each city, and then evaluating whether the material environment and the social and economic environment are balanced in urban development.
[0007] Preferably, before establishing the correlation model between the urban material environment and the urban indicator according to the urban indicator and the corresponding street view image, the street view image and the urban indicator are also preprocessed to screen out representative street view images and regularize the urban indicator data.
[0008] Preferably, the screening criteria for the representative street view images include that the neural network predicts the administrative division where it is located through the street view image, and the street view image that can correctly predict the area where it is located is considered a representative image.
[0009] Preferably, the process of regularizing the urban indicator data includes converting the urban indicator data from absolute values into decile segment intervals within the scope of the city, that is, the corresponding 0-10%, 10-20%,..., 90-100% segments.
[0010] Preferably, the process of obtaining the street view image includes: obtaining the urban indicators of the corresponding area, obtaining the location information of the corresponding community through address coding according to the community where the urban indicator data is located, and then collecting the street view data of the corresponding location, so as to obtain the street view image corresponding to the urban indicator of the city.
[0011] Preferably, the street view image includes urban infrastructure images and street landscape images.
[0012] This application also proposes an electronic device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the method described in any one of the above is implemented.
[0013] The present application also provides a computer-readable storage medium, which includes a stored computer program. When the computer program runs, it controls the device where the computer-readable storage medium is located to execute the method described in any one of the above.
[0014] The present invention captures the comprehensive urban physical environment by using street view images, and at the same time, through deep learning methods, embeds the urban socio-economic environment completely into the model, ensuring the comprehensiveness of the urban environment description in this analysis. The present invention realizes the comprehensive modeling of the urban physical environment and the socio-economic environment through deep learning and model migration methods, thereby taking into account the interaction and dynamics of different factors in the urban environment. Through deep learning methods, the present invention treats the physical environment and the socio-economic environment as a whole and unified, ensuring the comprehensiveness of the evaluation results and being able to better reflect the urban development situation. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] Figure 1 It is a schematic flowchart of an implementation manner of the method of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS
[0016] Next, the technical solutions of the present application will be described clearly and completely in conjunction with the drawings. Obviously, the described embodiments are part of the embodiments of the present application, rather than all of them. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative efforts belong to the scope of protection of the present application.
[0017] In the description of the present application, it should be noted that the orientation or positional relationship indicated by the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present application and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation of the present application. In addition, the terms "first", "second", and "third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.
[0018] In the description of the present application, it should be noted that unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected, or indirectly connected through an intermediate medium, and it can be the communication inside two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to specific situations.
[0019] The present invention will be further described in detail below with reference to the accompanying drawings. Referring to Figure 1 , in the present application, a method for evaluating the balance between the material environment and the economic environment uses a neural network to train a correlation model between the material environment of the urban agglomeration to be evaluated and urban indicators, and evaluates the balance between the material environment and the economic environment of the urban agglomeration to be evaluated by comparing the deviations of indicators between cities according to the model; the method steps include: obtaining the urban indicators of each city in the urban agglomeration to be evaluated and the street view images corresponding to the urban indicators; establishing a correlation model between the urban material environment and the urban indicators of each city using a neural network according to each urban indicator and the corresponding street view image; performing migration calculations on the correlation models of each city to obtain the quantitative differences between the material environments and urban indicators of different cities; according to the migration results, comparing the similarity of the transfer deviation trends between the material environments and urban indicators of each city, and further evaluating whether the material environment and the social and economic environment are balanced in urban development. In this embodiment, the process of evaluating the similarity between the material environment and the social and economic environment in different urban developments and whether the material environment and the social and economic environment are balanced includes, based on the migration deviations of the material environment and the social and economic environment calculated between the city and different cities to form vectors, obtaining the migration deviation vector of the material environment and the migration deviation vector of the social and economic environment of each city. Then, the cosine similarity is used to calculate the correlation between the migration deviation vectors of different cities, and according to the threshold, the similarity of the two cities is judged. When the cosine similarity is higher than the threshold, the two cities are considered similar, otherwise they are not. In the process of judging the balance, for all cities, based on the similarity discrimination between cities, the cities are classified into different categories. In each category, the sample closest to the center point in the space composed of the cosine similarities between samples is taken as the representative city, and the balance of the two environments of the city is judged based on the magnitude of the migration deviation values of the material environment and the social and economic environment of the city. If the migration deviations of the two environments are both negative, it is considered that the city develops well and is balanced; if the migration deviations of the two environments are both positive, it is considered that the city develops poorly and is balanced; if the migration deviations of the two environments are positive and negative inconsistently, it is considered that the city develops unbalanced.
[0020] On the basis of the above embodiment, further, before establishing a correlation model between the urban material environment and the urban indicators according to the urban indicators and the corresponding street view images, the street view images and urban indicators are also preprocessed to screen out representative street view images and regularize the urban indicator data.
[0021] On the basis of a certain above embodiment, further, the screening criteria for the representative street view images include that the street view images whose administrative division can be correctly predicted by the neural network through the street view images are considered representative images.
[0022] Based on a certain above - mentioned embodiment, further, the process of regularizing urban indicator data includes converting the urban indicator data from absolute values into decile segmentation intervals within the scope of the city, that is, the corresponding segmentation of 0 - 10%, 10 - 20%, …, 90 - 100%. In this embodiment, the decile segmentation interval is the segmentation of the maximum and minimum interval values of the urban indicator data. For example, for housing prices, they are evenly divided into ten segmentation intervals from the highest to the lowest according to price.
[0023] Based on a certain above - mentioned embodiment, further, the process of obtaining street - view images includes: obtaining the urban indicators of the corresponding region, obtaining the location information of the corresponding community through address coding according to the community where the urban indicator data is located, then collecting the street - view data at the corresponding location, and further obtaining the street - view images corresponding to the urban indicators of the city.
[0024] Based on a certain above - mentioned embodiment, further, the street - view images include urban infrastructure images and street - scene images.
[0025] For one or more of the above - mentioned embodiments, the present application further exemplifies the method. In the method of the present invention, the first step: obtaining the urban indicators and corresponding street - view images of different cities. For different cities included, respectively obtain the fine - grained urban indicators of the corresponding regions (such as housing prices, crime rates, safety perceptions, etc.), then obtain the location information of the corresponding community through address coding according to the community where the local urban indicator data is located, and then collect the street - view data at the corresponding location to obtain the corresponding street - view images as the physical - environment representation of this area of the city.
[0026] The second step: data pre - processing. Data pre - processing includes screening representative street - view images and data regularization. In order to select representative street - view images, using the method of a neural network model, predicting the administrative division where the street - view image is located through the street - view image, and the images that can be correctly predicted are considered representative images; for data regularization, convert the urban indicator data from absolute values into decile segmentation intervals within the scope of the city, that is, the corresponding segmentation of 0 - 10%, 10 - 20%, …, 90 - 100%, so as to ensure the efficient learning of the relationship between the subsequent urban physical environment and urban indicators.
[0027] The third step: establishing a model of the urban physical environment and urban indicators. Taking housing price as an example of urban indicators, for the cities included in the experiment, use a deep - learning model to predict the urban indicators at the corresponding location through street - view images to capture the relationship between the urban physical environment and urban housing prices.
[0028] Step 4: Quantify the differences in the physical environment and the socio-economic environment among different cities. Through different migration methods, the differences in the physical environment and indicators of different cities can be obtained. Still taking housing prices as an example, the street view images of city b are migrated to the model trained in city a for prediction. The difference between the prediction result and the real housing price of city b is the migration deviation in the socio-economic environment among different cities, which reflects the difference in the socio-economic environment between city a and city b; the street view images of city a and city b are migrated to the model of city c for prediction, and the difference between the prediction result and the real housing price of its corresponding city reflects the difference in the physical environment between city a and city b.
[0029] Step 5: Evaluate the similarity between the physical environment and the socio-economic environment in the development of different cities. For any city, based on the migration deviations of the physical environment and the socio-economic environment calculated between this city and different cities to form vectors, the migration deviation vector of the physical environment and the migration deviation vector of the socio-economic environment of each city are obtained. Then, the cosine similarity is used to calculate the correlation between the migration deviation vectors of different cities, and according to the threshold, the similarity between two cities is judged. When the cosine similarity is higher than the threshold, it is considered that the two cities are similar, otherwise they are not. In the present invention, the threshold is usually set to the empirical value of 0.5.
[0030] Step 6: Evaluate the balance between the physical environment and the socio-economic environment in the development of cities. For all cities, based on the similarity discrimination between cities calculated in Step 5, the cities are classified into different categories. In each category, the sample closest to the center point in the space composed of the cosine similarities between samples is taken as the representative city, and the balance of the two environments of this city is judged based on the magnitude of the migration deviation values of the physical environment and the socio-economic environment of this city. If the migration deviations of both environments are negative, it is considered that the city develops well and is balanced; if the migration deviations of both environments are positive, it is considered that the city develops poorly and is balanced; if the migration deviations of the two environments are positive and negative inconsistently, it is considered that the city develops unbalanced.
[0031] Taking street view images as physical environment data and city indicators as socio-economic environment data, hierarchical clustering is used to compare the similarity of the transfer deviation trends of the physical and socio-economic environments among cities respectively, and the balanced and unbalanced classifications of urban development are obtained.
[0032] Furthermore, taking housing prices as an example of city indicators and 10 cities in the United States as the research area, the balance of the development of the physical environment and the socio-economic environment of these cities is evaluated. The specific steps are as follows: Step 1: Obtain the housing price data at the community level of 10 cities in the United States. Then, according to the location information of the area where the housing price data is located, obtain the Google street view image set.
[0033] Step 2: Preprocess the community-level housing price data and Google Street View image data obtained in Step 1. For the Google Street View image data, by training a deep learning classifier, filter out the street view images that can correctly predict the administrative division where they are located. These filtered images are considered representative images and enter the next process. For the housing price data, regularize these data to their decile segmentation intervals.
[0034] Step 3: For the 10 US cities covered in the experiment, use deep learning methods to predict the housing prices at corresponding locations in each city through the street view images of that city, and capture the relationship between the urban physical environment and urban housing prices. In this process, the deep learning model captures the influence of the socioeconomic environment (such as economic, cultural, social and other factors) on urban indicators. In the present invention, the urban socioeconomic environment characteristics implicitly affect the relationship between the urban physical environment and housing prices. The model may capture social exclusion phenomena in the environmental characteristics (such as the lack of public facilities in a specific area), and then expose this socioeconomic difference through indicators. For example, in this embodiment, the non-linear mapping relationship between the street view images (representations of the physical environment) and housing prices (urban economic indicators) will be affected by the socioeconomic environment. Therefore, by using deep learning methods, taking the street view images (representations of the physical environment) as inputs and housing prices (urban economic indicators) as outputs, a model is built to express this non-linear relationship, thereby quantifying the influence of the socioeconomic environment. In this step, decouple the physical environment and socioeconomic signals from the images. The model automatically identifies visual features in the street view through a convolutional neural network (CNN). Certain visual features (such as the proportion of high-end shops and luxury cars) imply the regional economic level; the architectural style (such as historic preservation buildings) reflects the cultural background; the design of community public spaces reflects the social class structure.
[0035] Step 4: Quantify the differences in the physical environment and socioeconomic environment between different cities. Through different transfer methods, we can obtain the differences in the physical environment and socioeconomic environment between different cities: Transfer the street view images of city a and city b to the model of city c for prediction. The difference between the prediction result and the true housing price of its corresponding city reflects the difference in the physical environment between city a and city b; Transfer the street view images of city b to the model trained in city a for prediction. The difference between the prediction result and the true housing price of city b is the transfer bias in the socioeconomic environment between different cities, which reflects the difference in the socioeconomic environment between city a and city b. In this step, through the operation in Step 4, "Transfer the street view images of city b to the model trained in city a for prediction. The difference between the prediction result and the true housing price of city b is the transfer bias in the socioeconomic environment between different cities, which reflects the difference in the socioeconomic environment between city a and city b.", the decoupling of the socioeconomic environment between cities is achieved.
[0036] Step 5: Evaluate the similarity between the physical environment and the socio-economic environment in the development of different cities. First, take City A. Based on the migration deviations of the physical environment and the socio-economic environment calculated between City A and different cities, form vectors. Each city has a migration deviation vector of the physical environment and a migration deviation vector of the socio-economic environment . Then, use the cosine similarity to calculate the correlation between the migration deviation vectors of different cities. For example, the similarity of the physical environment between City A and City B is , and the similarity of the socio-economic environment is . Set the threshold to 0.5. If the cosine similarity is higher than the threshold, the two cities are considered similar; otherwise, they are not
[0037] Step 6: Evaluate the balance between the physical environment and the socio-economic environment in the development of cities. For all cities, based on the similarity discrimination between cities obtained in Step 5, classify the cities into different categories. In each category, select a representative city and judge the balance of the two environments of the city based on the magnitude of the migration deviation values of the physical environment and the socio-economic environment. If the migration deviations of both environments are negative, it is considered that the city has good and balanced development; if the migration deviations of both environments are positive, it is considered that the city has poor and balanced development; if the migration deviations of the two environments are positive and negative inconsistently, it is considered that the city has unbalanced development
[0038] The present application also provides an electronic device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the method described in any one of the above is implemented
[0039] The present application also provides a computer-readable storage medium. The computer-readable storage medium includes a stored computer program. When the computer program runs, it controls the device where the computer-readable storage medium is located to execute the method described in any one of the above
[0040] The analysis process of this example can clearly classify the development situations of the physical environment and the socio-economic environment of different cities in the experiment into two categories: "good development" and "insufficient development", and based on the development states of these two environments, judge the balance of the current development of the two environments of the city. The results of this experiment can better reflect the development status of the selected US cities in the current experiment
[0041] The above has introduced in detail the method for quantifying the balance of the development of the urban physical environment and the social and economic environment proposed by the present invention. The above examples have elaborated on the principle and implementation manner of the present solution, but are only used to help understand the method and its core idea of the present invention. At the same time, according to different application scenarios, there will be changes in the specific implementation manner. In summary, the content of the above examples should not be construed as a limitation to the present solution.
Claims
1. A method for evaluating the balance between the physical environment and the economic environment, which uses a neural network to construct a correlation model between the physical environment and urban indicators of the urban agglomeration to be evaluated, compares the deviations of indicators between cities according to the correlation model, and then evaluates the balance between the physical environment and the economic environment of the urban agglomeration to be evaluated; characterized in that: The method steps include: Obtain the city indicators of each city in the urban cluster to be evaluated and the street view images corresponding to the city indicators; Based on each city indicator and the corresponding street view image, a neural network is used to establish a correlation model between the urban material environment and the city indicator of each city; Migrate the correlation models of each city to obtain the quantitative differences in the material environment and urban indicators between different cities; Based on the migration results, the similarities between the material environment and the urban indicator transfer deviation trends among cities are compared, and then the balance between the material environment and the socio-economic environment in urban development is evaluated.
2. The method according to claim 1, characterized in that: Before establishing the correlation model between the urban material environment and the urban indicators based on the urban indicators and the corresponding street view images, the street view images and urban indicators are preprocessed to screen out representative street view images and regularize the urban indicator data.
3. The method according to claim 2, characterized in that The representative street view image screening criteria include predicting the administrative division where the street view image is located by a neural network, and a street view image that can correctly predict the area is considered to be a representative image.
4. The method according to claim 2, characterized in that: The regularization process of the city indicator data includes converting the city indicator data from absolute values into decile segmentation intervals within the city, namely, the corresponding 0-10%, 10-20%, ..., 90-100% segments.
5. The method according to claim 1, characterized in that The street view image acquisition process includes: obtaining the city index of the corresponding area, obtaining the location information of the corresponding community through address coding according to the community where the city index data is located, and then collecting the street view data of the corresponding location, and then obtaining the street view image corresponding to the city index of the city.
6. The method according to claim 1, characterized in that The street view images include urban infrastructure images and street landscape images.
7. An electronic device, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor implements the method according to any one of claims 1 to 6 when executing the computer program.
8. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the method according to any one of claims 1 to 6.