Intelligent old city boundary extraction and texture calculation system based on vector topology analysis and semantic segmentation

Abstract

The invention discloses an old city boundary intelligent extraction and texture calculation system based on vector topology analysis and semantic segmentation. According to the method, the topological semantic association module provides a multi-dimensional association basis for boundary extraction by integrating spatial structure information of vector topology and functional attribute tags of semantic segmentation. In the coarse extraction stage, the objective space law of the topological features and the subjective function orientation of the semantic tags are mutually verified, so that the misjudgment of'similar forms but inconsistent functions' possibly caused by single dependence on the topological structure or the deviation of'tag coverage but space breakage 'possibly caused by only dependence on the semantic tags is avoided; in the fine correction stage, the module further dynamically calibrates the boundary range through historical-current situation space-time correlation and multi-source data conflict detection, so that the continuity of historical stable elements is reserved, the updating requirement of current situation semantic tags is included, the final boundary better fits the essential characteristics of'active inheritance 'of the old city, and the accuracy of the final boundary is improved. And human experience interference and errors are obviously reduced.

Description

Technical Field

[0001] This invention belongs to the field of old city boundary calculation technology, specifically an intelligent extraction and texture calculation system for old city boundaries based on vector topology analysis and semantic segmentation. Background Technology

[0002] Old town boundary extraction and texture calculation is a comprehensive research method combining remote sensing technology, Geographic Information System (GIS), image processing, and urban morphology analysis. It aims to accurately identify and quantitatively assess the boundary range and spatial texture characteristics of historical urban areas. This technology extracts the spatial boundaries of the old town through the fusion analysis of multi-source data (such as historical maps, aerial imagery, satellite remote sensing data, and 3D laser scanning). It then uses morphological, texture analysis, and spatial syntax methods to model and calculate texture elements such as street networks, building density, spatial scale, and land use types, thereby revealing the characteristics of the old town in terms of spatial organization, morphological evolution, and cultural heritage. This method has significant application value in the fields of historical district protection, urban renewal planning, and cultural landscape research, providing not only a scientific basis for urban heritage protection but also technical support for sustainable urban development and the preservation of cultural memory.

[0003] However, existing technologies for extracting the boundaries of old towns often rely on a single data source (such as based solely on morphological features or a single semantic label), which can easily lead to misjudgments such as "similar in form but inconsistent in function" or "label coverage but spatial fragmentation." Furthermore, they lack mechanisms for historical-current spatiotemporal correlation and multi-source data conflict detection, resulting in boundary delineation being highly susceptible to human experience and prone to errors. Summary of the Invention

[0004] The purpose of this invention is to provide an intelligent extraction and texture calculation system for old city boundaries based on vector topology analysis and semantic segmentation in order to solve the problems mentioned above.

[0005] The technical solution adopted in this invention is as follows: an intelligent extraction and texture calculation system for old city boundaries based on vector topology analysis and semantic segmentation, the system comprising: a data acquisition and preprocessing module, a boundary coarse extraction module, a boundary fine correction module, a topological semantic association module, an old city texture quantitative calculation module, and a visualization and output module;

[0006] The internal components of the topological semantic association module include: a data association rule base construction module, a cross-source topological-semantic mapping module, a spatiotemporal association calibration module, and an association verification and feedback module.

[0007] The output of the data acquisition and preprocessing module is connected to the input of the boundary coarse extraction module, providing it with preprocessed multi-source data;

[0008] The output of the boundary coarse extraction module is connected to the input of the boundary fine correction module, and the coarse extraction boundary is used as the initial range for fine correction.

[0009] The output of the topological semantic association module is connected to the input of the boundary coarse extraction module, the boundary fine correction module, and the old city texture quantification calculation module, respectively, providing the former two with the association data of topological features and semantic labels, and providing the latter with the quantitative basis for topological-semantic fusion;

[0010] The output of the boundary fine correction module is connected to the input of the old city texture quantification calculation module, and the fine correction boundary is used as the spatial range for texture calculation.

[0011] The output of the old city texture quantitative calculation module is connected to the input of the visualization and output module. The quantitative results such as street density and functional mixing index are input into the visualization process, and finally the visualization and output module integrates the output of each module.

[0012] In a preferred embodiment, the data acquisition and preprocessing module internally includes a multi-source data integration module, a coordinate unification processing module, a noise filtering and alignment module, and a data format standardization module. The multi-source data integration module is responsible for collecting remote sensing imagery (such as 2025 high-resolution satellite imagery), historical maps, vector topology data, semantic segmentation results, and supplementary data. The coordinate unification processing module converts all data to the WGS84 geographic coordinate system using a geographic information system to ensure spatial consistency. The noise filtering and alignment module removes clouds from the remote sensing imagery, corrects broken nodes in the vector data, and uses a feature matching algorithm to achieve spatial registration between historical maps and current imagery. The data format standardization module converts unstructured data into structured attribute tables, providing a unified data input basis for subsequent modules.

[0013] In a preferred embodiment, the boundary coarse extraction module internally includes a topological feature extraction module, a semantic label filtering module, a threshold joint determination module, and a coarse boundary generation module. The topological feature extraction module calculates key parameters such as street connectivity, building clustering, and fault zone indicators based on preprocessed vector data. The semantic label filtering module uses semantic segmentation results to extract the spatial distribution range of core labels such as "traditional streets and alleys," "cultural heritage buildings," and "historical courtyards." The threshold joint determination module filters areas that conform to the characteristics of the old city by setting topological feature thresholds and semantic label coverage thresholds. Based on the above filtering results, the coarse boundary generation module extracts the approximate outline of the old city to form the initial boundary range.

[0014] In a preferred embodiment, the boundary refinement module internally includes a spatiotemporal sequence correlation analysis module, a multi-source data conflict detection module, a weight dynamic adjustment module, and a refined boundary refinement output module. The spatiotemporal sequence correlation analysis module introduces historical-current time series data, calculates the spatial overlap between historical and current boundaries (based on the IoU index), and identifies "stable boundary elements" and "boundary drift zones." The multi-source data conflict detection module compares the topological analysis results with semantic label data to identify spatial offsets or attribute contradictions. The weight dynamic adjustment module assigns weights to drift zones and conflict zones: the weight of stable elements decreases with time span but is strengthened by overlap, while the weight of current semantic labels increases with the degree of drift. The refined boundary refinement output module generates a "refined boundary" through weighted refinement, balancing historical continuity and current accuracy.

[0015] In a preferred embodiment, the data association rule base construction module internally includes a multi-source rule extraction unit, a rule classification and standardization unit, a rule storage and indexing unit, and a dynamic update triggering unit. The multi-source rule extraction unit extracts different types of association rules based on historical data, domain expert experience, and algorithmic derivation (mining implicit relationships between topological features and semantic tags through association analysis algorithms). The rule classification and standardization unit categorizes the extracted rules according to data type and application scenario, and standardizes the rule expression format. The rule storage and indexing unit stores the standardized rules in a structured database, creating an index by category tags to support fast retrieval. The dynamic update triggering unit sets a rule timeliness threshold, triggering the rule update process when new data is input or rule application feedback occurs.

[0016] In a preferred embodiment, the cross-source topology-semantic mapping module first spatially aligns the vector topology data with the pixel-level probability map output by semantic segmentation based on a preprocessed unified coordinate system, ensuring that the topological features and semantic labels of the same geographic unit strictly correspond in spatial location. Secondly, it extracts key features from the vector topology data: connectivity C (connectivity strength of the street network, value 0-1), clustering AA (density of building clusters, value 0-1), and fault zone index FF (identification of abrupt changes in topological structure, value 0-1, 0 for no faults, 1 for strong faults). Then, using these topological features as input and semantic labels LL (discrete variables, e.g., L=1 represents "traditional streets," L=2 represents "protected cultural relics") as output, it trains the mapping relationship between topological features and semantic labels using a weighted Bayesian probability model. Finally, it generates a "topology-semantic association matrix," where each element Mi,j represents the association probability of the j-th type of semantic label for the i-th spatial unit, used for weight allocation in subsequent boundary verification and texture calculation.

[0017] The mapping between topological features and semantic labels is achieved through the following probability formula:

[0018] P(L=l∣C,A,F)=σ(w C ·C+w A ·A+w F ·F+b)

[0019] In the formula:

[0020] P(L=l|C,A,F) represents the probability that a spatial unit belongs to a semantic label l (such as "traditional streets and alleys") given topological features C (connectivity), A (clustering), and F (fracture zone index).

[0021] σ(·) represents the Sigmoid activation function, which maps the linear combination result to the interval [0,1] and outputs a probability value.

[0022] wC, wA, and wF are the weight coefficients for connectivity, clustering, and fault zone indices, respectively. They are obtained through training on historical labeled data and reflect the degree of contribution of each topological feature to the target semantic label.

[0023] b represents the bias term, which is used to adjust the base probability offset.

[0024] The innovation of this formula lies in the fact that by explicitly introducing a weighted combination of multi-dimensional topological features, it breaks through the limitations of traditional single-feature mapping. At the same time, the introduction of the Sigmoid function makes the probability output more consistent with the semantic association characteristics of "non-absolute matching" in real-world scenarios (such as highly connected streets and alleys may partially belong to "traditional commercial streets" rather than being 100% matched), thus improving the robustness of the mapping results.

[0025] In a preferred embodiment, the spatiotemporal correlation calibration module resolves the boundary discrepancy between historical and current data through cross-calibration of time and spatial dimensions. The specific process is as follows: First, time-series data alignment is performed—historical maps and current data are unified to the WGS84 coordinate system, and "stable boundary elements" (such as unmodified river channels, preserved city wall foundations, and other geographic entities that have not changed significantly over time) are extracted. Second, spatial difference detection is conducted—the spatial overlap between historical and current boundaries is calculated, "boundary drift zones" are identified, and the degree of drift is quantified. Finally, calibration rules are applied—for drift zones, based on the historical continuity of stable elements and the accuracy of current semantic labels, the weights of historical and current data are dynamically adjusted to perform weighted correction on the boundaries: the weight of stable elements decreases with increasing time span but is strengthened by spatial overlap; the weight of current semantic labels increases with increasing drift, ultimately generating a "spatiotemporally calibrated correlation boundary."

[0026] The weight allocation for boundary correction is achieved using the following formula:

[0027] whist =γ·e -δ·ΔT ·O;

[0028] w hist The weighting coefficient (0-1) represents the impact of historical stable factors on boundary correction, reflecting the degree of influence of historical data on the current boundary.

[0029] γ represents the initial weight coefficient (empirical value, such as 0.6), which represents the basic weight of historically stable elements without time decay.

[0030] e -δ·ΔT This represents the time decay term, where δ is the time decay coefficient (controlling the decay rate, such as 0.05 / year), and ΔT is the time span between historical and current data. The longer the time period, the smaller the direct impact of historical data on the current situation, and the lower its weight.

[0031] O represents the spatial overlap between the historical boundary and the current boundary (0-1), calculated using the IoU index (O = intersection area of ​​historical and current boundaries / union area of ​​historical and current boundaries). The higher the overlap, the stronger the continuity of the historical boundary, and the greater its weight.

[0032] In a preferred embodiment, the association verification and feedback module internally includes a sample data verification unit, a conflict detection and annotation unit, a feedback correction execution unit, and a rule version management unit. The sample data verification unit extracts verification samples from multi-source data (e.g., randomly selecting 10% of historical-current boundary data), applies association rules from the rule base to the samples, and calculates the rule matching accuracy (e.g., the percentage of samples correctly associated by the rule). The conflict detection and annotation unit compares the rule application results with the actual data, identifies rule conflicts (e.g., the rule requires "high connectivity streets and alleys to match traditional commercial street labels," but the actual data does not match), and annotates the conflict type (spatial misalignment, attribute contradiction, etc.). The feedback correction execution unit generates correction strategies based on the conflict type: adjusting parameters for rule threshold deviations (e.g., excessively high connectivity thresholds) and supplementing new rules for incomplete rule coverage (e.g., missing "historical courtyard" label association). The rule version management unit records the time, content, and verification accuracy of each rule correction, forming a rule version chain (e.g., V1.0 initial rule, V1.1 corrected rule), supporting version backtracking and comparison.

[0033] In a preferred embodiment, the old city fabric quantification module quantifies spatial characteristics through multi-dimensional indicators, providing data support for conservation planning. The specific process is as follows: First, basic data is extracted from vector topology analysis results and semantic segmentation labels, and analysis units are divided into 10m×10m grids. Second, quantification indicators are designed for the three major fabric dimensions: street and alley fabric calculates street and alley density and directional entropy based on the connectivity and directional distribution of vector topology; architectural fabric extracts building density (building base area / unit area), height variance (reflecting spatial hierarchy; the larger the variance, the more pronounced the staggered heights), and the proportion of historical buildings (historical building area / total building area) based on the semantic segmentation of building outlines; functional fabric utilizes the classification statistics of semantic labels to calculate the land use ratio of residential / commercial / public spaces, and quantifies functional diversity through a mixed-function index (the index significantly increases when a unit contains more than three types of functions). Finally, after normalizing each indicator (0-1), a fabric heatmap is generated and overlaid onto the old city boundary vector map, visually displaying the differences in fabric characteristics between different areas.

[0034] The functional mix index is calculated using the following formula:

[0035]

[0036] H represents the functional mixing index (Shannon entropy), which takes a value of 0-1 (after normalization). The larger the value, the higher the degree of functional mixing and the stronger the spatial vitality.

[0037] n represents the number of functional types contained within the unit (such as residential, commercial, public space, etc., n≥1).

[0038] pi represents the proportion of the area of ​​the i-th functional land use to the total area of ​​the unit. ).

[0039] In a preferred embodiment, the visualization and output module internally includes a texture index heatmap generation module, a multi-layer spatial overlay module, a multi-format output module, and a basic information annotation module. The texture index heatmap generation module normalizes quantitative indicators such as street density, building height variance, and functional mixing index (0-1), and generates intuitive heatmaps through color gradation mapping (cool colors represent low texture values, and warm colors represent high texture values). The multi-layer spatial overlay module spatially overlays refined boundary vector data with the heatmap and semantic label map to form a "boundary-texture-function" composite layer. The multi-format output module supports various formats, including PDF analysis reports (containing heatmaps and indicator descriptions), SHP vector boundary files (for GIS system calls), and PNG visualization images (for planning reports). The basic information annotation module adds indicator explanations, legends, and other content to the output to improve readability.

[0040] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0041] 1. In this invention, the topological semantic association module integrates spatial structural information of vector topology (such as quantitative features like street connectivity and building clustering) with functional attribute labels of semantic segmentation (such as semantic information like "traditional streets" and "cultural heritage buildings"), providing multi-dimensional association basis for boundary extraction. In the coarse extraction stage, the objective spatial laws of topological features and the subjective functional orientation of semantic labels mutually verify each other, avoiding misjudgments such as "similar form but inconsistent function" that may be caused by relying solely on topological structure, or deviations such as "label coverage but spatial fragmentation" that may occur by relying solely on semantic labels. In the fine correction stage, the module further dynamically calibrates the boundary range through historical-current spatiotemporal association and multi-source data conflict detection—preserving the continuity of historically stable elements (such as unmodified river channels and city wall foundations) while incorporating the updating needs of current semantic labels (such as boundary drift caused by urban expansion), making the final boundary more in line with the essential characteristics of the "living heritage" of the old city. Compared with traditional boundary delineation methods based solely on form or single data, this significantly reduces human experience interference and errors.

[0042] 2. In this invention, the topological semantic association module provides the underlying support for the quantitative calculation of urban fabric by integrating "space and function," enabling fabric indicators to not only reflect physical form (such as street density and building height variance) but also to correlate with functional attributes (such as the coverage of "cultural heritage building" labels and the matching degree of "traditional commercial street" labels). For example, the generation of fabric heatmaps is no longer limited to simple spatial density distribution, but through topological semantic association, "areas with high street connectivity" are superimposed and mapped with "traditional commercial street label coverage areas," intuitively presenting the core protection area of ​​"complete form and active function." The multi-layer overlay output further integrates boundary, fabric, and functional information, providing a composite reference of "spatial scope, morphological characteristics, and cultural value" for planning decisions. This integrated calculation method breaks through the limitations of traditional fabric analysis that "emphasizes form and neglects function," enabling the results to guide the protection and restoration of physical space and support the revitalization and utilization of cultural functions, effectively improving the operability and accuracy of old city protection and renewal. Attached Figure Description

[0043] Figure 1 This is an overall system block diagram of the present invention;

[0044] Figure 2 This is a system block diagram of the topological semantic association module in this invention. Detailed Implementation

[0045] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0046] Reference Figure 1-2 A smart system for extracting the boundaries and calculating the texture of old city areas based on vector topology analysis and semantic segmentation. The system includes: a data acquisition and preprocessing module, a coarse boundary extraction module, a fine boundary correction module, a topological semantic association module, a quantitative calculation module for the texture of old city areas, and a visualization and output module.

[0047] The topological semantic association module has the following internal components: a data association rule base construction module, a cross-source topological-semantic mapping module, a spatiotemporal association calibration module, and an association verification and feedback module.

[0048] The output of the data acquisition and preprocessing module is connected to the input of the boundary coarse extraction module to provide it with preprocessed multi-source data;

[0049] The output of the boundary coarse extraction module is connected to the input of the boundary fine correction module, and the coarse extraction boundary is used as the initial range for fine correction.

[0050] The output of the topological semantic association module is connected to the input of the boundary coarse extraction module, the boundary fine correction module, and the old city texture quantization calculation module, respectively, providing the former two with the association data of topological features and semantic labels, and providing the latter with the quantitative basis for topological-semantic fusion;

[0051] The output of the boundary fine correction module is connected to the input of the old city texture quantification calculation module, and the fine correction boundary is used as the spatial range for texture calculation.

[0052] The output of the old city texture quantitative calculation module is connected to the input of the visualization and output module. The quantitative results such as street density and functional mixing index are input into the visualization stage, and finally the visualization and output module integrates the output of each module.

[0053] The data acquisition and preprocessing module internally comprises a multi-source data integration module, a coordinate unification processing module, a noise filtering and alignment module, and a data format standardization module. The multi-source data integration module is responsible for collecting remote sensing imagery (such as 2025 high-resolution satellite imagery), historical maps, vector topology data, semantic segmentation results, and supplementary data. The coordinate unification processing module converts all data to the WGS84 geographic coordinate system using a geographic information system to ensure spatial consistency. The noise filtering and alignment module removes clouds from the remote sensing imagery, corrects broken nodes in the vector data, and uses feature matching algorithms to achieve spatial registration between historical maps and current imagery. The data format standardization module converts unstructured data into structured attribute tables, providing a unified data input basis for subsequent modules.

[0054] The coarse boundary extraction module internally comprises a topological feature extraction module, a semantic label filtering module, a threshold joint determination module, and a coarse boundary generation module. The topological feature extraction module calculates key parameters such as street connectivity, building clustering, and fault zone indicators based on preprocessed vector data. The semantic label filtering module utilizes semantic segmentation results to extract the spatial distribution range of core labels such as "traditional streets and alleys," "cultural heritage buildings," and "historical courtyards." The threshold joint determination module filters areas that conform to the characteristics of the old city by setting topological feature thresholds and semantic label coverage thresholds. Based on the above filtering results, the coarse boundary generation module extracts the approximate outline of the old city, forming the initial boundary range.

[0055] The boundary refinement module internally comprises a spatiotemporal sequence correlation analysis module, a multi-source data conflict detection module, a dynamic weight adjustment module, and a refined boundary refinement output module. The spatiotemporal sequence correlation analysis module incorporates historical-current time series data to calculate the spatial overlap between historical and current boundaries (based on the IoU index), identifying "stable boundary elements" and "boundary drift zones." The multi-source data conflict detection module compares topological analysis results with semantic label data to identify spatial offsets or attribute contradictions. The dynamic weight adjustment module assigns weights to drift and conflict zones: the weight of stable elements decreases over time but is strengthened by overlap, while the weight of current semantic labels increases with the degree of drift. The refined boundary refinement output module generates a "refined boundary" through weighted refinement, balancing historical continuity with current accuracy.

[0056] The data association rule base construction module internally comprises a multi-source rule extraction unit, a rule classification and standardization unit, a rule storage and indexing unit, and a dynamic update triggering unit. The multi-source rule extraction unit extracts different types of association rules based on historical data, domain expert experience, and algorithmic derivation (mining implicit relationships between topological features and semantic tags through association analysis algorithms). The rule classification and standardization unit categorizes the extracted rules according to data type and application scenario, and standardizes the rule expression format. The rule storage and indexing unit stores the standardized rules in a structured database, creating an index by category tags to support fast retrieval. The dynamic update triggering unit sets rule timeliness thresholds, triggering the rule update process when new data is input or rule application feedback occurs.

[0057] The cross-source topology-semantic mapping module first spatially aligns the vector topology data with the pixel-level probability map output from semantic segmentation, based on a preprocessed unified coordinate system, ensuring a strict spatial correspondence between the topological features and semantic labels of the same geographic unit. Second, it extracts key features from the vector topology data: connectivity C (connectivity strength of the street network, value 0-1), clustering AA (density of building clusters, value 0-1), and fault zone index FF (identification of abrupt changes in topological structure, value 0-1, 0 for no faults, 1 for strong faults). Then, using these topological features as input and semantic labels LL (discrete variables, e.g., L=1 represents "traditional streets," L=2 represents "protected cultural relics") as output, it trains the mapping relationship between topological features and semantic labels using a weighted Bayesian probability model. Finally, it generates a "topology-semantic association matrix," where each element Mi,j represents the association probability of the j-th type of semantic label for the i-th spatial unit, used for weight allocation in subsequent boundary verification and texture calculation.

[0058] The mapping between topological features and semantic labels is achieved through the following probability formula:

[0059] P(L=l∣C,A,F)=σ(w C ·C+w A ·A+w F ·F+b)

[0060] In the formula:

[0061] P(L=l|C,A,F) represents the probability that a spatial unit belongs to a semantic label l (such as "traditional streets and alleys") given topological features C (connectivity), A (clustering), and F (fracture zone index).

[0062] σ(·) represents the Sigmoid activation function, which maps the linear combination result to the interval [0,1] and outputs a probability value.

[0063] wC, wA, and wF are the weight coefficients for connectivity, clustering, and fault zone indices, respectively. They are obtained through training on historical labeled data and reflect the degree of contribution of each topological feature to the target semantic label.

[0064] b represents the bias term, which is used to adjust the base probability offset.

[0065] The innovation of this formula lies in the fact that by explicitly introducing a weighted combination of multi-dimensional topological features, it breaks through the limitations of traditional single-feature mapping. At the same time, the introduction of the Sigmoid function makes the probability output more consistent with the semantic association characteristics of "non-absolute matching" in real-world scenarios (such as highly connected streets and alleys may partially belong to "traditional commercial streets" rather than being 100% matched), thus improving the robustness of the mapping results.

[0066] The spatiotemporal correlation calibration module addresses the boundary discrepancies between historical and current data through cross-calibration of time and space dimensions. The specific process is as follows: First, time-series data alignment is performed—historical maps and current data are unified to the WGS84 coordinate system, extracting "stable boundary elements" (such as unmodified river channels, preserved city wall foundations, and other geographic entities that have not changed significantly over time). Second, spatial difference detection is conducted—the spatial overlap between historical and current boundaries is calculated, "boundary drift zones" are identified, and the degree of drift is quantified. Finally, calibration rules are applied—for drift zones, based on the historical continuity of stable elements and the accuracy of current semantic labels, the weights of historical and current data are dynamically adjusted to perform weighted correction on the boundaries: the weight of stable elements decreases with increasing time span but is strengthened by spatial overlap; the weight of current semantic labels increases with increasing drift, ultimately generating a "spatiotemporally calibrated correlation boundary."

[0067] The weight allocation for boundary correction is achieved using the following formula:

[0068] w hist =γ·e -δ·ΔT ·O;

[0069] w hist The weighting coefficient (0-1) represents the impact of historical stable factors on boundary correction, reflecting the degree of influence of historical data on the current boundary.

[0070] γ represents the initial weight coefficient (empirical value, such as 0.6), which represents the basic weight of historically stable elements without time decay.

[0071] e -δ·ΔT This represents the time decay term, where δ is the time decay coefficient (controlling the decay rate, such as 0.05 / year), and ΔT is the time span between historical and current data. The longer the time period, the smaller the direct impact of historical data on the current situation, and the lower its weight.

[0072] O represents the spatial overlap between the historical boundary and the current boundary (0-1), calculated using the IoU index (O = intersection area of ​​historical and current boundaries / union area of ​​historical and current boundaries). The higher the overlap, the stronger the continuity of the historical boundary, and the greater its weight.

[0073] The innovation of this formula lies in its first-ever combination of time decay and spatial overlap, dynamically adjusting the weight of historical data. This avoids both the boundary obsolescence problem caused by "complete reliance on historical data" and the drawback of "over-reliance on current data" which ignores historical continuity. For example, if a region has a high overlap between historical and current boundaries (O = 0.8) and a short time span (ΔT = 30 years), then w hist When the values ​​are relatively high (e.g., γ = 0.6, δ = 0.05), w hist =0.6·e -0.05×30 (0.8≈0.6×0.223×0.8≈0.107), historical data has a more significant impact on boundary correction;

[0074] Conversely, if the overlap is low (O = 0.3) and the time span is large (ΔT = 75 years), then w hist Reduce (e.g., 0.6×e) -0.05×75 ×0.3≈0.6×0.023×0.3≈0.004), the current data dominates the boundary correction to ensure that the results conform to the current reality.

[0075] The association verification and feedback module internally comprises a sample data verification unit, a conflict detection and annotation unit, a feedback correction execution unit, and a rule version management unit. The sample data verification unit extracts verification samples from multi-source data (e.g., randomly selecting 10% of historical-current boundary data), applies association rules from the rule base to the samples, and calculates the rule matching accuracy (e.g., the percentage of samples correctly associated by the rule). The conflict detection and annotation unit compares the rule application results with the actual data, identifies rule conflicts (e.g., the rule requires "high connectivity streets and alleys to match traditional commercial street labels," but the actual data does not match), and annotates the conflict type (spatial misalignment, attribute contradiction, etc.). The feedback correction execution unit generates correction strategies based on the conflict type: adjusting parameters for rule threshold deviations (e.g., excessively high connectivity thresholds) and supplementing new rules for incomplete rule coverage (e.g., missing "historical courtyard" label association). The rule version management unit records the time, content, and verification accuracy of each rule correction, forming a rule version chain (e.g., V1.0 initial rule, V1.1 corrected rule), supporting version backtracking and comparison.

[0076] The Old Town Fabric Quantification Calculation Module quantifies spatial characteristics through multi-dimensional indicators, providing data support for conservation planning. The specific process is as follows: First, basic data is extracted from vector topology analysis results and semantic segmentation labels, and analysis units are divided into 10m×10m grids. Second, quantitative indicators are designed for the three fabric dimensions: Street and alley fabric calculates street and alley density and directional entropy based on the connectivity and directional distribution of vector topology; architectural fabric extracts building density (building base area / unit area), height variance (reflecting spatial hierarchy; the larger the variance, the more pronounced the staggered heights), and the proportion of historical buildings (historical building area / total building area) based on the semantically segmented building outlines; functional fabric utilizes the classification statistics of semantic labels to calculate the land use ratio of residential / commercial / public spaces, and quantifies functional diversity through a mixed-function index (the index significantly increases when a unit contains more than three types of functions). Finally, after normalizing each indicator (0-1), a fabric heatmap is generated and overlaid onto the Old Town boundary vector map, visually displaying the differences in fabric characteristics between different areas.

[0077] The functional mix index is calculated using the following formula:

[0078]

[0079] H represents the functional mixing index (Shannon entropy), which takes a value of 0-1 (after normalization). The larger the value, the higher the degree of functional mixing and the stronger the spatial vitality.

[0080] n represents the number of functional types contained within the unit (such as residential, commercial, public space, etc., n≥1).

[0081] pi represents the proportion of the area of ​​the i-th functional land use to the total area of ​​the unit. ).

[0082] The visualization and output module includes a texture index heatmap generation module, a multi-layer spatial overlay module, a multi-format output module, and a basic information annotation module. The texture index heatmap generation module normalizes quantitative indicators such as street density, building height variance, and functional mixing index (0-1), and generates intuitive heatmaps through color mapping (cool colors represent low texture values, and warm colors represent high texture values). The multi-layer spatial overlay module spatially overlays refined boundary vector data with heatmaps and semantic label maps to form a composite layer of "boundary-texture-function". The multi-format output module supports various formats, including PDF analysis reports (containing heatmaps and indicator descriptions), SHP vector boundary files (for GIS system calls), and PNG visualization images (for planning reports). The basic information annotation module adds indicator explanations and legends to the outputs to improve readability.

[0083] From the above, we can conclude that:

[0084] In this invention, the topological semantic association module integrates spatial structural information of vector topology (such as quantitative features like street connectivity and building clustering) with functional attribute labels of semantic segmentation (such as semantic information like "traditional streets" and "cultural heritage buildings"), providing multi-dimensional association basis for boundary extraction. In the coarse extraction stage, the objective spatial laws of topological features and the subjective functional orientation of semantic labels mutually verify each other, avoiding misjudgments such as "similar form but inconsistent function" that may result from relying solely on topological structure, or deviations such as "label coverage but spatial fragmentation" that may occur from relying solely on semantic labels. In the fine correction stage, the module further dynamically calibrates the boundary range through historical-current spatiotemporal association and multi-source data conflict detection—preserving the continuity of historically stable elements (such as unmodified river channels and city wall foundations) while incorporating the updating needs of current semantic labels (such as boundary drift caused by urban expansion), making the final boundary more in line with the essential characteristics of the old city's "living heritage." Compared with traditional boundary delineation methods based solely on form or single data, this significantly reduces interference and errors from human experience.

[0085] In this invention, the topological semantic association module provides the underlying support for the quantitative calculation of urban fabric by integrating "space and function," enabling fabric indicators to not only reflect physical form (such as street density and building height variance) but also to correlate with functional attributes (such as the coverage of "cultural heritage building" labels and the matching degree of "traditional commercial street" labels). For example, the generation of fabric heatmaps is no longer limited to simple spatial density distribution, but through topological semantic association, "areas with high street connectivity" are superimposed and mapped with "traditional commercial street label coverage areas," intuitively presenting core protected areas that are "morphologically complete and functionally active." The multi-layer overlay output further integrates boundary, fabric, and functional information, providing a composite reference of "spatial scope, morphological characteristics, and cultural value" for planning decisions. This integrated calculation method breaks through the limitations of traditional fabric analysis that "emphasizes form and neglects function," enabling the results to guide the protection and restoration of physical space and support the revitalization and utilization of cultural functions, effectively improving the operability and accuracy of old city protection and renewal.

[0086] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the term "comprising" or any other variations thereof is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0087] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A system for intelligent extraction and texture calculation of old city boundaries based on vector topology analysis and semantic segmentation, characterized in that: The system includes: a data acquisition and preprocessing module, a boundary coarse extraction module, a boundary fine correction module, a topological semantic association module, an old city texture quantitative calculation module, and a visualization and output module; The internal components of the topological semantic association module include: a data association rule base construction module, a cross-source topological-semantic mapping module, a spatiotemporal association calibration module, and an association verification and feedback module. The output of the data acquisition and preprocessing module is connected to the input of the boundary coarse extraction module, providing it with preprocessed multi-source data; The output of the boundary coarse extraction module is connected to the input of the boundary fine correction module, and the coarse extraction boundary is used as the initial range for fine correction. The output of the topological semantic association module is connected to the input of the boundary coarse extraction module, the boundary fine correction module, and the old city texture quantification calculation module, respectively. The output of the boundary fine correction module is connected to the input of the old city texture quantification calculation module; The output of the old city texture quantitative calculation module is connected to the input of the visualization and output module. The quantitative results such as street density and functional mixing index are input into the visualization process, and finally the visualization and output module integrates the output of each module.

2. The intelligent extraction and texture calculation system for old city boundaries based on vector topology analysis and semantic segmentation as described in claim 1, characterized in that: The data acquisition and preprocessing module is internally configured with a multi-source data integration module, a coordinate unification processing module, a noise filtering and alignment module, and a data format standardization module.

3. The intelligent extraction and texture calculation system for old city boundaries based on vector topology analysis and semantic segmentation as described in claim 1, characterized in that: The boundary coarse extraction module internally includes a topological feature extraction module, a semantic label filtering module, a threshold joint determination module, and a coarse boundary generation module.

4. The intelligent extraction and texture calculation system for old city boundaries based on vector topology analysis and semantic segmentation as described in claim 1, characterized in that: The boundary fine correction module internally includes a spatiotemporal sequence correlation analysis module, a multi-source data conflict detection module, a weight dynamic adjustment module, and a fine boundary correction output module.

5. The intelligent extraction and texture calculation system for old city boundaries based on vector topology analysis and semantic segmentation as described in claim 1, characterized in that: The data association rule base construction module is internally configured with a multi-source rule extraction unit, a rule classification and standardization unit, a rule storage and indexing unit, and a dynamic update triggering unit.

6. The intelligent extraction and texture calculation system for old city boundaries based on vector topology analysis and semantic segmentation as described in claim 1, characterized in that: The mapping probability formula between the topological features and semantic tags of the cross-source topology-semantic mapping module is as follows: P(L=l∣C,A,F)=σ(w C ·C+w A ·A+w F ·F+b) In the formula: P(L=l|C,A,F) represents the probability that a spatial unit belongs to semantic label l given topological features C, A, and F; σ(·) represents the Sigmoid activation function, which maps the linear combination result to the interval [0,1] and outputs a probability value; wC, wA, and wF are the weight coefficients of connectivity, clustering, and fault zone indices, respectively. They are obtained through training with historical labeled data and reflect the degree of contribution of each topological feature to the target semantic label. b represents the bias term, which is used to adjust the base probability offset.

7. The intelligent extraction and texture calculation system for old city boundaries based on vector topology analysis and semantic segmentation as described in claim 1, characterized in that: The weight allocation formula for the boundary correction of the spatiotemporal correlation calibration module is as follows: w hist =γ·e -δ·ΔT ·O; w hist The weighting coefficients of historical stability factors on boundary correction reflect the degree of influence of historical data on the current boundary. γ represents the initial weight coefficient, which represents the basic weight of historically stable elements without time decay; e -δ·ΔT This represents the time decay term, where δ is the time decay coefficient and ΔT is the time span between historical data and current data; the longer the time, the smaller the direct impact of historical data on the current situation, and the lower the weight accordingly. O represents the spatial overlap between the historical boundary and the current boundary, calculated using the IoU index; the higher the overlap, the stronger the continuity of the historical boundary, and the greater the weight.

8. The intelligent extraction and texture calculation system for old city boundaries based on vector topology analysis and semantic segmentation as described in claim 1, characterized in that: The associated verification and feedback module is internally configured with a sample data verification unit, a conflict detection and annotation unit, a feedback correction execution unit, and a rule version management unit.

9. The intelligent extraction and texture calculation system for old city boundaries based on vector topology analysis and semantic segmentation as described in claim 1, characterized in that: The formula for calculating the functional hybrid index of the old city texture quantitative calculation module is as follows: H represents the functional mixing index, which ranges from 0 to 1. The larger the value, the higher the degree of functional mixing and the stronger the spatial vitality. n represents the number of functional types contained in the unit; pi represents the proportion of the area of ​​the i-th functional land use to the total area of ​​the unit.

10. The intelligent extraction and texture calculation system for old city boundaries based on vector topology analysis and semantic segmentation as described in claim 1, characterized in that: The visualization and output module includes a texture index heatmap generation module, a multi-layer spatial overlay module, a multi-format output module, and a basic information annotation module.

Images