Street space gene recognition and extraction method based on ancient customs pictures

Abstract

The invention discloses a street space gene recognition and extraction method based on ancient customs drawings, which comprises the following steps: collecting the ancient customs drawings related to a to-be-studied regional range, and screening and classifying the ancient customs drawings; dividing a characteristic space scene in the ancient custom picture in combination with the space scene conversion marker; performing extraction on the spatial features of the street by combining the divided spatial scenes, and performing clustering on each street space; extracting space elements of each street space in the street space image and converting the space elements into text labels; based on the street space in the scene, analyzing to obtain the configuration relationship of the space elements, and combining with the function interaction relationship in the ancient customs drawing to form a combination relationship and performing screening; verifying a spatial element configuration relationship in the screened combination relationship to obtain a final street space gene; according to the method, the suitability of the path flow of space gene recognition extraction and the technical method is effectively improved.

Description

Technical Field

[0001] The invention relates to a neighborhood space gene identification and extraction method, in particular to a neighborhood space gene identification and extraction method based on ancient folk paintings, and belongs to the technical field of information extraction. Background Art

[0002] The historical context of a city is the accumulation and charm of urban development. In the context of urban construction, neighborhood space, as an important carrier of urban context and a key space for urban renewal actions, is a powerful tool for realizing the integration of urban context inheritance and urban renewal. The proposal of spatial genes provides a new guiding path for the inheritance of urban context, and the identification and extraction of neighborhood space genes can also better guide urban renewal work. At present, many scholars have conducted research on the identification and extraction of spatial genes in terms of path processes and technical methods, but there are still problems such as the mismatch between the universal process based on multi-materials and the technical methods for specific materials and local links. Therefore, how to construct a set of targeted spatial gene identification and extraction path processes and specific methods for specific materials, and support the basic links of the spatial gene analysis and control technology system, is still a key proposition in spatial gene research.

[0003] Based on the principle of selection and rejection in traditional painting, the content of ancient folk paintings depicting urban landscapes is the most representative of urban spatial characteristics, selected from the perspective of local painters at that time. It is also the most important urban spatial scene, which meets the requirement of spatial gene identification and extraction to recognize the space with urban regional characteristics; secondly, the picture expression of traditional paintings is spatial. When ancient folk paintings depict urban landscapes, their specific depiction objects are the elements of urban space. The picture expression of such folk paintings is the organization and arrangement of the elements of urban space in the picture space. The spatial relationship of urban space elements constructed in the picture space is essentially the abstraction of the spatial relationship of the real world from the perspective of the painter. Therefore, ancient folk paintings can further support the extraction of spatial element configuration relationships within the neighborhood space gene. At present, there is little research on spatial gene extraction technology for ancient folk paintings, and general information extraction technology is difficult to adapt to the spatial gene extraction of ancient folk paintings. Summary of the invention

[0004] Purpose of the invention: The purpose of the present invention is to provide a neighborhood space gene identification and extraction method based on ancient customs paintings, which can improve the adaptability of space gene identification and extraction.

[0005] Technical solution: The method for identifying and extracting neighborhood space genes based on ancient folk paintings of the present invention comprises:

[0006] S1: Collect ancient custom paintings related to the region to be studied, obtain data sets, and screen and classify the data sets to obtain main material data sets and auxiliary material data sets;

[0007] S2: Based on the scattered perspective and narrative characteristics of ancient genre paintings, combined with the space scene conversion markers, the characteristic space scenes in ancient genre paintings are divided;

[0008] S3: Combine the functional activities, regional sections, natural environment characteristics, and architectural characteristics of each neighborhood space in the divided space scene to refine the neighborhood space characteristics, and cluster each neighborhood space based on the neighborhood space characteristics;

[0009] S4: Use image analysis and recognition tools to perform label recognition on the clustered neighborhood space images, extract the spatial elements of each neighborhood space in the neighborhood space images and convert them into text labels;

[0010] S5: Based on the neighborhood space in the scene, the configuration relationship of spatial elements is analyzed, combined with the functional interaction relationship in ancient folk paintings, and then a combination relationship is formed and screened;

[0011] S6: Verify the spatial element configuration relationship in the screened combination relationship to obtain the final neighborhood space gene.

[0012] Furthermore, the step S1 includes: using ancient custom paintings that use fine brushwork techniques and have more than or equal to 4 urban space elements in the image as the main material data set D a , the remaining ancient customs paintings are used as auxiliary material data set D b .

[0013] Furthermore, the method for determining the narrative characteristics in step S2 is: combining the spatial narrative characteristics displayed in the ancient genre paintings, judging the spatial narrative plot according to the spatial regional attributes of the elements of the paintings, and combining the theme depicted in the works.

[0014] Furthermore, the method for determining the scattered perspective characteristic in step S2 is:

[0015] Different spatial scenes in the picture are judged according to the changes in perspective and visual focus under the scattered perspective; the perspective change refers to the different observation perspectives of each scene produced by the painter adjusting the relative height between the viewpoint and the observed scene when depicting different scenes; the visual focus change refers to the different composition centers of gravity constructed by the painter when depicting different spatial scenes.

[0016] Furthermore, in step S2, the characteristic spatial scenes in the ancient genre paintings are divided in combination with the spatial scene conversion markers, including:

[0017] Determining a spatial scene transition marker; the spatial scene transition marker is a spatial element that forms an occlusion or separation relationship between spatial scenes;

[0018] Based on the judgment of narrative plot, combined with the analysis of perspective, changes in visual focus and spatial scene transition markers under scattered perspective, the spatial scenes of ancient genre paintings are divided.

[0019] Furthermore, the step S3 comprises:

[0020] S3.1: Assign functional activity labels A to the neighborhood spaces in each spatial scene X ; The functional activities include residence, commerce, public services and leisure and entertainment;

[0021] According to the regional segment attributes of each spatial scene, the neighborhood space in each spatial scene is assigned a regional segment label Z X ;

[0022] According to the proportion of natural environment in each neighborhood space N X 、Building structure ratio SR X , Building Height E X 、Architectural form characteristics B X The indicators are used to refine the natural environment characteristics and architectural characteristics of each neighborhood space in each space scene; X , Z X 、N X , SR X 、E X With B X As an indicator for comprehensively determining the characteristics of neighborhood space X;

[0023] S3.2: A X With Z X The same neighborhood spaces are clustered into one category, which is determined as the set {T1, T2, … ,T K}, K is the number of sets; then, determine N in equal proportion X , SR X 、E X With B X The weights of the four items are standardized to obtain N X ′、SR X ′、E X ′ and B X ′, according to the formula Quantify the characteristics of each neighborhood space, in T k In the above example, the neighborhood spaces whose final comprehensive numerical error is less than the threshold α are clustered into type set t k .

[0024] Furthermore, the step S4 comprises:

[0025] S4.1: Based on the neighborhood scale, the spatial elements of the neighborhood space are classified according to the natural environment and artificial environment classification; the spatial element classification includes rivers, lakes, mountains, farmland, streets, buildings, city walls, water piers, bridges and archways;

[0026] S4.2: For each type set t k The scene images of the neighborhood spaces are stratified and sampled according to a preset ratio. The image analysis and recognition tool is called to perform feature detection on the sampled images to obtain a multi-label set of feature elements with confidence scores for the corresponding images, and the labels with confidence levels higher than a preset threshold are extracted. The corresponding feature elements are screened and merged to obtain spatial element labels. According to the spatial element labels, multiple labels that are not automatically identified as neighborhood space elements are added to all images, and the upper and lower limits of the number of images annotated with each element label are set, and the annotated images are split into training set, validation set and test set. The machine learning platform is used to optimize the scene images of various types of neighborhood spaces, and the confidence threshold is set to θ to automatically quantify the spatial elements of each neighborhood space.

[0027] Furthermore, the step S5 comprises:

[0028] The spatial elements identified and extracted from each neighborhood space are arranged and combined exhaustively to form a spatial element matrix; the spatial element configuration relationship of the neighborhood space is analyzed from multiple angles, and the functional interaction relationship between the main groups of people in each spatial element is analyzed based on the functional interaction relationship between space and crowd activities. Based on the combination relationship formed by the spatial element configuration relationship and the functional interaction relationship, the type set of the same neighborhood space is t k In the clustering, the spatial element combination relationship extracted from the neighborhood space of the main material is clustered to obtain t k The multiple sets of spatial element combination relationships under the above are selected, and based on the characteristics of spatial genes controlling the expression of urban spatial characteristics and the broad consensus characteristics of spatial genes, the multiple sets of spatial element combination relationships are screened, and the spatial element combination relationships belonging to the spatial element configuration relationships that will not affect the formation of the neighborhood spatial characteristics are deleted. In combination with auxiliary materials, the corresponding spatial element combination relationships that are only involved in individual spatial scenes are deleted.

[0029] Furthermore, the multi-angle analysis includes juxtaposition combination relationship analysis, link dependency relationship analysis, sequence relationship analysis and connectivity relationship analysis;

[0030] The juxtaposition combination relationship analysis is based on the principle of spatial geometry, according to the visual perception completeness principle and the visual perception topology theory, and through combination, the relationship analysis of spatial elements with combination tendency is carried out;

[0031] The link dependency analysis is based on the attributes of the spatial elements themselves.

[0032] The sequence relationship analysis is to analyze the spatial sequence formed based on the streamline and positional relationship of the spatial elements;

[0033] The connectivity analysis is specifically to determine whether two points can be connected by a path when any point is selected in two spatial elements.

[0034] Furthermore, the step S6 comprises:

[0035] S6.1: Based on the characteristics that the spatial element configuration relationship is organized by the concept of urban construction, the spatial element configuration relationship that does not conform to the concept of urban construction is eliminated;

[0036] S6.2: Based on the stable existence of spatial genes, according to the principle of diachrony, the neighborhood space type set extracted from ancient folk paintings is k The spatial scenes in the type set are compared with the corresponding modern urban spaces, or neighborhood spaces that can reflect the characteristics of this neighborhood are found in the modern urban spaces. Through on-site surveys, combined with photos and documentary materials, it is judged whether the configuration relationships of each group of spatial elements under this type set can be established in the modern urban neighborhood space. If the neighborhood space characteristics originally extracted from ancient genre paintings and controlled by the spatial gene are changed due to the absence or change of the configuration relationship of the spatial elements, it is judged that the neighborhood space gene has not undergone diachronic screening and is eliminated.

[0037] Beneficial effects: Compared with the prior art, the present invention improves the adaptability of the path process and technical methods of spatial gene identification and extraction from the perspective of ancient folk paintings as extraction materials, enriches the technical methods of spatial gene identification and extraction; deepens the value of using ancient folk paintings to study the space of historical urban neighborhoods, and expands the practical application value of ancient folk paintings as a material in the study of urban historical context for cultural heritage. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 is a flow chart of the method of the present invention;

[0039] Figure 2 A schematic diagram of the occlusion relationship of the space scene conversion markers according to an embodiment of the present invention;

[0040] Figure 3 A schematic diagram of the separation relationship of space scene conversion markers according to an embodiment of the present invention;

[0041] Figure 4 A schematic diagram of a spatial element label and data format according to an embodiment of the present invention;

[0042] Figure 5 This is a schematic diagram of spatial element recognition and extraction results according to an embodiment of the present invention;

[0043] Figure 6 A schematic diagram of a spatial element matrix according to an embodiment of the present invention;

[0044] Figure 7 Schematic diagram of the spatial element combination relationship of an embodiment of the present invention. DETAILED DESCRIPTION

[0045] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only 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.

[0046] Taking a certain city as an example, the technical solution of the present invention is described in detail; Figure 1 As shown, the neighborhood space gene identification and extraction method based on ancient customs paintings in this embodiment includes:

[0047] S1: Collect ancient custom paintings related to the region to be studied, obtain data sets, and screen and classify the data sets to obtain main material data sets and auxiliary material data sets;

[0048] S2: Based on the scattered perspective and narrative characteristics of ancient genre paintings, combined with the space scene conversion markers, the characteristic space scenes in ancient genre paintings are divided;

[0049] S3: Combine the functional activities, regional sections, natural environment characteristics, and architectural characteristics of each neighborhood space in the divided space scene to refine the neighborhood space characteristics, and cluster each neighborhood space based on the neighborhood space characteristics;

[0050] S4: Use image analysis and recognition tools to perform label recognition on the clustered neighborhood space images, extract the spatial elements of each neighborhood space in the neighborhood space images and convert them into text labels;

[0051] S5: Based on the neighborhood space in the scene, the configuration relationship of spatial elements is analyzed, combined with the functional interaction relationship in ancient folk paintings, and then a combination relationship is formed and screened;

[0052] S6: Verify the spatial element configuration relationship in the screened combination relationship to obtain the final neighborhood space gene.

[0053] Specifically, the above S1 includes two steps S1.1 and S1.2:

[0054] S1.1. Collection of research materials: Through five sources of information: collections of major museums at home and abroad, public publications, application software specifically for collecting traditional calligraphy and painting works, relevant literature and online information, we collected ancient genre paintings depicting the urban landscape of a specific research area and obtained dataset D.

[0055] S1.2, Research material screening - Based on the three criteria of authenticity, realism and image content, the research materials are screened, and the main materials and auxiliary materials for extracting neighborhood space genes are distinguished. Specifically, based on the authenticity of the works, paintings with questionable age, forgeries and controversial authors are excluded. Based on the nature of the image content, paintings that only reflect natural landscapes such as mountains and rivers and pastoral areas are excluded. Works with high realism and rich image content (using fine brushwork techniques and more than or equal to 4 urban space elements in the image) are used as the main materials to obtain the data set D. a , the rest of the materials are used as auxiliary materials to obtain the data set D b .

[0056] For this embodiment, the main material data set D a The content is shown in Table 1. Supplementary material dataset D b As shown in Table 2.

[0057] Table 1 Main materials

[0058]

[0059]

[0060] Table 2 Supplementary materials

[0061]

[0062]

[0063] The above S2 includes four steps: S2.1, S2.2, S2.3 and S2.4:

[0064] S2.1. Narrative plot judgment: Based on the spatial narrative characteristics of ancient genre paintings, according to the spatial and regional attributes of the elements of the painting, and combined with the theme of the work, make a judgment on the spatial narrative plot.

[0065] In this embodiment, taking "Prosperous Suzhou" as an example, the development of its narrative plot can be judged according to the painter's route, and the development of the narrative plot of the scroll can be preliminarily analyzed according to the geographical sections where the painter stayed. Therefore, "Prosperous Suzhou" can be mainly divided into five geographical sections: Lingyan Mountain, Mudu Scenery, Market Outside the City, Street Market Inside the City, and Shantang Street Section.

[0066] S2.2, scattered perspective judgment - combined with the scattered perspective characteristics of ancient folk paintings, combined with the changes in perspective and visual focus under scattered perspective, the different spatial scenes contained in the picture are judged. Specifically, the change in perspective refers to the different observation perspectives of each scene produced by the painter when depicting different scenes by adjusting the relative height between the viewpoint and the observed scene; the change in visual focus refers to the different composition centers of gravity constructed by the painter when depicting different spatial scenes. This embodiment is analyzed from two perspectives: changes in perspective and changes in visual focus.

[0067] S2.3. Determination of spatial scene conversion markers - When ancient folk paintings convert spatial scenes, they usually place spatial elements such as mountains, rivers, forests, rocks or buildings between two different scenes to play a role of natural transition. Specifically, it can be judged by the occlusion and separation relationship formed between such elements and the spatial scenes. In this embodiment, "Prosperous Map of Suzhou" mainly uses spatial elements such as mountains, rivers, smoke, clouds, bridges, etc. to divide and convert various spatial scenes through occlusion and separation relationships. A total of 12 types of conversion markers were analyzed, and the analysis results (partial) are as follows: Figure 2 , Figure 3 shown.

[0068] S2.4, spatial scene division - based on the judgment of narrative plot, combined with the analysis of perspective, visual focus change and spatial scene conversion markers under scattered perspective, the spatial scene of ancient folk paintings is divided. In this embodiment, "Suzhou Prosperous Picture" can be divided into thirteen spatial scenes: Lingyan Mountain Front, Shanyou Yaji, Mudu Street and No. 1 Scholar Ship, Suichu Garden, Lion, He Ershan, Zaoshi Street, Xumen Port, Wannian Bridge, Government Office, Huanglifang, Shuilu Changmen, Shantang Market, and Shantang River Market.

[0069] The above S3 includes two steps, S3.1 and S3.2:

[0070] S3.1. Extraction of neighborhood characteristics: First, according to the four functional activities of residence, commerce, public services and leisure and entertainment, the neighborhood spaces in each space scene are given functional interaction labels A. x ; Secondly, according to the regional segment attributes of each spatial scene, the neighborhood space in each spatial scene is assigned a regional segment label Z x ; Then, according to the proportion of natural environment in each neighborhood space N x 、Building structure ratio SR x , Building Height E x 、Architectural form characteristics B x The indicators are used to refine the natural environment characteristics and architectural characteristics of each neighborhood space in each space scene. x , Z x 、N x , SRx 、E x With B x As an indicator for comprehensively determining the spatial characteristics of neighborhood X.

[0071] S3.2 Clustering of characteristic neighborhood spaces - First, x With Z x The same neighborhood spaces are clustered into one category, which is determined as the set {T1, T2, … ,T K}; Then, determine N in proportion X , SR X 、E X With B X The weights of the four items are standardized to obtain N X ′、SR X ′、E X ′ and B X 'After that, according to The calculation formula is, at T k In the above example, the neighborhood space scenes with the final comprehensive numerical error less than α are further clustered into one category and determined as the type set t k , t k It can be regarded as a representation of the characteristics of the neighborhood space formed under the influence of specific neighborhood space genes.

[0072] In this embodiment, the best result is obtained when α is 15%. After the spatial scene division of 31 paintings depicting the urban landscape of a certain city, since this study aims to identify and extract the spatial genes of urban neighborhoods, it is necessary to exclude the spatial scenes that represent rural areas and purely suburban environments, such as the scene 1 "In front of Lingyan Mountain", the scene 2 "Mountain Tour and Elegant Gathering" and the scene 5 "Lion and He Mountain" divided from "Suzhou Prosperous Picture". On the basis of excluding such spatial scenes, the characteristics of urban neighborhoods in the remaining spatial scenes are further refined, and it is found that there are mainly four types of characteristic neighborhood spaces: town streets and rivers, city streets, city gates and commercial markets, and leisure gardens and temples. Based on the characteristics of these four types of neighborhood spaces, the divided spatial scenes are clustered.

[0073] The above S4 includes two steps, S4.1 and S4.2:

[0074] S4.1. Classification of neighborhood spatial elements - Based on the neighborhood scale and according to the classification of natural and artificial environments, the neighborhood spatial elements are classified into 10 types of spatial elements: rivers, lakes, mountains, farmlands, streets, buildings, city walls, piers, bridges, and archways.

[0075] S4.2. Neighborhood spatial element identification and extraction - for each t kThe characteristic spatial scene images in the dataset were stratified and sampled at a ratio of 10%-15%. The VisionAPI in Google Cloud Vision was called, and the label detection function (LABEL_DETECTION) in the API was used to detect the features of the sampled images. Each image was returned in Json format with a confidence score (0-1) and a multi-label set of characteristic features was obtained. Only labels with a confidence score higher than 0.7 were extracted for statistics. After screening and merging such characteristic features, spatial feature labels were obtained. Furthermore, multi-labels that were not automatically identified as neighborhood spatial features were manually added to each image sample based on the obtained spatial feature multi-labels. It was necessary to ensure that at least 50 images were annotated for each feature label. To ensure data balance, the number of images with the most labels should not exceed β times the number of images with the least labels. Then, the annotated training dataset was split into training set, validation set, and test set for training the automatic machine learning model. Finally, the trained Auto ML model was used on the Vertex AI platform to make offline batch requests for various types of characteristic spatial scene images, and the confidence threshold was set to θ to achieve automatic quantitative extraction of image spatial features.

[0076] In this embodiment, after multiple experiments, the extraction effect is best when β = 15 and θ = 0.7. Figure 4 An example of the recognition and extraction results is shown in Figure 5 shown.

[0077] The above S5 includes four steps: S5.1, S5.2, S5.3 and S5.4:

[0078] S5.1. Construction of spatial element matrix: exhaustively arrange and combine the spatial elements of each neighborhood scale identified and extracted from the neighborhood space of each characteristic scene to form a spatial element matrix. Based on the matrix, analyze the spatial element combination relationship based on the characteristic neighborhood space in each spatial scene.

[0079] S5.2. Analysis of spatial element configuration relationships - Analyze the spatial element configuration relationships of neighborhoods in ancient folk paintings from four perspectives: juxtaposition combination relationship, link dependency relationship, sequence relationship, and connectivity relationship. Specifically, the analysis of juxtaposition combination relationship can be based on spatial geometric principles such as parallelism, verticality, and continuity, and visual perception completeness principles such as proximity, similarity, and integrity, and visual perception topology theories such as overall priority. Through close combination, similar combination, and graphic combination, the relationship analysis of spatial elements with combination tendencies can be carried out; the link dependency combination relationship can be judged based on the spatial element's own attributes, that is, the spatial element that can exist as a link dependency relationship, its own attributes determine that the spatial element cannot exist independently from other spatial elements; the sequence relationship is mainly reflected in the spatial sequence formed based on the streamline and positional relationship of the spatial element; in the connectivity relationship, it means that any point in each of the two spatial elements can be connected by a path.

[0080] S5.3. Analysis of the interactive relationship between functional subjects - In combination with the realistic presentation of social life in ancient genre paintings, based on the functional interactive relationship between space and human activities, the functional behavioral interaction between human subjects within each spatial element is used to further analyze and verify the combination relationship between spatial elements.

[0081] S5.4 Clustering and screening of spatial element combination relationships - clustering the same characteristic neighborhood space types k In the process, the spatial elements extracted from the spatial scenes of the main materials are clustered to obtain t k The multiple sets of spatial element combination relationships under the above conditions are selected. Based on the two characteristics that spatial genes control the expression of urban spatial characteristics and that spatial genes are two characteristics that can form a broad consensus, these multiple sets of spatial element combination relationships are screened. First, the spatial element combination relationships that belong to the spatial element configuration relationships that will not affect the formation of the neighborhood spatial characteristics are deleted. Secondly, in combination with auxiliary materials, a set of spatial element combination relationships that are only involved in individual spatial scenes is deleted.

[0082] In this embodiment, for each characteristic scene, based on the neighborhood space elements extracted in each scene, on the basis of verifying the authenticity of its relative spatial position, in order to avoid the loss of key spatial configuration relationships, the spatial elements contained in each spatial gene exogenous trait can be arranged and combined exhaustively to form a spatial element matrix. Then, the local expression forms of various types of spatial configuration relationships are associated with the matrix through the above-mentioned analysis methods. The systematic and open nature of the spatial element matrix can support the mutual complementation and verification of multiple results. It mainly includes ten types of scene combination relationships such as Mudu Street Market and Zhuangyuan Boat, Suichu Garden, Zaoshi Street, Xumen Port, and Wannian Bridge. This patent takes Mudu Street Market and Zhuangyuan Boat as an example for detailed description.

[0083] Construct the spatial element matrix of "Mudu Town and Champion Ship", such as Figure 6 As shown in the figure, there are 30 types of spatial element combinations. Similarly, the spatial element combination relationships that cannot exist independently in the scene are merged into the spatial element combination relationships that exist with it. According to the combination relationships provided by the matrix, the combination relationships that can be established are screened according to the analysis of spatial configuration relationships and functional interaction relationships. The existence of the matrix can ensure that no omissions occur. Finally, 9 types of spatial relationships are obtained, among which some spatial configuration relationships, functional interaction relationships and their combination relationships are shown in the figure. Figure 7 shown.

[0084] The above S6 includes two steps, S6.1 and S6.2:

[0085] S6.1. Verification of urban construction concepts: Based on the characteristics that the spatial element configuration relationship is organized by the urban construction concept, eliminate the spatial element configuration relationship that does not conform to the urban construction concept generated by the harmonious interactive relationship of "space-nature-humanity".

[0086] S6.2. Diachronic screening: Based on the stable existence of spatial genes, according to the diachronic principle, the characteristic neighborhood space types extracted from ancient folk paintings are selected. k The spatial scenes in the genre are compared with the corresponding modern urban spaces, or the neighborhood spaces that can still reflect the characteristics of this neighborhood are found in the modern urban spaces. Through on-site surveys, combined with photos, documents and other relevant materials, it is judged whether the configuration relationships of each group of spatial elements under this type set can still be established in the modern urban neighborhood space. If the characteristics of the neighborhood space originally extracted from ancient genre paintings and controlled by this spatial gene are changed due to the absence or change of these groups of spatial element configuration relationships, then the neighborhood space gene has not undergone diachronic screening and should be eliminated.

[0087] In this embodiment, the spatial element configuration relationship is first verified and screened based on the concept of town construction. The spatial element configuration relationship contains the concept of town construction. Under the organization of the four concepts of ritual and music customization, adapting to local conditions, the unity of heaven, earth and man, and the functional practicality of meeting the needs of production and life, combined with the interactive process of space, nature and humanities, various spatial element configuration relationships in various characteristic neighborhoods in a certain city are formed. Therefore, based on the concept of town construction and the interactive mode of space, nature and humanities, the spatial element configuration relationship under each characteristic neighborhood space type extracted from the ancient customs paintings of a certain city can be further verified and screened.

[0088] Secondly, based on the spatial scenes in the five types of characteristic neighborhood spaces extracted from ancient genre paintings, we look for corresponding spatial scenes or spatial scenes that can reflect the characteristics of the neighborhood space in a certain city today, and based on the current neighborhood space, we carry out diachronic screening of spatial genes to determine whether the spatial element configuration relationship within the spatial gene that controls the characteristic neighborhood can still exist stably in the current neighborhood space, and whether the mutation of the spatial element configuration relationship affects the characteristic expression of the neighborhood space under the control of the spatial gene, thereby forming the final neighborhood space gene of a certain city extracted from ancient genre paintings.

[0089] After screening, this embodiment extracts four types of spatial genes: town street rivers, city streets, city gates, commercial rivers, leisure parks and temples.

[0090] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0091] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, and these changes and improvements all fall within the scope of the present invention to be protected.

Claims

1. A neighborhood space gene identification and extraction method based on ancient folk paintings, characterized in that: include: S1: Collect ancient custom paintings related to the region to be studied, obtain data sets, and screen and classify the data sets to obtain main material data sets and auxiliary material data sets; S2: Based on the scattered perspective and narrative characteristics of ancient genre paintings, combined with the space scene conversion markers, the characteristic space scenes in ancient genre paintings are divided; S3: Combine the functional activities, regional sections, natural environment characteristics, and architectural characteristics of each neighborhood space in the divided space scene to refine the neighborhood space characteristics, and cluster each neighborhood space based on the neighborhood space characteristics; S4: Use image analysis and recognition tools to perform label recognition on the clustered neighborhood space images, extract the spatial elements of each neighborhood space in the neighborhood space images and convert them into text labels; S5: Based on the neighborhood space in the scene, the configuration relationship of spatial elements is analyzed, combined with the functional interaction relationship in ancient folk paintings, and then a combination relationship is formed and screened; S6: Verify the spatial element configuration relationship in the screened combination relationship to obtain the final neighborhood space gene.

2. The method for identifying and extracting neighborhood space genes based on ancient folk paintings according to claim 1, characterized in that: The step S1 includes: taking ancient folk paintings using fine brushwork techniques and having more than or equal to 4 urban space elements in the images as the main material data set D a , the remaining ancient customs paintings are used as auxiliary material data set D b .

3. The neighborhood space gene identification and extraction method based on ancient folk paintings according to claim 1 is characterized in that: The method for judging the narrative characteristics in step S2 is: judging the spatial narrative plot according to the spatial regional attributes of the elements of the picture and the theme depicted in the work in combination with the spatial narrative characteristics shown in the ancient genre painting.

4. The method for identifying and extracting neighborhood space genes based on ancient folk paintings according to claim 1, characterized in that: The method for determining the scattered perspective characteristic in step S2 is: Different spatial scenes in the picture are judged according to the changes in perspective and visual focus under the scattered perspective; the perspective change refers to the different observation perspectives of each scene produced by the painter adjusting the relative height between the viewpoint and the observed scene when depicting different scenes; the visual focus change refers to the different composition centers of gravity constructed by the painter when depicting different spatial scenes.

5. The method for identifying and extracting neighborhood space genes based on ancient folk paintings according to claim 1, characterized in that: Step S2 combines the space scene conversion markers to divide the characteristic space scenes in the ancient folk paintings, including: Determining a spatial scene transition marker; the spatial scene transition marker is a spatial element that forms an occlusion or separation relationship between spatial scenes; Based on the judgment of narrative plot, combined with the analysis of perspective, changes in visual focus and spatial scene transition markers under scattered perspective, the spatial scenes of ancient genre paintings are divided.

6. The method for identifying and extracting neighborhood space genes based on ancient folk paintings according to claim 1, characterized in that: The step S3 comprises: S3.1: Assign functional activity labels A to the neighborhood spaces in each spatial scene X ; The functional activities include residence, commerce, public services and leisure and entertainment; According to the regional segment attributes of each spatial scene, the neighborhood space in each spatial scene is assigned a regional segment label Z X ; According to the proportion of natural environment in each neighborhood space N X 、Building structure ratio SR X , Building Height E X 、Architectural form characteristics B X The indicators are used to refine the natural environment characteristics and architectural characteristics of each neighborhood space in each space scene; X , Z X 、N X , SR X 、E X With B X As an indicator for comprehensively determining the characteristics of neighborhood space X; S3.2: A X With Z X The same neighborhood spaces are clustered into one category, which is determined as the set {T1, T2, … ,T K }, K is the number of sets; then, determine N in equal proportion X , SR X 、E X With B X The weights of the four items are standardized to obtain N X ′、SR X ′、E X ′ and B X ′, according to the formula Quantify the characteristics of each neighborhood space, in T k In the above example, the neighborhood spaces whose final comprehensive numerical error is less than the threshold α are clustered into type set t k .

7. The method for identifying and extracting neighborhood space genes based on ancient folk paintings according to claim 6 is characterized in that: The step S4 comprises: S4.1: Based on the neighborhood scale, the spatial elements of the neighborhood space are classified according to the natural environment and artificial environment classification; the spatial element classification includes rivers, lakes, mountains, farmland, streets, buildings, city walls, water piers, bridges and archways; S4.2: For each type set t k The scene images of the neighborhood spaces are stratified and sampled according to a preset ratio. The image analysis and recognition tool is called to perform feature detection on the sampled images to obtain a multi-label set of feature elements with confidence scores for the corresponding images, and the labels with confidence levels higher than a preset threshold are extracted. The corresponding feature elements are screened and merged to obtain spatial element labels. According to the spatial element labels, multiple labels that are not automatically identified as neighborhood space elements are added to all images, and the upper and lower limits of the number of images annotated with each element label are set, and the annotated images are split into training set, validation set and test set. The machine learning platform is used to optimize the scene images of various types of neighborhood spaces, and the confidence threshold is set to θ to automatically quantify the spatial elements of each neighborhood space.

8. The method for identifying and extracting neighborhood space genes based on ancient folk paintings according to claim 1, characterized in that: The step S5 comprises: The spatial elements identified and extracted from each neighborhood space are arranged and combined exhaustively to form a spatial element matrix; the spatial element configuration relationship of the neighborhood space is analyzed from multiple angles, and the functional interaction relationship between the main groups of people in each spatial element is analyzed based on the functional interaction relationship between space and crowd activities. Based on the combination relationship formed by the spatial element configuration relationship and the functional interaction relationship, the type set of the same neighborhood space is t k In the clustering, the spatial element combination relationship extracted from the neighborhood space of the main material is clustered to obtain t k The multiple sets of spatial element combination relationships under the above are selected, and based on the characteristics of spatial genes controlling the expression of urban spatial characteristics and the broad consensus characteristics of spatial genes, the multiple sets of spatial element combination relationships are screened, and the spatial element combination relationships belonging to the spatial element configuration relationships that will not affect the formation of the neighborhood spatial characteristics are deleted. In combination with auxiliary materials, the corresponding spatial element combination relationships that are only involved in individual spatial scenes are deleted.

9. The method for identifying and extracting neighborhood space genes based on ancient folk paintings according to claim 8, characterized in that: The multi-angle analysis includes juxtaposition relationship analysis, link dependency relationship analysis, sequence relationship analysis and connectivity relationship analysis; The juxtaposition combination relationship analysis is based on the principle of spatial geometry, according to the visual perception completeness principle and the visual perception topology theory, and through combination, the relationship analysis of spatial elements with combination tendency is carried out; The link dependency analysis is based on the attributes of the spatial elements themselves. The sequence relationship analysis is to analyze the spatial sequence formed based on the streamline and positional relationship of the spatial elements; The connectivity analysis is specifically to determine whether two points can be connected by a path when any point is selected in two spatial elements.

10. The method for identifying and extracting neighborhood space genes based on ancient folk paintings according to claim 1, characterized in that: The step S6 comprises: S6.1: Based on the characteristics that the spatial element configuration relationship is organized by the concept of urban construction, the spatial element configuration relationship that does not conform to the concept of urban construction is eliminated; S6.2: Based on the stable existence of spatial genes, according to the principle of diachrony, the neighborhood space type set extracted from ancient folk paintings is k The spatial scenes in the type set are compared with the corresponding modern urban spaces, or neighborhood spaces that can reflect the characteristics of this neighborhood are found in the modern urban spaces. Through on-site surveys, combined with photos and documentary materials, it is judged whether the configuration relationships of each group of spatial elements under this type set can be established in the modern urban neighborhood space. If the neighborhood space characteristics originally extracted from ancient genre paintings and controlled by the spatial gene are changed due to the absence or change of the configuration relationship of the spatial elements, it is judged that the neighborhood space gene has not undergone diachronic screening and is eliminated.

Images