Old cell reconstruction space perception and optimization path evaluation method based on deep learning

Abstract

The invention provides an old cell reconstruction space perception and optimization path evaluation method based on deep learning, and belongs to the technical field of cell reconstruction. The method comprises the steps of establishing a research method and an evaluation system module, selecting a research object and acquiring research data, determining a processing method of the research data, constructing a measurement index system of street space objective environment perception, and evaluating the optimal path of the old cell transformation. Performing standardization processing on the measurement indexes of the street space objective environment perception; and performing weight setting on the measurement indexes of the street space objective environment perception. By constructing a measurement index system for objective environment perception of a street space, measurement is carried out from two aspects of material environment characteristics of an old community and subjective feeling of people, a theoretical method and support are provided for research, objective characteristics of a physical environment are concerned, subjective feeling of residents on the environment of the residents is also included, and the objective environment perception of the residents is realized. Therefore, a more comprehensive evaluation framework is provided.

Description

Technical Field

[0001] The present invention relates to the technical field of community renovation, and in particular to a deep learning-based spatial perception and optimization path evaluation method for old community renovation. Background Art

[0002] With the continuous development of urbanization, many cities are facing many problems such as aging buildings in old communities, outdated infrastructure, and declining environmental quality. The renovation of old communities has become the focus of urban renewal work. The continuous expansion of urban development ideas has caused inadaptability and incompatibility with the new urban development stage, thus producing a series of changes and evolutions. The stage of urban development has shifted from "large-scale incremental construction" to "equal emphasis on stock quality improvement and incremental structural adjustment." Residents attach great importance to the survey results.

[0003] Currently, there are few studies on the spatial perception measurement of old community renovation. Most of them are based on urban streets and are based on the material composition of the streets and the adaptability of users. The data and conclusions obtained may be biased and cannot truly reflect the overall picture of the spatial perception measurement of old community renovation. Summary of the invention

[0004] In order to make up for the above shortcomings, the present invention provides a method for spatial perception and optimized path evaluation for renovation of old communities based on deep learning, which overcomes the above technical problems or at least partially solves the above problems.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A deep learning-based spatial perception and optimization path evaluation method for old residential area renovation, including:

[0007] Establish research methods and build evaluation system modules, including literature analysis, big data collection, quantitative analysis, and empirical analysis;

[0008] Selection of research subjects and acquisition of research data. The research subjects should preferably be selected from typical areas of historical perception measurement;

[0009] Identify methods for processing research data, including semantic segmentation of street view data and building color recognition;

[0010] Construct a measurement index system for objective environmental perception of street space. The objective environmental perception is a preliminary integration and understanding based on the basic internal elements of street space. The objective environment of street space includes cultural elements, artificial environmental elements and natural elements. The measurement index system includes the walkability of old residential streets, the enclosure of old residential streets, the visibility of characteristic landscapes of old residential streets, the perception of cultural atmosphere of old residential streets and the perception of characteristic functions of old residential streets.

[0011] Standardization of measurement indicators of objective environmental perception of street space, wherein the standardization of measurement indicators of objective environmental perception of street space is combined with the identification results of the elements of the street and lane space environment of old residential areas and relevant indicators are selected for analysis;

[0012] The weights of the measurement indicators of the objective environmental perception of the street space are set. The weights of the measurement indicators of the objective environmental perception of the street space are compared by calculating the eigenvectors and eigenvalues ​​and finally obtaining the final weights of each measurement indicator after the consistency test of the matrix discrimination results.

[0013] Determine the environmental constituent element characteristics of the street space in the old residential area, wherein the environmental constituent element characteristics of the street space in the old residential area include the three-level distribution characteristics presented by the overall environmental constituent elements, the visible characteristics of the streets in the old residential area dominated by natural elements, and the data distribution characteristics of each element dominated by positive skewed distribution, wherein the visible characteristics of the streets in the old residential area dominated by natural elements are dominated by natural elements;

[0014] The subjective perception measurement of street space in old residential areas includes evaluating street view pictures in a small range by scoring, learning the relationship between the score of street view pictures and their element composition by using an artificial neural network model, and automatically evaluating all street view images within the range by a computer, thereby realizing the subjective environmental perception prediction of all street sampling points in the old residential areas.

[0015] By constructing a measurement index system for objective environmental perception of street space, the objective environment of street space includes cultural elements, artificial environmental elements and natural elements, and constitutes a multi-dimensional spatial perception method. Compared with the traditional single-dimensional spatial perception research, this method adopts a multi-dimensional spatial perception quantitative analysis method, which mainly measures the material environmental characteristics of old communities and people's subjective feelings, providing theoretical methods and support for research. This method not only focuses on the objective characteristics of the physical environment, but also includes residents' subjective feelings about their environment, thereby providing a more comprehensive evaluation framework. In one embodiment of the present invention, the method described.

[0016] In one embodiment of the present invention, the data includes road vector data, street network data and street view impact data. The area of ​​the research object contains a sufficient number of old residential communities and the architectural culture and architectural heritage are well preserved. The street image data is merged and topologically processed on a geographic information system software platform.

[0017] In one embodiment of the present invention, the semantic segmentation of the street view data aims to quantitatively deconstruct the street material environment while identifying the main types of visually perceptible environmental elements captured in the street visual scene and accurately quantifying and unifying their proportions in the street view image, and the building color recognition aims to count and identify the color composition, distribution range and proportion of architectural elements in the street view image.

[0018] In one embodiment of the present invention, the measure of the walkability of the streets in the old community includes a walking environment index, a traffic sign index and a vehicle interference index. The walking environment index is used to measure the walking quality in the streets in the old community. The vehicle interference index is used to measure the degree of interference of motor vehicles on the walking environment. The traffic sign index is used to evaluate the adequacy of traffic facilities in the streets in the old community. The enclosure of the streets in the old community is intended to illustrate the degree of enclosure of the streets by the interfaces on both sides of the streets in the old community.

[0019] In one embodiment of the present invention, the basic elements of the characteristic landscape of the streets in the old residential area include green trees, the Yangtze River and the blue sky. The green trees correspond to the environmental green vision index. The Yangtze River and the blue sky together constitute the environmental openness index. The measurement indicators of the perception of the cultural atmosphere of the streets in the old residential area include the proportion of historical buildings, traditional buildings and building interfaces of cultural relics protection units in the old residential area, building color pixels and crowd gathering index. The measurement of the building color pixels selects three colors: red, yellow and gray.

[0020] In one embodiment of the present invention, the relevant indicators selected for the standardized processing of the measurement indicators of the objective environment perception of the street space include three dimensions of artificial environment, natural landscape and culture in the objective environment perception measurement, and five levels of perception of walkability of streets in old communities, perception of enclosure of streets in old communities, perception of visibility of characteristic landscapes, and perception of cultural atmosphere and perception of characteristic functions.

[0021] In one embodiment of the present invention, the method for determining the weights of the measurement indicators of the objective environmental perception of the street space includes a questionnaire survey method and an interview method, and the weight hierarchy of the measurement indicators of the objective environmental perception of the street space includes a target layer and a quasi-side layer. The target layer involves the objective environmental perception of the space of old communities, and the quasi-side layer involves the perception of walkability of streets in old communities, the perception of enclosure of streets in old communities, the perception of visibility of characteristic landscapes, the perception of cultural atmosphere, and the perception of characteristic functions.

[0022] In one embodiment of the present invention, the first-level elements of the three-level distribution characteristics presented by the overall environmental elements include sky elements, building elements, road elements and vegetation elements, the second-level elements of the three-level distribution characteristics presented by the overall environmental elements include car elements, wall elements, sidewalk elements and fence elements, and the third-level elements of the three-level distribution characteristics presented by the overall environmental elements include street light elements, truck elements, traffic sign elements, traffic light elements, pedestrian elements, motorcycle elements, bus elements and bicycle elements.

[0023] In one embodiment of the present invention, the analysis means for measuring the subjective perception of street space in old communities includes spatial distribution structure analysis and data statistical analysis. The spatial distribution structure analysis links the measurement results of the subjective environmental perception evaluation platform with the old community street evaluation source data, and then discusses the distribution characteristics of the dominant perception types of the old community street environment and the subjective perception evaluation structure characteristics of the street environment through road network mapping and kernel density analysis.

[0024] In one embodiment of the present invention, the evaluation system module comprises:

[0025] A data collection module, the data source of which comes from the network map API port. The data collected by the data collection module includes drone aerial images, ground photos taken by high-definition cameras, laser radar point cloud data, and existing community planning drawings and basic geographic information data;

[0026] A data preprocessing module, wherein the data preprocessing module processes data using methods including denoising and enhancement of image data, filtering and three-dimensional reconstruction of point cloud data, unified data format processing, and construction of an initial database for multi-source data fusion;

[0027] A deep learning-based spatial perception module, which analyzes the spatial structure of a community through a deep learning model to identify problem areas and potential transformation targets;

[0028] A transformation path optimization module, wherein the optimization objectives of the transformation path optimization module include space utilization, cost optimization, and satisfaction;

[0029] A transformation effect evaluation module, wherein the evaluation dimensions of the transformation effect evaluation module include space transformation indicators, economic benefits and social benefits.

[0030] The present invention provides a method for spatial perception and optimization path evaluation of old residential area renovation based on deep learning, and its beneficial effects include:

[0031] 1. By constructing a measurement index system for the objective environmental perception of street space, the objective environment of street space includes cultural elements, artificial environmental elements and natural elements, and constitutes a multi-dimensional spatial perception method. Compared with the traditional single-dimensional spatial perception research, this method adopts a multi-dimensional spatial perception quantitative analysis method, which mainly measures the material environment characteristics of old communities and people's subjective feelings, providing theoretical methods and support for the research. This method not only focuses on the objective characteristics of the physical environment, but also includes the subjective feelings of residents about their environment, thus providing a more comprehensive evaluation framework.

[0032] 2. Through the subjective perception measurement of street space in old residential areas, an artificial neural network model is used to learn the relationship between the score of street view pictures and their element composition. Now, the spatial characteristics of old residential areas can be efficiently identified and analyzed, and a measurement system for spatial perception of old residential area renovation can be constructed. This can extract deeper insights from large amounts of data and provide more accurate support for renovation strategies.

[0033] 3. By clarifying the processing method of research data, semantically segmenting street view data, and identifying building colors, the visual appearance characteristics of old residential areas are quantified at multiple scales, and a multi-feature fusion method based on street view images of old residential areas is established, which helps to accurately capture and analyze the environmental characteristics within the residential area, provide detailed data support for the transformation, and at the same time explore the impact of objective environmental factors on residents' subjective perception. In the objective material perception measurement, effective factor analysis is conducted through the sustainability of the transformation of old residential areas, the optimization of the transformation, the visibility of landscape features, the perception of cultural atmosphere, and the perception of characteristic functions. In the subjective environmental perception measurement, the key factors affecting residents' perception are further clarified, providing a basis for formulating effective transformation strategies. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0035] Figure 1 is a flow chart of a method provided by an embodiment of the present invention;

[0036] Figure 2 A schematic diagram of environmental components provided for an embodiment of the present invention;

[0037] Figure 3 A schematic diagram of measurement indicators provided by an embodiment of the present invention;

[0038] Figure 4A schematic diagram of a quasi-side layer structure provided in an embodiment of the present invention;

[0039] Figure 5 A schematic diagram of a research method provided for an embodiment of the present invention;

[0040] Figure 6 A schematic diagram of an optimization object provided in an embodiment of the present invention;

[0041] Figure 7 A schematic diagram of an evaluation system module provided in an embodiment of the present invention;

[0042] Figure 8 A flow chart of the spatial subjective perception measurement process provided by an embodiment of the present invention.

[0043] 1. Data acquisition module; 2. Data preprocessing module; 3. Spatial perception module based on deep learning; 4. Transformation path optimization module; 5. Transformation effect evaluation module. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0045] In the description of the present invention, it should be noted that the terms "upper", "lower", "left", "right", "inner", "outer", "top / bottom" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific position, be constructed and operated in a specific position, and therefore cannot be understood as limiting the present invention. In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance.

[0046] Example

[0047] See also Figure 1-8The present invention provides a technical solution: a method for evaluating spatial perception and optimization path of old residential area reconstruction based on deep learning, including establishing a research method and establishing an evaluation system module, selecting a research object and obtaining research data, clarifying a processing method for research data, constructing a measurement index system for objective environmental perception of street space, standardizing the measurement index of objective environmental perception of street space, setting the weight of the measurement index of objective environmental perception of street space, determining the characteristics of environmental constituent elements of street space in old residential areas and the subjective perception measurement of street space in old residential areas, the research methods include literature analysis, big data collection, quantitative analysis and empirical analysis, the research object should be selected from typical areas of historical perception measurement, the processing methods include semantic segmentation of street view data and architectural color recognition, objective environmental perception is a preliminary integration and understanding based on the basic internal constituent elements of street space, the objective environment of street space includes cultural elements, human The measurement index system includes the walkability of old community streets, the enclosure of old community streets, the visibility of characteristic landscapes of old community streets, the perception of cultural atmosphere of old community streets and the perception of characteristic functions of old community streets. The standardized processing of the measurement index of objective environmental perception of street space is combined with the identification results of the spatial environmental constituent elements of streets and lanes in old communities and relevant indicators are selected for analysis. The weights of the measurement index of objective environmental perception of street space are compared through the calculation of eigenvectors and eigenvalues, and the final weights of each measurement index are finally obtained after the consistency test of the matrix discrimination results. The characteristics of the environmental constituent elements of the street space in old communities include the three-level distribution characteristics presented by the overall environmental constituent elements, the visible characteristics of the old community streets dominated by natural elements and the data distribution characteristics of each element dominated by positive skewed distribution. The visible characteristics of the old community streets dominated by natural elements are dominated by natural elements.

[0048] See also Figure 1-8 The data include road vector data, street network data and streetscape impact data. The research area contains sufficient old residential areas and the architectural culture and architectural heritage are well preserved. The street image data are merged and topologically processed on the geographic information system software platform.

[0049] See also Figure 1-8, The measurement of the walkability of old community streets includes the walking environment index, traffic sign index and vehicle interference index. The walking environment index is used to measure the walking quality in old community streets, the vehicle interference index is used to measure the interference degree of motor vehicles on the walking environment, and the traffic sign index is used to evaluate the adequacy of traffic facilities in old community streets. The enclosure of old community streets aims to illustrate the enclosure degree of the streets on both sides of the old community streets. The walking environment index is an indicator used to measure the walking quality in old community streets, among which the proportion of pedestrian paths is an important consideration. When the walking index is higher, it means that the old community streets can accommodate more people and their activities, and the walking environment is more pleasant, which will make people more willing to walk, thereby increasing people’s contact and understanding of old community streets, and deepening their cognition and perception of old community streets. The correlation perception measurement index formula of the walking environment index, traffic sign index and vehicle interference index is:

[0050]

[0051] Among them, WEIq refers to the walking environment index; Wp is the number of sidewalk pixels identified in the street view image; Rw is the number of road feature pixels in the street view image;

[0052]

[0053] TSIq refers to the street traffic sign index; Tp is the number of pixels of traffic facilities such as street lights, signal lights, and traffic signs in the street view image; Rp is the total number of pixels of road elements in the street view image;

[0054]

[0055] Among them, VDIq refers to the vehicle interference index of the street environment; Vp is the number of pixels of motor vehicles such as cars, buses, motorcycles, and trucks in the street view image; Rp is the number of pixels of road elements in the street view image;

[0056] The enclosure index refers to the degree of enclosure of the street by the interfaces on both sides of the street, including building entities, enclosing walls, courtyard fences or other structures. The enclosure formula is:

[0057]

[0058] Among them, SIq refers to the street environment enclosure index; Sp is the number of pixels of walls, fences, vegetation, and buildings identified in the street view image; Ip is the total number of pixels in all areas of the street view image.

[0059] See also Figure 1-8The basic elements of the characteristic landscape of old residential streets include green trees, the Yangtze River and the blue sky. Green trees correspond to the environmental green vision index, and the Yangtze River and the blue sky together constitute the environmental openness index. The measurement indicators of the perception of the cultural atmosphere of old residential streets include the proportion of historical buildings, traditional buildings and building interfaces of cultural relics protection units in old residential areas, building color pixels and crowd gathering index. The measurement of building color pixels selects red, yellow and gray colors. The formulas of the environmental green tree index and environmental openness index are as follows:

[0060]

[0061] Among them, GVIq refers to the green visibility index of the street environment; Gp is ​​the number of pixels of green landscape elements identified in the street view image; Ip is the total number of pixels in the corresponding area of ​​the street view;

[0062]

[0063] Among them, EVIq refers to the openness index of the street environment; Ep is the sum of the pixels of the sky and Yangtze River elements identified in the street view image; Ip is the total number of pixels in all areas of the street view image.

[0064] See also Figure 1-8 The relevant indicators selected for the standardized processing of the measurement indicators of the objective environmental perception of street space include the three dimensions of artificial environment, natural landscape and culture in the objective environmental perception measurement, as well as the five levels of the perception of walkability of old residential streets, the perception of the enclosure of old residential streets, the perception of the visibility of characteristic landscapes, the perception of cultural atmosphere and the perception of characteristic functions. In order to ensure the reliability of the results and avoid deviations, the relevant indicators were standardized, and the Min-Max standardized public announcement was adopted:

[0065]

[0066] Where X' ij is the original data value, X jmax and X jmin They represent the upper and lower limits of the relevant measurement index values ​​respectively, i is the sample number of the street view sampling point, and j is the number corresponding to the relevant measurement index.

[0067] See also Figure 1-8The semantic segmentation of street view data aims to quantitatively deconstruct the physical environment of the street while identifying the main types of visually perceptible environmental elements captured in the street visual scene and accurately quantify and unify their proportions in the street view images. The architectural color recognition aims to count and identify the color composition, distribution range and proportion of architectural elements in street view images. The TensorFlow framework and deeplab v3+ decoder are selected. The advantage is that they effectively improve the segmentation quality and application scope of the model, making it superior in many semantic segmentation tasks. In the segmentation of small targets and scenes with rich details, the upsampling and multi-scale feature fusion capabilities of the decoder significantly improve the accuracy of the target boundary. Especially in complex scenes, this fine-grained segmentation performance is particularly prominent. The recognition of local architectural colors mainly counts and identifies the color composition, distribution range and proportion of architectural elements in each street view image. In order to construct refined color recognition and proposal, the recognition process is set to input street view data, extract architectural elements, optimize architectural images, identify HSV three-channel values, and propose Take each color block, count the number of pixels of each color, output the picture and value. When extracting architectural elements, first use the OpenCV computer vision library to semantically segment and identify architectural elements, then extract the architectural elements through masks. On this basis, perform color calibration on the proposed building facade image, use the white balance algorithm to reduce the color deviation problem that may be caused by brightness problems, and finally use the closed operation algorithm to refine the building edge to avoid the fragmented parts of the surrounding environment affecting color recognition. Then, based on the corresponding HSV parameters, extract the color blocks of each color and count their pixel numbers. The following are the relevant formulas for the historical and cultural interface index, local color atmosphere, and crowd gathering index:

[0068]

[0069] Among them, HIIq refers to the historical interface index in the street environment; Hp is the interface length of historical buildings, traditional buildings, and cultural relics protection units on both sides of the street; Bp is the total length of the building interface on both sides of the street;

[0070]

[0071] Among them, LCIq refers to the local color atmosphere index; Cp is the total number of local color pixels, and Ep is the number of pixels of all architectural elements in the architectural image;

[0072]

[0073] Among them, CGIq refers to the crowd gathering index; Pp is the sum of the pixels of pedestrians and bicycles identified in the street view image; Ip is the total number of pixels in all areas of the street view image.

[0074] See also Figure 1-8The weight determination methods of the measurement indicators of the objective environmental perception of street space include questionnaire survey method and interview method. The weight hierarchy of the measurement indicators of the objective environmental perception of street space includes target layer and quasi-layer. The target layer involves the objective environmental perception of the space of old residential communities, and the quasi-layer involves the perception of walkability of streets in old residential communities, the perception of enclosure of streets in old residential communities, the perception of visibility of characteristic landscapes, the perception of cultural atmosphere, and the perception of characteristic functions.

[0075] See also Figure 1-8 The first-level elements of the three-level distribution characteristics of the overall environmental elements include sky elements, building elements, road elements and vegetation elements. The second-level elements of the three-level distribution characteristics of the overall environmental elements include car elements, wall elements, sidewalk elements and fence elements. The third-level elements of the three-level distribution characteristics of the overall environmental elements include street light elements, truck elements, traffic sign elements, traffic light elements, pedestrian elements, motorcycle elements, bus elements and bicycle elements. By comparing and summarizing the data of each element, the following characteristics can be obtained:

[0076] Pa=(∑I f ×I w )÷D v ×100%

[0077] Where P a is the average number of pixels of a certain environmental element, I f is the frequency of the environmental factor in the overall measurement data, I w is the weight of the factor in the overall measurement data results, D v is the total number of samples.

[0078] See also Figure 1-8The analysis methods of subjective perception measurement of street space in old communities include spatial distribution structure analysis and data statistical analysis. The spatial distribution structure analysis links the measurement results of the subjective environmental perception evaluation platform with the source data of the old community street evaluation, and then discusses the distribution characteristics of the dominant perception types of the old community street environment and the structural characteristics of the subjective perception evaluation of the street environment through road network mapping and kernel density analysis. In the spatial perception model, the individual's subjective perception of the objective environment is the last stage of the model. On the one hand, people's subjective environmental perception can be understood as being affected by objective environmental elements and representations. After the primary perception process of the environment through subjective and objective interaction, it is combined with the subjective perception of the environment. The final comprehensive understanding and perception results of objective environmental information formed by the combination of personal attributes, motivations and needs, combined with the measurement content of the objective environment of the street space in the old residential area and the primary perception of the environment in the previous article, determined the individual evaluation indicators reflecting subjective environmental perception, and built a subjective perception platform based on unified street view data. The subjective perception measurement process is to evaluate street view pictures in a small range through the ELO scoring method, and then use the artificial neural network model to learn the relationship between the score of the street view picture and its element composition. All street view images within the range are automatically evaluated by computer, thereby realizing the prediction of subjective environmental perception of all street sampling points in the old residential area.

[0079] See also Figure 1-8 The evaluation system modules include a data acquisition module 1, a data preprocessing module 2, a spatial perception module 3 based on deep learning, a transformation path optimization module 4 and a transformation effect evaluation module 5. The data source of the data acquisition module 1 comes from the network map API port. The data collected by the data acquisition module 1 include drone aerial images, ground photos taken by high-definition cameras, lidar point cloud data, and existing planning drawings and basic geographic information data of the community. The methods used by the data preprocessing module 2 to process data include denoising and enhancement of image data, filtering and three-dimensional reconstruction of point cloud data, unified processing of data formats, and construction of an initial database for multi-source data fusion. The spatial perception module 3 based on deep learning analyzes the spatial structure of the community through a deep learning model to identify problem areas and potential transformation targets. The optimization goals of the transformation path optimization module 4 include space utilization, cost optimization and satisfaction. The evaluation dimensions of the transformation effect evaluation module 5 include spatial transformation indicators, economic benefits and social benefits.

[0080] Specifically, the working process or working principle of the spatial perception and optimization path evaluation method for the renovation of old communities based on deep learning is as follows: when in use, when conducting an evaluation, the first step is to select a typical area of ​​historical perception measurement as the research object, and the selection is based on the distribution of a large number of old community buildings, and the architectural culture and architectural heritage are well preserved; the second step is to obtain data through the data acquisition module 1, specifically to obtain street network data and street view image data, to obtain road vector data through OpenStreetMap, and to process the obtained road network information, to delete duplicate road sections, and to retain only road sections with real street view images, and to delete the road information without street views in the road network, and then based on the street network data processed in the previous step According to the data, further operations such as street merging and topological processing are performed in the ArcGis platform to merge the pre-processed roads into a line segment, and on this basis, the intersections are interrupted, and samples are taken at fixed distances along the center line of the road to obtain coordinate points. The collection program is used to obtain the front, back, left and right images of each street view sampling point from the network map platform such as Baidu Map; the third step is to use the data preprocessing module 2 to process the research data, specifically semantic segmentation and recognition to quantitatively deconstruct the street material environment, identify the main types of visually perceptible environmental elements captured in the street visual scene, and accurately quantify and unify the proportion of these constituent elements in the street view pictures. The recognition of local colors of buildings is mainly The color composition, distribution range and proportion of architectural elements in each street view image are counted and identified, and the identification process is set to input street view data, extract architectural elements, optimize architectural images, identify HSV three-channel values, extract color blocks of each color, count the number of pixels of each color, and output pictures and values; the fourth step is to construct a measurement index system for objective environmental perception of street space. The environmental components of street space in old communities are divided into cultural elements, artificial environmental elements and natural elements. On the basis of the three types of environmental components, the objective perception of this study is obtained. The selection of measurement indicators is mainly carried out from the following five aspects: walkability of streets in old communities, enclosure of streets in old communities, visibility of characteristic landscapes in streets in old communities, and visibility of streets in old communities. Perception of cultural atmosphere, and perception of characteristic functions of streets in old communities; the fifth step is to use the deep learning-based spatial perception module 3 to standardize the measurement indicators of objective environmental perception of street space, specifically combining the identification results of the spatial environmental elements of streets and lanes in historical and cultural blocks, and in the objective environmental perception measurement, from the three dimensions of artificial environment, natural landscape and culture, the perception of walkability of streets in old communities, the perception of enclosure of streets in old communities, the perception of visibility of characteristic landscapes, and the perception of cultural atmosphere and characteristic functions. Since the values ​​of some indicators vary greatly, in order to ensure the reliability of the results and avoid deviations, the relevant indicators are standardized, and Min-Max standardization is used;The sixth step is to set the weights of measurement indicators. In order to determine the influence weights of measurement indicators, questionnaire surveys and interviews were conducted, and then the AHP hierarchical analysis method was used to establish a hierarchical structure. The objective environmental perception of the old community space was set as the target layer, and the five levels of old community street walkability perception, old community street enclosure perception, characteristic landscape visibility perception, cultural atmosphere perception, and characteristic function perception were set as the quasi-layer. The corresponding measurement indicators were set as the factor scheme layer, and a paired comparison matrix was established to determine the relationship between indicators. The weights were compared by calculating the eigenvectors and eigenvalues, and the consistency test of the matrix discrimination results was used to determine the relationship between the indicators. This shows that the entire decision-making process is reliable and effective, and finally the final weight of each measurement indicator is obtained; the seventh step is to determine the characteristics of the environmental components of the street space in the old community. By performing semantic segmentation and recognition on the street view images collected in the streets of the old community, the proportion of each environmental component is obtained, and it is visualized in space through ArcGIS to observe the spatial distribution characteristics of each element. The analysis of its basic situation can better help decision makers understand the environmental characteristics and features of the block, thereby providing data support and decision-making reference for the renovation and management of the streets in the old community; the eighth step is to analyze the subjective perception of the street space in the old community. The subjective perception platform is built based on the subjective perception process. After the primary perception process of the environment through subjective and objective interaction, the final comprehensive understanding and perception results of the objective environmental information are formed in combination with the individual attributes, motivations and needs. The individual evaluation indicators reflecting subjective environmental perception are determined by combining the objective environment of the street space in the old residential area and the measurement content of the primary perception of the environment. Based on the unified street view data, the subjective perception platform is built. The subjective perception measurement process is to evaluate the street view pictures in a small range through the ELO scoring method, and then use the artificial neural network model to learn the relationship between the score of the street view picture and its element composition. The computer automatically evaluates all the street view images within the range. Evaluation is conducted to predict the subjective environmental perception of all street sampling points in the old residential area; finally, using the transformation effect evaluation module 5, according to the subjective environmental perception measurement results of the old residential area streets, according to the guidance form of the data analysis results, the measurement results are analyzed for spatial distribution structure and data statistics, and the measurement results of the subjective environmental perception evaluation platform are linked with the evaluation source data of the old residential area streets, and then the data is input into the GIS platform. Through road network mapping and kernel density analysis, the distribution characteristics of the dominant perception types of the street environment in the old residential area and the structural characteristics of the subjective perception evaluation of the street environment are discussed. ;

[0081] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for spatial perception and optimization path evaluation of old residential area reconstruction based on deep learning, characterized in that: The method comprises: Establish research methods and build evaluation system modules, including literature analysis, big data collection, quantitative analysis, and empirical analysis; Selection of research subjects and acquisition of research data. The research subjects should preferably be selected from typical areas of historical perception measurement; Identify methods for processing research data, including semantic segmentation of street view data and building color recognition; Construct a measurement index system for objective environmental perception of street space. The objective environmental perception is a preliminary integration and understanding based on the basic internal elements of street space. The objective environment of street space includes cultural elements, artificial environmental elements and natural elements. The measurement index system includes the walkability of old residential streets, the enclosure of old residential streets, the visibility of characteristic landscapes of old residential streets, the perception of cultural atmosphere of old residential streets and the perception of characteristic functions of old residential streets. Standardization of measurement indicators of objective environmental perception of street space, wherein the standardization of measurement indicators of objective environmental perception of street space is combined with the identification results of the elements of the street and lane space environment of old residential areas and relevant indicators are selected for analysis; The weights of the measurement indicators of the objective environmental perception of the street space are set. The weights of the measurement indicators of the objective environmental perception of the street space are compared by calculating the eigenvectors and eigenvalues ​​and finally obtaining the final weights of each measurement indicator after the consistency test of the matrix discrimination results. Determine the environmental constituent element characteristics of the street space in the old residential area, wherein the environmental constituent element characteristics of the street space in the old residential area include the three-level distribution characteristics presented by the overall environmental constituent elements, the visible characteristics of the streets in the old residential area dominated by natural elements, and the data distribution characteristics of each element dominated by positive skewed distribution, wherein the visible characteristics of the streets in the old residential area dominated by natural elements are dominated by natural elements; The subjective perception measurement of street space in old residential areas includes evaluating street view pictures in a small range by scoring, learning the relationship between the score of street view pictures and their element composition by using an artificial neural network model, and automatically performing a computer-generated evaluation on all street view images within the range.

2. According to the method of deep learning-based spatial perception and optimization path evaluation for old residential area reconstruction according to claim 1, it is characterized in that: The data include road vector data, street network data and street view impact data. The area of ​​the research object contains a sufficient number of old residential communities and the architectural culture and architectural heritage are well preserved. The street image data is merged and topologically processed on a geographic information system software platform.

3. According to claim 2, a method for spatial perception and optimization path evaluation of old residential area reconstruction based on deep learning is characterized in that: The semantic segmentation of street view data aims to quantitatively deconstruct the physical environment of the street while identifying the main types of visually perceptible environmental elements captured in the street visual scene and accurately quantifying and unifying their proportions in the street view image. The architectural color recognition aims to count and identify the color composition, distribution range and proportion of architectural elements in street view images.

4. According to claim 3, a method for spatial perception and optimization path evaluation of old residential area reconstruction based on deep learning is characterized in that: The measurement of the walkability of the old community streets includes the walking environment index, the traffic sign index and the vehicle interference index. The walking environment index is used to measure the walking quality in the old community streets, the vehicle interference index is used to measure the degree of interference of motor vehicles on the walking environment, the traffic sign index is used to evaluate the adequacy of traffic facilities in the old community streets, and the enclosure of the old community streets is intended to illustrate the degree of enclosure of the streets by the interfaces on both sides of the old community streets.

5. According to claim 4, a method for spatial perception and optimization path evaluation of old residential area reconstruction based on deep learning is characterized in that: The basic elements of the characteristic landscape of the streets in the old residential areas include green trees, the Yangtze River and the blue sky. The green trees correspond to the environmental green vision index, and the Yangtze River and the blue sky together constitute the environmental openness index. The measurement indicators of the perception of the cultural atmosphere of the streets in the old residential areas include the proportion of the building interfaces of historical buildings, traditional buildings and cultural relics protection units in the old residential areas, building color pixels and the crowd gathering index. The measurement of the building color pixels selects three colors: red, yellow and gray.

6. According to claim 5, a method for spatial perception and optimization path evaluation of old residential area reconstruction based on deep learning is characterized in that: The relevant indicators selected for the standardized processing of the measurement indicators of the objective environmental perception of the street space include the three dimensions of artificial environment, natural landscape and culture in the objective environmental perception measurement, as well as the five levels of perception of walkability of streets in old communities, perception of enclosure of streets in old communities, perception of visibility of characteristic landscapes, and perception of cultural atmosphere and perception of characteristic functions.

7. According to claim 6, a method for spatial perception and optimization path evaluation of old residential area reconstruction based on deep learning is characterized in that: The weight determination method of the measurement indicators of the objective environmental perception of the street space includes a questionnaire survey method and an interview method. The weight hierarchy structure of the measurement indicators of the objective environmental perception of the street space includes a target layer and a quasi-side layer. The target layer involves the objective environmental perception of the old community space, and the quasi-side layer involves the perception of walkability of streets in old communities, the perception of enclosure of streets in old communities, the perception of visibility of characteristic landscapes, the perception of cultural atmosphere, and the perception of characteristic functions.

8. According to claim 7, a method for spatial perception and optimization path evaluation of old residential area reconstruction based on deep learning is characterized in that: The first-level elements of the three-level distribution characteristics presented by the overall environmental elements include sky elements, building elements, road elements and vegetation elements; the second-level elements of the three-level distribution characteristics presented by the overall environmental elements include car elements, wall elements, sidewalk elements and fence elements; the third-level elements of the three-level distribution characteristics presented by the overall environmental elements include street light elements, truck elements, traffic sign elements, traffic light elements, pedestrian elements, motorcycle elements, bus elements and bicycle elements.

9. According to claim 8, a method for spatial perception and optimization path evaluation of old residential area reconstruction based on deep learning is characterized in that: The analysis methods for measuring the subjective perception of street space in old communities include spatial distribution structure analysis and data statistical analysis. The spatial distribution structure analysis links the measurement results of the subjective environmental perception evaluation platform with the source data of street evaluation in old communities, and then discusses the distribution characteristics of the dominant perception types of the street environment in old communities and the structural characteristics of the subjective perception evaluation of the street environment through road network mapping and kernel density analysis.

10. According to the method of deep learning-based spatial perception and optimization path evaluation for old residential area reconstruction in claim 1, it is characterized in that: The evaluation system modules include: A data collection module (1), wherein the data source of the data collection module (1) comes from a network map API port, and the data collected by the data collection module (1) includes drone aerial images, ground photos taken by a high-definition camera, laser radar point cloud data, and existing community planning drawings and basic geographic information data; A data preprocessing module (2), wherein the data preprocessing module (2) processes data using methods including denoising and enhancement of image data, filtering and three-dimensional reconstruction of point cloud data, data format unification processing, and construction of an initial database for multi-source data fusion; A deep learning-based spatial perception module (3), wherein the deep learning-based spatial perception module (3) analyzes the spatial structure of the community through a deep learning model to identify problem areas and potential transformation targets; A transformation path optimization module (4), wherein the optimization objectives of the transformation path optimization module (4) include space utilization, cost optimization, and satisfaction; A transformation effect evaluation module (5), wherein the evaluation dimensions of the transformation effect evaluation module (5) include space transformation indicators, economic benefits and social benefits.

Images