A confidence field-based spatial intelligent historical block style evaluation method and system

Abstract

The application provides a confidence field-based space intelligent historical block style evaluation method and system, relates to the technical field of building planning, and comprises the following steps: acquiring multi-dimensional features of a historical block; based on the multi-dimensional features, an evaluation model corresponding to a decision type, a cultural context of a local style feature set, and previous expert evaluation experience of the region, comprehensively evaluating a preliminary evaluation result of the historical block and corresponding comprehensive confidence, and positioning weak dimensions; if the comprehensive confidence does not meet the target requirement, the weak link is reviewed to obtain a review result; based on the preliminary evaluation result and the review result, a final evaluation result is determined; the model cross-region generalization confidence, the local cultural context compliance degree and the consistency with the expert evaluation historical experience are coupled into dynamic confidence to provide the accuracy of the confidence evaluation, and the review is performed for the weak link, the resource input is more accurate, only the evaluation conclusion with a cross-region generalization risk triggers directional review, and the evaluation efficiency is improved.

Description

Technical Field

[0001] This application relates to the field of architectural planning technology, specifically to a spatial intelligent historical district appearance assessment method and system based on confidence fields. Background Technology

[0002] The assessment and preservation of the historical character of districts is a crucial issue in urban renewal and cultural heritage management. With the rapid development of artificial intelligence technology, visual language models offer a new technological approach for the automated identification and assessment of the historical character of buildings.

[0003] Currently, in the field of architectural style identification, researchers have constructed a quantitative analysis framework for architectural styles based on visual language models, achieving automatic identification, comparison, and accurate classification of architectural images, and outputting descriptive language to clarify stylistic differences. In the digital assessment of historical building cultural heritage, existing research has utilized large-scale language models to detect the cultural heritage value in building facade images, achieving classification and assessment of multi-family residential and non-residential buildings. In urban renewal planning decision support, researchers have explored the application of large-scale visual models in diagnosing urban landscape issues, as well as an intelligent urban planning support framework driven by the synergy of large-scale language models and visual large-scale models. These studies provide planners with intelligent assistance throughout the entire process, from textual concepts to visual design schemes.

[0004] However, as the above methods are applied to actual conservation planning scenarios in different historical districts, significant shortcomings still exist: First, the model's cross-regional generalization confidence is insufficient and lacks quantification methods. The landscape assessment ability of visual language models is highly dependent on the geographical distribution of the training data. The model is relatively reliable in areas with sufficient coverage of architectural styles in the training data, but its reliability drops significantly in areas with insufficient coverage of local or niche architectural styles.

[0005] Secondly, there is a lack of adaptation mechanisms for local cultural context. The assessment of the appearance of historical blocks involves a profound local cultural context. The same architectural element may have completely different historical value judgments in different regional cultural backgrounds. Existing visual language models mainly rely on visual input for scene description, which has obvious limitations in understanding cultural heritage that requires professional historical depth.

[0006] In summary, existing methods for assessing the appearance of historical districts based on visual language models have significant shortcomings in cross-regional generalization confidence measurement, adaptation to local cultural context, and verification of consistency with expert evaluations. Therefore, there is an urgent need to propose a spatially intelligent historical district appearance assessment inference method based on a confidence field, which unifies the cross-regional generalization confidence and the conformity to local cultural context into a confidence field for appearance assessment inference. Summary of the Invention

[0007] To address the aforementioned technical problems, this application is proposed. Embodiments of this application provide a spatially intelligent historical district appearance assessment method and system based on a confidence field.

[0008] According to one aspect of this application, a spatial intelligent historical district appearance assessment method based on confidence field is provided, comprising: acquiring an appearance image of the historical district and extracting multi-dimensional features from the appearance image; identifying the decision type to which the current assessment belongs; comprehensively assessing the preliminary assessment result and corresponding comprehensive confidence of the historical district based on the multi-dimensional features, the assessment model corresponding to the decision type, the cultural context of the local appearance feature set, and the previous expert assessment experience of the region, and locating the weak dimensions of the multi-dimensional features based on the comprehensive confidence; if the comprehensive confidence does not meet the target requirements, reviewing the multi-dimensional features based on the weak links corresponding to the weak dimensions to obtain a review result; and determining the final assessment result of the historical district based on the preliminary assessment result, the review result, and the final assessment basis.

[0009] In one embodiment, the extraction of multi-dimensional features of the appearance image includes: extracting material features, color features, and shape features of the appearance image.

[0010] In one embodiment, the spatial intelligent historical district appearance assessment method based on confidence field further includes: based on the multi-dimensional features, associating the local appearance feature library and historical and cultural background data to obtain the local appearance feature set.

[0011] In one embodiment, the step of comprehensively evaluating the preliminary evaluation results and corresponding comprehensive confidence of the historical block based on the multi-dimensional features, the evaluation model corresponding to the decision type, the cultural context of the local landscape feature set, and the previous expert evaluation experience of the region includes: evaluating the coverage degree of the multi-dimensional features based on the sufficiency of the coverage of the multi-dimensional features in the evaluation model corresponding to the decision type; evaluating the cultural context conformity of the multi-dimensional features based on the cultural context recorded in the local landscape feature set; evaluating the multi-dimensional features from multiple dimensions to obtain single-dimensional evaluation values; determining the consistency degree of the multi-dimensional features based on the single-dimensional evaluation values ​​and the previous expert evaluation experience of the region to which the historical block belongs; and combining the coverage degree, the cultural context conformity, the consistency degree, and the single-dimensional evaluation values ​​to obtain the preliminary evaluation results and corresponding comprehensive confidence of the historical block.

[0012] In one embodiment, the step of evaluating the multi-dimensional features from multiple dimensions to obtain a single-dimensional evaluation value includes: evaluating the multi-dimensional features from the perspectives of material, color, form, and historical evolution to obtain a single-dimensional evaluation value.

[0013] In one embodiment, the step of locating the weak dimension of the multi-dimensional features based on the comprehensive confidence includes: if the comprehensive confidence is lower than a preset confidence threshold, then the dimension with the lowest confidence among the coverage, cultural context conformity, consistency, and single-dimensional evaluation value is determined as the weak dimension.

[0014] In one embodiment, the review of the multi-dimensional features based on the weak links corresponding to the weak dimensions includes: diverting features with weak cultural context or weak consistency to local experts for review; and diverting features with weak single-dimensional assessments to historical archives for review.

[0015] In one embodiment, the step of diverting weak features in the single-dimensional assessment to historical archives for review includes: independently comparing the weak features with existing historical building survey archives and historical photographs in the historical archives to obtain comparison results; if the comparison results are consistent, the confidence level of the weak features in the single-dimensional assessment is increased; otherwise, it is marked as needing further verification.

[0016] In one embodiment, the spatial intelligent historical district appearance assessment method based on confidence field further includes: calibrating the parameters of the assessment model for the region based on cases that have been reviewed, overturned, or confirmed by local experts.

[0017] According to another aspect of this application, a spatial intelligent historical district appearance assessment system based on a confidence field is provided, comprising: a district feature extraction module for acquiring appearance images of historical districts and extracting multi-dimensional features from the appearance images; a decision type identification module for identifying the decision type to which the current assessment belongs; a comprehensive confidence assessment module for comprehensively assessing the preliminary assessment results and corresponding comprehensive confidence of the historical district based on the multi-dimensional features, the assessment model corresponding to the decision type, the cultural context of the local appearance feature set, and the previous expert assessment experience of the region, and locating the weak dimensions of the multi-dimensional features based on the comprehensive confidence; a weak link verification module for verifying the multi-dimensional features based on the weak links corresponding to the weak dimensions if the comprehensive confidence does not meet the target requirements, and obtaining a verification result; and an assessment result determination module for determining the final assessment result of the historical district based on the preliminary assessment result, the verification result, and the final assessment basis.

[0018] This application provides a spatial intelligent historical district appearance assessment method and system based on confidence fields. The method acquires exterior images of historical districts and extracts multi-dimensional features from these images. It identifies the decision type of the current assessment. Based on the multi-dimensional features, the assessment model corresponding to the decision type, the cultural context of the local appearance feature set, and the past expert assessment experience in the region, it comprehensively assesses the preliminary assessment results and corresponding comprehensive confidence of the historical district, and identifies the weak dimensions of the multi-dimensional features based on the comprehensive confidence. If the comprehensive confidence does not meet the target requirements, the multi-dimensional features are reviewed based on the weak links corresponding to the weak dimensions, resulting in a review result. Based on the preliminary assessment results, review results, and final assessment basis, the final assessment result of the historical district is determined. The method unifies the model's cross-regional generalization confidence, local cultural context conformity, and consistency with historical expert assessment experience into dynamic confidence to improve the accuracy of the confidence assessment. Furthermore, it reviews weak links, making the review resource allocation more precise. Targeted reviews are triggered only for specific assessment conclusions that truly have cross-regional generalization risks, thereby improving assessment efficiency. Attached Figure Description

[0019] The above and other objects, features, and advantages of this application will become more apparent from the more detailed description of the embodiments of this application in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the embodiments of this application to explain this application and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0020] Figure 1 This is a flowchart illustrating an exemplary embodiment of the spatial intelligent historical district appearance assessment method based on a confidence field provided in this application.

[0021] Figure 2 This is a schematic diagram of the structure of a spatial intelligent historical district appearance assessment system based on a confidence field provided in an exemplary embodiment of this application.

[0022] Figure 3 This is a structural diagram of an electronic device provided in an exemplary embodiment of this application. Detailed Implementation

[0023] Hereinafter, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments of this application. It should be understood that this application is not limited to the exemplary embodiments described herein.

[0024] Figure 1 This is a flowchart illustrating an exemplary embodiment of the spatial intelligent historical district appearance assessment method based on a confidence field provided in this application. Figure 1As shown, the spatial intelligent historical district appearance assessment method based on confidence field includes the following steps: Step 110: Obtain exterior images of the historical blocks and extract multi-dimensional features from the exterior images.

[0025] This application acquires exterior images of historical blocks and extracts multi-dimensional features from these images.

[0026] Step 120: Identify the decision type to which the current assessment belongs.

[0027] This application identifies the decision type of the current assessment based on the assessment task. The assessment task includes the assessment purpose, administrative matter, scope of the target, proposed measures, use of results, reversibility of alterations, legal consequences, and approval level. Decision types include inclusion or adjustment in the directory, demolition or major renovation permits, review of repair plans, planning permits, routine inspections, and landscape surveys. The decision type described in the current assessment is determined through rule mapping between the assessment task and the decision type. Preferably, when the rule mapping conflicts with the assessment task, key fields of the assessment task are missing, or multiple types are simultaneously matched, the candidate type, matching rule number, conflicting fields, and pending confirmation status are output. Target confidence is not generated before confirmation. Importance is divided into four levels: Level 1 corresponds to filing and scientific research statistics; Level 2 corresponds to landscape surveys and general filing; Level 3 corresponds to review of repair plans, planning permits, and other adjustable matters that may change the physical state of buildings; and Level 4 corresponds to inclusion or removal from the directory, adjustment of protection level, and demolition or major renovation permits. Each level sets corresponding comprehensive target confidence, minimum confidence of key dimensions, required evidence types, mandatory verification paths, number of experts, and queue priority; the numerical thresholds are calibrated for reliability by samples from the same region that have final verification conclusions.

[0028] Step 130: Based on the multi-dimensional features, the evaluation model corresponding to the decision type, the cultural context of the local landscape feature set, and the previous expert evaluation experience of the region, comprehensively evaluate the preliminary evaluation results and corresponding comprehensive confidence of the historical block, and locate the weak dimensions of the multi-dimensional features based on the comprehensive confidence.

[0029] Based on multi-dimensional features, evaluation models corresponding to decision types, the cultural context of local landscape feature sets, and the previous expert evaluation experience in the region, this application comprehensively evaluates the preliminary evaluation results and corresponding comprehensive confidence of historical blocks, and identifies the weak dimensions of multi-dimensional features based on the comprehensive confidence.

[0030] Step 140: If the overall confidence does not meet the target requirements, the multi-dimensional features are reviewed based on the weak links corresponding to the weak dimensions to obtain the review results.

[0031] If the overall confidence level meets the target requirements, it is directly adopted; if the overall confidence level does not meet the target requirements, the multi-dimensional features are split and reviewed according to the weak links corresponding to the weak dimensions, and the review results are obtained.

[0032] Step 150: Based on the preliminary assessment results, the review results, and the final assessment criteria, determine the final assessment results for the historical district.

[0033] Based on the preliminary assessment results, the review results, and the final assessment criteria, this application determines the final assessment result of the historical district.

[0034] This application provides a spatial intelligent historical district appearance assessment method based on confidence fields. The method acquires exterior images of historical districts and extracts multi-dimensional features from these images. It identifies the decision type of the current assessment. Based on the multi-dimensional features, the assessment model corresponding to the decision type, the cultural context of the local appearance feature set, and the past expert assessment experience in the region, it comprehensively assesses the preliminary assessment results and corresponding comprehensive confidence of the historical district, and identifies the weak dimensions of the multi-dimensional features based on the comprehensive confidence. If the comprehensive confidence does not meet the target requirements, the multi-dimensional features are reviewed based on the weak links corresponding to the weak dimensions, resulting in a review result. Based on the preliminary assessment results, review results, and final assessment basis, the final assessment result of the historical district is determined. The method unifies the model's cross-regional generalization confidence, local cultural context conformity, and consistency with historical expert assessment experience into dynamic confidence to improve the accuracy of the confidence assessment. Furthermore, it reviews weak links, making the review resource allocation more precise. Targeted reviews are triggered only for specific assessment conclusions that truly have cross-regional generalization risks, thereby improving assessment efficiency.

[0035] In one embodiment, step 110 can be implemented by extracting the material features, color features, and shape features of the appearance image.

[0036] This application identifies the material, color, and structural features of historical blocks within exterior images, using these features as fundamental data for subsequent assessment of the historical block's appearance. Specifically, this application first performs radial distortion correction, perspective correction, white balance correction, and local illumination normalization on the exterior images based on camera intrinsic parameters or facade vanishing points. Then, semantic segmentation and instance segmentation are performed on roofs, walls, doors, windows, columns, foundations, and decorative components to generate component masks. Material features are identified only within the corresponding component masks. Combined with Lab color statistics, local texture, frequency domain texture, and visual coding features, these features are matched against local brick, stone, wood, tile, and plaster samples. Multi-view results are fused based on sharpness, occlusion rate, incident angle, and viewpoint consistency. The form vector (form characteristics) consists of the building height-to-width ratio, number of stories, roof type and slope, ridge and eave curvature, bay and column grid rhythm, door and window height-to-width ratio and spacing, facade solid-to-void ratio, axial symmetry, component adjacency relationship and hierarchical relationship. The geometric quantities are normalized according to facade width, story height or component enclosure frame.

[0037] In one embodiment, the above-mentioned spatial intelligent historical district appearance assessment method based on confidence field may further include: obtaining a local appearance feature set by associating local appearance feature database and historical and cultural background data based on multi-dimensional features.

[0038] This application obtains a local landscape feature set by associating local landscape feature databases and historical and cultural background data based on multi-dimensional features. Specifically, this application can perform feature association based on the features of each dimension in the multi-dimensional features to obtain multiple related features from the local landscape feature database and historical and cultural background data, thereby obtaining a local landscape feature set. The local landscape feature set includes regional codes, time periods, building types, component types, materials, craftsmanship, form, cultural semantics, evidence sources, data levels, and version numbers, and establishes structured indexes, image vector indexes, and text vector indexes. Based on the indexes, hard filtering is performed by coordinates or blocks, time periods, and building types. The entire building image features, component image features, component-material-craftsmanship attributes, and task text are encoded into query vectors to retrieve multiple candidate records (related features). Candidate records must simultaneously satisfy vector similarity thresholds, time compatibility rules, and relationship constraints. Preferably, this application can filter the associated features, for example, by determining whether there is a contradiction between the associated features and the multi-dimensional features of the historical blocks. If there is a contradiction, the corresponding associated features are directly deleted; otherwise, they are retained in the local landscape feature set.

[0039] In one embodiment, step 130 can be implemented as follows: assess the coverage of multi-dimensional features based on their sufficiency in the evaluation model corresponding to the decision type; assess the cultural context conformity of multi-dimensional features based on their cultural context recorded in the local landscape feature set; evaluate the multi-dimensional features from multiple dimensions to obtain single-dimensional evaluation values; determine the consistency of multi-dimensional features based on the single-dimensional evaluation values ​​and the previous expert evaluation experience of the historical district; and obtain the preliminary evaluation results and corresponding comprehensive confidence of the historical district by comprehensively considering the coverage, cultural context conformity, consistency, and single-dimensional evaluation values.

[0040] This application assesses the coverage of multi-dimensional features in the evaluation model corresponding to the decision type based on the sufficiency of multi-dimensional features. For regions with insufficient coverage, the evaluation confidence is reduced accordingly from the initial regional evaluation confidence (the reduction value is inversely correlated with the coverage level). The evaluation model includes a visual or visual language encoder, a task head for architectural style and component attributes, a task head for protection level classification or ranking, and a probability calibrator. The training set or the same encoder reference support set is used to establish vector indexes by region, period, building type, material, and form, and to store class centers, covariance, local sample density, effective sample count, and label distribution. For the building to be evaluated, multiple neighbors are retrieved by dimension, and the average neighbor distance, local density, effective sample count, neighbor label consistency rate, and class-conditional Mahalanobis distance are calculated. The initial regional evaluation confidence is reduced when the local density is lower than a preset density, the Mahalanobis distance exceeds a threshold, the neighbor label distribution density is lower than a set value, or the image quality is lower than a preset quality. The cultural contextual compliance of multi-dimensional features is assessed by comparing their consistency with the cultural context (including region, era, building type, components, materials, craftsmanship, cultural meaning, and relationship diagrams of historical events) recorded in the local landscape feature set. Significant discrepancies result in a lower confidence level (the corresponding decrease is inversely correlated with the compliance level). Multi-dimensional features are evaluated separately from multiple dimensions to obtain single-dimensional evaluation values. Based on the single-dimensional evaluation values ​​and previous expert evaluation experience in the historical district (including region, period, building type, components, materials, landscape category, orderly protection level, evaluation standard version, evidence completeness, and expert consensus), the consistency of multi-dimensional features is determined. (Specifically, based on the single-dimensional feature, corresponding previous expert evaluation results are searched, and the consistency between the expert evaluation results and the model's results is calculated). Specifically, after hard filtering by region-period-type, multiple similar cases are retrieved using image and structural features. The weights of corresponding cases are calculated based on case similarity, consensus, evidence completeness, and standard version compatibility to form a historical expert level distribution. The Jensen-Shannon distance or Wasserstein distance is calculated based on the current model probability distribution and historical distribution. Normalized rank differences are then calculated based on the ordinal rank. These two parameters are mapped to expert historical consistency confidence using locally validated samples. When the number of cases falls below the statistical lower limit, historical evidence is considered missing, and the weight of this factor is reduced. By combining coverage, cultural context fit, consistency, and single-dimensional evaluation values, a preliminary assessment of the historical district and its corresponding overall confidence level are obtained.

[0041] In one embodiment, step 130 can be implemented by evaluating the multi-dimensional features from the perspectives of material, color, form, and historical evolution to obtain a single-dimensional evaluation value.

[0042] This application evaluates the historical district from multiple dimensions, including material, color, form, and historical development, and obtains a single-dimensional evaluation value for each dimension. After obtaining the single-dimensional evaluation value, this application can also calculate an overall evaluation value (e.g., by weighted summation) based on multiple single-dimensional evaluation values, which serves as a comprehensive indicator for the preliminary evaluation results of the historical district.

[0043] In one embodiment, step 130 can be implemented as follows: if the overall confidence level is lower than a preset confidence threshold, the dimension with the lowest confidence level among coverage, cultural context conformity, consistency, and single-dimensional evaluation value is selected as the weak dimension.

[0044] If the overall confidence level is lower than a preset confidence threshold, the dimension with the lowest confidence level among coverage, cultural context fit, consistency, and single-dimensional assessment values ​​is selected as the weak dimension. Preferably, since the threshold levels for each dimension are different, this application can normalize the confidence levels among coverage, cultural context fit, consistency, and single-dimensional assessment values ​​before comparing them to select the actual weak dimension. Specifically, dimensions below the indeterminate lower limit, containing hard conflicts, or with internal evidence disagreements exceeding the threshold are listed as candidate weak dimensions. Sensitivity analysis is performed on the candidate weak dimensions: other dimensions are fixed, and this dimension is replaced with a verified value or removed from the aggregation. The overall confidence change and final level change are calculated to obtain the sensitivity of this dimension. The candidate weak dimension with the highest sensitivity is selected as the final weak dimension.

[0045] In one embodiment, step 140 can be implemented by: diverting features with weak cultural context or weak consistency to local experts for review; and diverting features with weak single-dimensional assessment to historical archives for review.

[0046] This application diverts features with weak cultural context (cultural relationship conflicts) or weak consistency (expert historical deviation) to local experts for review, and diverts features with weak single-dimensional assessment (weak training coverage, poor image quality, or missing archives) to historical archive review, in order to achieve targeted review and thus improve the review effect and efficiency.

[0047] In one embodiment, step 140 can be implemented by independently comparing the weak features in the single-dimensional assessment with existing historical building survey archives and historical photographs in the historical archives to obtain the comparison results; if the comparison results are consistent, the confidence of the weak features in the single-dimensional assessment is increased; otherwise, it is marked that further verification is needed.

[0048] This application independently compares the weak features of a single-dimensional assessment with existing historical building survey archives and historical photographs in historical archives to obtain comparison results. The existing historical building survey archives include building identification, surveying time, measurement unit, facade orientation, number of floors, number of bays, grid, roof slope, door and window dimensions, component location, material records, and source grade. Historical photographs are retrieved by building identification, coordinates, facade orientation, shooting time, and global image features. Local or dense corresponding points are extracted between candidate photographs and the current image, and the homography matrix or camera pose is estimated. Mismatches are eliminated by effective interior point rate, interior point spatial distribution, and reprojection error. The existence, outline position, normalized size, topological relationship, texture, and material of the same component area are compared after processing. Color is only included in color archives that can be color-corrected. Based on numerical deviation, effective interior point rate, reprojection error, component attribute consistency rate, difference mask, and archive quality, three-state conclusions are obtained: consistent, inconsistent, and insufficient evidence. Insufficient evidence is not considered inconsistent. If the comparison results are consistent, the confidence level of the weak feature in that single dimension assessment is increased. The formula for calculating the increased confidence level is as follows: To improve the confidence level, To improve the confidence level before, The calibration coefficient is determined from historical cases with final conclusions. The strength of positive archival evidence is determined by the reliability of the evidence source, the quality of geometric matching, the consistency rate of attributes, and the independence of the data; otherwise, it is marked as requiring further verification and expert review, and the confidence level is updated when the evidence is sufficient. The specific update formula is as follows: Maintain the confidence level when there is insufficient evidence.

[0049] In one embodiment, the aforementioned spatial intelligent historical district appearance assessment method based on confidence field may further include: calibrating the parameters of the assessment model for the region based on cases that have been reviewed, overturned, or confirmed by local experts.

[0050] This application calibrates the parameters of the evaluation model for the region based on cases that have been reviewed, overturned, or confirmed by local experts, thus forming a localized calibration dataset.

[0051] The technical solution of this application will be illustrated below with specific examples: This embodiment uses a historical building style assessment project in a traditional ethnic minority village in Southwest China as the application environment, involving the assessment of the style protection level of approximately 95 traditional buildings. The specific steps are as follows: Image and local landscape feature database integration: Access to exterior images of 95 buildings, extraction of multi-dimensional features such as material, color, and form, and association with the local ethnic minority architectural landscape feature database of the region.

[0052] Importance Analysis: Among them, about 20 buildings are involved in the major decision of whether to be included in the provincial protection list, with a high target confidence level of 0.85. The remaining 75 buildings are subject to general landscape harmony surveys, with a medium target confidence level of 0.60.

[0053] Confidence field construction: The assessment found that the model training data had limited coverage of the traditional architectural styles of ethnic minorities in the region, and the cross-regional generalization confidence of about 38 buildings was significantly low; further assessment of cultural context conformity revealed that the assessment basis of 15 buildings deviated from the traditional craft features recorded in the local landscape feature database; comparison with the historical assessment experience of experts showed that the assessment conclusions of 8 buildings were significantly inconsistent with previous expert assessments.

[0054] Triggering judgment and scheduling: 15 buildings with cultural context deviation and 8 buildings with expert experience deviation (a total of about 20 buildings, with some overlap) were diverted to local experts for review, while the remaining 18 buildings with low cross-regional generalization confidence but acceptable cultural context and expert experience assessments were diverted to independent historical archive verification.

[0055] Targeted review by local experts: Local experts on ethnic minority architectural cultural heritage were organized to review 20 buildings. After the review, the preliminary assessment conclusions of the model were adjusted for 18 of them, mainly involving a reassessment of the historical value of specific traditional decorative patterns.

[0056] Independent historical archive verification: The 18 buildings were compared with existing village survey archives and historical photos. The comparison of 15 buildings was consistent, confirming the reliability of the original assessment dimensions. The 3 buildings were transferred to on-site inspection and verification by experts due to missing archives.

[0057] Confidence archiving and feedback: Archive the processing path of the 38 buildings to be verified in this round, and condense the 18 cases that were overturned by the experts into a localized calibration dataset for this region, so as to reduce the uncertainty in the cross-regional generalized confidence reduction of subsequent assessments of similar ethnic minority traditional buildings.

[0058] Implementation Effects and Technical Functions: Without the technical solution of this application, the style assessment conclusions of 95 buildings would be directly accepted, and misjudgments in 38 buildings with actual cross-regional generalization risks could affect the accuracy of the protection list decision. If expert review were uniformly implemented for all 95 buildings, approximately 60% of the buildings that already meet the standards would be unnecessarily re-reviewed. Without a continuous tracking mechanism for changes in building status, style changes caused by residents' self-renovation might go undetected for a long time. The results show that the core technology of this application lies in using the confidence field of style assessment inference as a consistent variable, triggering targeted verification only for specific assessment conclusions with insufficient confidence, rather than treating the model assessment conclusions as general conclusions that can be directly applied across regions.

[0059] Figure 2 This is a schematic diagram of the structure of a spatially intelligent historical district appearance assessment system based on a confidence field, provided in an exemplary embodiment of this application. Figure 2 As shown, the spatial intelligent historical district appearance assessment system 20 based on confidence field includes: a district feature extraction module 21, used to acquire appearance images of historical districts and extract multi-dimensional features of the appearance images; a decision type identification module 22, used to identify the decision type to which the current assessment belongs; a comprehensive confidence assessment module 23, used to comprehensively assess the preliminary assessment results and corresponding comprehensive confidence of historical districts based on multi-dimensional features, assessment models corresponding to decision types, cultural context of local appearance feature sets, and previous expert assessment experience in the region, and to locate the weak dimensions of multi-dimensional features based on comprehensive confidence; a weak link verification module 24, used to verify the multi-dimensional features based on the weak links corresponding to the weak dimensions if the comprehensive confidence does not meet the target requirements, and to obtain the verification results; and an assessment result determination module 25, used to determine the final assessment result of historical districts based on the preliminary assessment results, verification results, and final assessment basis.

[0060] This application provides a spatial intelligent historical district appearance assessment system based on a confidence field. The system comprises a district feature extraction module 21 that acquires exterior images of the historical district and extracts multi-dimensional features from these images; a decision type identification module 22 that identifies the decision type to which the current assessment belongs; and a comprehensive confidence assessment module 23 that, based on the multi-dimensional features, the assessment model corresponding to the decision type, the cultural context of the local appearance feature set, and the previous expert assessment experience in the region, comprehensively assesses the preliminary assessment results and corresponding comprehensive confidence of the historical district, and identifies the weak dimensions of the multi-dimensional features based on the comprehensive confidence. If the comprehensive confidence does not meet the target requirements, the system will proceed as follows: The weak link verification module 24 verifies the multi-dimensional features based on the weak links corresponding to the weak dimensions, and obtains the verification results. The evaluation result determination module 25 determines the final evaluation result of the historical block based on the preliminary evaluation results, verification results, and final evaluation basis. The model's cross-regional generalization confidence, local cultural context conformity, and consistency with the historical experience of expert evaluation are uniformly coupled into dynamic confidence to provide the accuracy of confidence evaluation. Furthermore, the verification is carried out on weak links, making the verification resource investment more precise. Targeted verification is only triggered for specific evaluation conclusions that truly have cross-regional generalization risks, thereby improving evaluation efficiency.

[0061] In one embodiment, the street feature extraction module 21 can be further configured to extract the material features, color features, and shape features of the appearance image.

[0062] In one embodiment, the aforementioned spatial intelligent historical district appearance assessment system 20 based on confidence field can be further configured to: obtain a local appearance feature set by associating local appearance feature database and historical and cultural background data based on multi-dimensional features.

[0063] In one embodiment, the aforementioned comprehensive confidence assessment module 23 can be further configured to: assess the coverage of multi-dimensional features based on the sufficiency of coverage of multi-dimensional features in the assessment model corresponding to the decision type; assess the cultural context conformity of multi-dimensional features based on the cultural context recorded in the local landscape feature set; assess multi-dimensional features from multiple dimensions to obtain single-dimensional assessment values; determine the consistency of multi-dimensional features based on the single-dimensional assessment values ​​and the previous expert assessment experience of the historical block's region; and obtain the preliminary assessment results of the historical block and the corresponding comprehensive confidence by comprehensively considering the coverage, cultural context conformity, consistency, and single-dimensional assessment values.

[0064] In one embodiment, the aforementioned comprehensive confidence assessment module 23 can be further configured to: assess the multi-dimensional features from the perspectives of material, color, form, and historical evolution, and obtain a single-dimensional assessment value.

[0065] In one embodiment, the comprehensive confidence assessment module 23 can be implemented as follows: if the comprehensive confidence is lower than a preset confidence threshold, the dimension with the lowest confidence among coverage, cultural context conformity, consistency and single-dimensional assessment value is selected as the weak dimension.

[0066] In one embodiment, the aforementioned weak link review module 24 can be further configured to: divert features with weak cultural context or weak consistency to local experts for review; and divert features with weak single-dimensional assessment to historical archives for review.

[0067] In one embodiment, the aforementioned weak link verification module 24 can be further configured to: independently compare the weak features in the single-dimensional assessment with existing historical building survey archives and historical photographs in the historical archives to obtain the comparison results; if the comparison results are consistent, the confidence of the weak features in the single-dimensional assessment is increased; otherwise, it is marked that further verification is needed.

[0068] In one embodiment, the aforementioned confidence field-based spatial intelligent historical district appearance assessment system 20 can be further configured to: calibrate the parameters of the assessment model for the region based on cases that have been reviewed, overturned, or confirmed by local experts.

[0069] Below, for reference Figure 3 To describe an electronic device according to embodiments of this application. For example... Figure 3 As shown, the electronic device 10 includes one or more processors 11 and memory 12.

[0070] The processor 11 may be a central processing unit (CPU) or other form of processing unit with data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device 10 to perform desired functions.

[0071] The memory 12 may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 11 may execute the program instructions to implement the methods of the various embodiments of this application described above and / or other desired functions. Various contents such as input signals, signal components, and noise components may also be stored in the computer-readable storage medium.

[0072] In one example, the electronic device 10 may also include an input device 13 and an output device 14, which are interconnected via a bus system and / or other forms of connection mechanism (not shown).

[0073] When the electronic device is a standalone device, the input device 13 can be a communication network connector for receiving the collected input signals from the first device and the second device.

[0074] In addition, the input device 13 may also include, for example, a keyboard, a mouse, etc.

[0075] The output device 14 can output various information to the outside, including determined distance information, direction information, etc. The output device 14 may include, for example, a display, a speaker, a printer, and a communication network and its connected remote output devices, etc.

[0076] Of course, for the sake of simplicity, Figure 3 Only some of the components of the electronic device 10 relevant to this application are shown in this illustration; components such as buses, input / output interfaces, etc., are omitted. In addition, the electronic device 10 may include any other suitable components depending on the specific application.

[0077] In addition to the methods and apparatus described above, embodiments of this application may also be computer program products, which include computer program instructions that, when executed by a processor, cause the processor to perform the steps in the methods according to various embodiments of this application described in the "Exemplary Methods" section above.

[0078] The computer program product can be written in any combination of one or more programming languages ​​to perform the operations of the embodiments of this application. The programming languages ​​include object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0079] Furthermore, embodiments of this application may also be computer-readable storage media storing computer program instructions thereon, which, when executed by a processor, cause the processor to perform the steps in the methods according to various embodiments of this application described in the "Exemplary Methods" section above.

[0080] The computer-readable storage medium may be any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof.

[0081] The basic principles of this application have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in this application are merely examples and not limitations, and should not be considered as essential features of each embodiment of this application. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the application to the necessity of employing the aforementioned specific details for implementation.

[0082] The block diagrams of devices, apparatuses, devices, and systems involved in this application are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as “comprising,” “including,” “having,” etc., are open-ended terms meaning “including but not limited to,” and are used interchangeably with them. The terms “or” and “and” as used herein refer to the terms “and / or,” and are used interchangeably with them unless the context clearly indicates otherwise. The term “such as” as used herein refers to the phrase “such as but not limited to,” and is used interchangeably with it.

[0083] It should also be noted that in the apparatus, equipment, and methods of this application, the components or steps can be disassembled and / or recombined. These disassemblies and / or recombinations should be considered as equivalent solutions of this application.

[0084] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of this application. Therefore, this application is not intended to be limited to the aspects shown herein, but rather to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0085] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this application to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. A spatially intelligent historical district appearance assessment method based on confidence fields, characterized in that, include: Acquire exterior images of the historical block and extract multi-dimensional features from the exterior images; Identify the decision type to which the current assessment belongs; Based on the multi-dimensional features, the evaluation model corresponding to the decision type, the cultural context of the local landscape feature set, and the previous expert evaluation experience of the region, the preliminary evaluation results and corresponding comprehensive confidence of the historical block are comprehensively evaluated, and the weak dimensions of the multi-dimensional features are located based on the comprehensive confidence. If the overall confidence does not meet the target requirements, the multi-dimensional features are reviewed based on the weak links corresponding to the weak dimensions to obtain the review results; Based on the preliminary assessment results, the review results, and the final assessment criteria, the final assessment results for the historical district are determined.

2. The spatial intelligent historical district appearance assessment method based on confidence field according to claim 1, characterized in that, The extraction of multi-dimensional features from the appearance image includes: Extract the material features, color features, and shape features of the appearance image.

3. The spatial intelligent historical district appearance assessment method based on confidence field according to claim 1, characterized in that, The spatial intelligent historical district appearance assessment method based on confidence fields also includes: Based on the aforementioned multi-dimensional features, the local landscape feature library and historical and cultural background data are associated to obtain the local landscape feature set.

4. The spatial intelligent historical district appearance assessment method based on confidence field according to claim 1, characterized in that, The preliminary assessment results and corresponding comprehensive confidence levels of the historical district, based on the multi-dimensional features, the assessment model corresponding to the decision type, the cultural context of the local landscape feature set, and the previous expert assessment experience of the region, include: The coverage of the multi-dimensional features is evaluated based on the sufficiency of their coverage in the evaluation model corresponding to the decision type. Based on the cultural context recorded by the multi-dimensional features and the local landscape feature set, the cultural context conformity of the multi-dimensional features is evaluated. The multi-dimensional features are evaluated from multiple dimensions to obtain a single-dimensional evaluation value. Based on the single-dimensional evaluation value and the previous expert evaluation experience of the region to which the historical block belongs, the degree of consistency of the multi-dimensional characteristics is determined; By combining the coverage, cultural context fit, consistency, and single-dimensional evaluation value, a preliminary evaluation result and corresponding comprehensive confidence level for the historical district are obtained.

5. The spatial intelligent historical district appearance assessment method based on confidence field according to claim 4, characterized in that, The evaluation of the multi-dimensional features from multiple dimensions to obtain a single-dimensional evaluation value includes: The multi-dimensional characteristics are evaluated from the aspects of material, color, form, and historical evolution to obtain a single-dimensional evaluation value.

6. The spatial intelligent historical district appearance assessment method based on confidence field according to claim 4, characterized in that, The weak dimensions of the multi-dimensional features located based on the comprehensive confidence include: If the overall confidence level is lower than a preset confidence threshold, then the dimension with the lowest confidence level among the coverage, cultural context conformity, consistency, and single-dimensional evaluation value is determined as the weak dimension.

7. The spatial intelligent historical district appearance assessment method based on confidence field according to claim 1, characterized in that, The step of reviewing the multi-dimensional features based on the weak links corresponding to the weak dimensions includes: Features with weak cultural context or low consistency will be assigned to local experts for review. Weak features in single-dimensional assessments will be diverted to historical archives for review.

8. The spatial intelligent historical district appearance assessment method based on confidence field according to claim 7, characterized in that, The process of diverting weak features in single-dimensional assessments to historical archives for review includes: The weak features in the single-dimensional assessment are independently compared with the existing historical building survey archives and historical photographs in the historical archives to obtain the comparison results. If the comparison results are consistent, the confidence level of the weak feature in the single-dimensional assessment is increased; otherwise, it is marked as needing further verification.

9. The spatial intelligent historical district appearance assessment method based on confidence field according to claim 1, characterized in that, The spatial intelligent historical district appearance assessment method based on confidence fields also includes: Based on cases that have been reviewed, refuted, or confirmed by local experts, the parameters of the assessment model for that region are calibrated.

10. A spatially intelligent historical district appearance assessment system based on confidence fields, characterized in that, include: The street feature extraction module is used to acquire the appearance images of historical streets and extract multi-dimensional features from the appearance images; The decision type identification module is used to identify the decision type to which the current evaluation belongs; The comprehensive confidence assessment module is used to comprehensively assess the preliminary assessment results and corresponding comprehensive confidence of the historical block based on the multi-dimensional features, the assessment model corresponding to the decision type, the cultural context of the local landscape feature set, and the previous expert assessment experience of the region, and to locate the weak dimensions of the multi-dimensional features based on the comprehensive confidence. The weak link verification module is used to verify the multi-dimensional features based on the weak links corresponding to the weak dimensions if the comprehensive confidence does not meet the target requirements, and to obtain the verification result. The assessment result determination module is used to determine the final assessment result of the historical block based on the preliminary assessment result, the review result, and the final assessment basis.

Images