Urban renewal project progress management and evaluation system based on unmanned aerial vehicle image

Abstract

The application provides a city renewal project progress management and evaluation system based on unmanned aerial vehicle images, relates to the technical field of project management, and comprises: an acquisition module for regularly collecting sequence images of a city renewal project area by an unmanned aerial vehicle, and pre-processing the collected images to obtain standardized image data; a calculation module for extracting features from the standardized image data, combining gray scale information and gradient information of feature point neighborhoods in the images to generate binary feature descriptors, and obtaining local differential binary feature data with high distinguishability and fast matching advantages; and constructing a hierarchical feature structure of a construction scene based on the local differential binary feature data. The application realizes automatic, accurate and visual management and evaluation of city renewal project progress.

Description

Technical Field

[0001] This invention relates to the field of project management technology, and in particular to an urban renewal project progress management and evaluation system based on UAV imagery. Background Technology

[0002] As urban renewal projects move towards larger scale and more refined management, traditional progress management methods are no longer sufficient to meet the needs of real-time monitoring and precise decision-making. While drone imaging technology is being gradually applied due to its flexibility and efficiency, it still faces the core challenge of data integration and analysis.

[0003] In a large-scale urban village redevelopment project in a major Chinese city, the construction unit used ordinary drones to collect images of the construction area. After manually stitching these images into a two-dimensional orthophoto, the project team monitored the progress of demolition and resettlement, new residential construction, and supporting facilities by visually comparing them with the planned progress data recorded in tables. However, the technical shortcomings were significant: the drone-collected image data lacked automated correlation and integration with the project's planned progress and historical construction data. This made it impossible to quantify the completion rate and time deviation of the work in each construction area. Furthermore, manual image processing was time-consuming and made it difficult to capture dynamic changes such as delays in demolition and deviations in pipeline laying in real time. As a result, management personnel could not accurately pinpoint the root causes of progress problems, leading to delayed decision-making and increased risk of project delays. Summary of the Invention

[0004] The technical problem to be solved by this invention is to provide an urban renewal project progress management and evaluation system based on UAV imagery, so as to realize automated, accurate and visualized management and evaluation of urban renewal project progress.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows:

[0006] The first aspect is an urban renewal project progress management and evaluation system based on drone imagery, including:

[0007] The acquisition module is used to periodically collect sequential images of urban renewal project areas using drones, and to preprocess the collected images to obtain standardized image data.

[0008] The computation module is used to extract features from standardized image data, combine the grayscale information and gradient information of the neighborhood of feature points in the image to generate binary feature descriptors, and obtain local differential binary feature data with the advantages of high discriminativeness and fast matching; and construct a hierarchical feature structure of the construction scene based on the local differential binary feature data.

[0009] The detection module is used to set the baseline analysis range in the hierarchical feature structure and divide the baseline analysis range into multiple levels to obtain multiple interrelated feature analysis regions; the extracted local differential binary feature data is mapped to the corresponding feature analysis regions, and the engineering entities and state changes represented by the features of each region are identified by fast matching of the local differential binary features of images in adjacent time periods, so as to obtain the change detection results containing hierarchical structure information.

[0010] The comparison module is used to compare and analyze the actual progress with the planned progress based on the change detection results and the project schedule data, and to generate a schedule deviation report.

[0011] The assessment module is used to evaluate the overall progress trend and risk level of the project based on schedule deviation reports and historical schedule data, in order to obtain the assessment results;

[0012] The processing module integrates the evaluation results with the schedule deviation report to obtain a visual schedule management interface, which supports project managers in making schedule interventions and decision adjustments.

[0013] Furthermore, drones are used to periodically collect sequential images of the urban renewal project area, and the collected images are preprocessed to obtain standardized image data, including:

[0014] Receive the raw image sequence uploaded by the drone at regular intervals, and perform dehazing and illumination normalization processing on each image in turn to obtain the dehazed image.

[0015] By performing camera distortion correction and geometric correction on the dehazed image, the corrected image is obtained;

[0016] The corrected image is then subjected to feature point matching and image stitching to obtain a complete orthophoto of the project area.

[0017] Standardized image data is obtained by resampling the resolution of the complete orthophoto and normalizing the coordinate system.

[0018] Furthermore, by extracting features from standardized image data and combining the grayscale and gradient information of the neighborhood of feature points in the image to generate binary feature descriptors, local differential binary feature data with both high discriminative power and fast matching advantages is obtained. Based on the local differential binary feature data, a hierarchical feature structure for the construction scene is constructed, including:

[0019] Based on standardized image data, multiple feature points are extracted using a preset feature point detection algorithm.

[0020] By sampling each feature point in its neighborhood according to a preset sampling mode, simultaneously calculating the gray-level difference and gradient direction difference, and encoding the comparison results into a binary string, a binary feature descriptor corresponding to each feature point is generated, resulting in a local difference binary feature dataset.

[0021] By using the local differential binary feature dataset, feature points with spatial proximity and descriptor similarity are clustered to form primary feature clusters for characterizing specific construction areas.

[0022] Based on the primary feature clusters, and according to the preset semantic segmentation rules of construction scenarios, they are combined according to spatial location and functional association to form upper-level feature groups for different construction stage partitions.

[0023] Based on primary feature clusters and upper-level feature groups, a hierarchical feature structure for the construction scenario is constructed by organizing them according to the inclusion and contained relationship.

[0024] Furthermore, for each feature point, the gray-level difference and gradient direction difference are calculated simultaneously within the pixel neighborhood of that feature point according to a preset sampling mode. The comparison results are then encoded into binary strings to generate binary feature descriptors for each feature point, resulting in a local difference binary feature dataset, including:

[0025] Based on the extracted multiple feature points, a pixel neighborhood region of a preset size is determined for each feature point;

[0026] Within the pixel neighborhood, multiple pairs of sampling points are selected according to a preset sampling mode, and each pair of sampling points includes a first sampling point and a second sampling point;

[0027] Based on the selected sampling point pairs, calculate the gray-level difference and gradient direction difference between each pair of sampling points;

[0028] The calculated grayscale difference value is compared with the preset grayscale threshold, and the gradient direction difference value is compared with the preset gradient threshold to obtain the comparison result.

[0029] The comparison results are converted into binary bits according to a preset encoding rule, and all binary bits are combined to form a binary string describing the features.

[0030] Each feature point is assigned a binary string describing its features as a binary feature descriptor. The binary feature descriptors of all feature points are then aggregated to obtain the local difference binary feature dataset.

[0031] Furthermore, a baseline analysis range is defined within the hierarchical feature structure, and this range is divided into multiple levels to obtain several interrelated feature analysis regions. The extracted local differential binary feature data is mapped to the corresponding feature analysis regions. Through rapid matching of local differential binary features from images of adjacent time periods, the engineering entities and state changes represented by the features of each region are identified, resulting in change detection results containing hierarchical structural information, including:

[0032] Based on the constructed hierarchical feature structure, the benchmark analysis range is set within the hierarchical feature structure according to the project planning area;

[0033] By dividing the benchmark analysis scope into multiple levels according to spatial scale and construction stage, we obtain the first-level analysis area, the second-level analysis area, and the third-level analysis area.

[0034] The local differential binary feature data is mapped to the corresponding third-level analysis region according to the spatial coordinates, and then aggregated upwards to form a correspondence with the multi-level analysis region;

[0035] Based on the correspondence, local differential binary feature data of the same spatial location in images of adjacent time periods are quickly matched to identify the spatial transformation relationship between feature point pairs.

[0036] Based on the spatial transformation relationship, the feature similarity of each third-level analysis region is calculated. When the similarity is lower than the preset threshold, it is determined that a state change has occurred. The engineering entity type and state change type are identified by combining the construction knowledge base to obtain the state change result.

[0037] The state change results are aggregated upwards according to the hierarchical relationship to obtain the hierarchical structure change detection results.

[0038] Furthermore, based on the change detection results and combined with the project schedule data, a comparative analysis of the actual progress and the planned progress is conducted to obtain a schedule deviation report, including:

[0039] Based on the hierarchical structure change detection results, extract the change type, change area and change time node of each hierarchical analysis area;

[0040] Retrieve the planned progress data for the corresponding time node from the project management database. The planned progress data includes the planned completion time, planned workload, and planned construction content for each construction area.

[0041] The extracted change time points are compared with the retrieved planned completion time to calculate the time deviation value of each level of analysis area; at the same time, the actual change area is quantitatively compared with the planned engineering quantity to calculate the engineering quantity completion rate.

[0042] Based on the time deviation value and the completion rate of the project, combined with the preset deviation threshold standard, the progress status of each construction area is determined, including ahead of schedule, normal, behind schedule, and severely behind schedule.

[0043] By integrating the determined progress status, time deviation value, and work completion rate, a progress deviation report is generated, which includes progress deviation analysis, problem area location, and preliminary judgment of the cause of deviation.

[0044] Furthermore, based on schedule deviation reports and historical schedule data, the overall project progress trend and risk level are assessed to obtain assessment results, including:

[0045] The schedule deviation report is used to extract the schedule deviation value, work completion rate and schedule status information for each analysis area.

[0046] Retrieve historical progress data for similar construction areas from the project history database, including historical completion time, historical deviation trends, and historical risk management records;

[0047] By comparing and analyzing the progress status information with historical progress data over time, the development trend and changing pattern of the current progress deviation can be identified.

[0048] Based on the changing patterns and combined with the preset progress prediction rules, the overall project completion time deviation and the probability of achieving key milestones are calculated to obtain the overall project progress trend assessment results.

[0049] Based on the progress trend assessment results, combined with historical risk management effectiveness records, and in accordance with the risk level quantification standards, the current risk index of the project is calculated and the risk level is determined.

[0050] By integrating the overall project progress trend assessment results with the determined risk level, a comprehensive assessment result is obtained, which includes completion time prediction, risk warning level, and targeted handling suggestions.

[0051] Furthermore, the evaluation results are integrated with the schedule deviation report to obtain a visual schedule management interface, supporting project managers in schedule intervention and decision-making adjustments, including:

[0052] Based on the comprehensive assessment results and schedule deviation reports, the completion time forecast, risk warning level, handling suggestions, and progress status information for each region are extracted.

[0053] Based on the obtained status information and combined with the hierarchical feature structure of the corresponding spatial coordinates, a progress status dataset containing time and spatial dimensions is constructed; based on the progress status dataset, multi-dimensional visualization components are generated.

[0054] The system integrates multi-dimensional visualization components into a unified schedule management interface. Based on this interface, it provides a schedule intervention toolset, receives adjustment instructions from project managers, generates schedule adjustment plans, and feeds them back to the project plan database.

[0055] In a second aspect, a computing device includes:

[0056] One or more processors;

[0057] A storage device for storing one or more programs that, when executed by one or more processors, cause the one or more processors to execute the system.

[0058] Thirdly, a computer-readable storage medium storing a program that, when executed by a processor, performs the system.

[0059] The above-described solution of the present invention has at least the following beneficial effects:

[0060] Because it employs drones to periodically collect sequential images and performs standardized preprocessing such as defogging, correction, stitching, and coordinate unification, it generates local differential binary feature data by combining the grayscale and gradient information of feature points' neighborhoods and constructs a hierarchical feature structure for the construction scene. Through multi-level regional division and feature mapping matching, it achieves hierarchical change detection. It also links project plans and historical data to complete progress deviation analysis, progress trend prediction, and risk assessment. Furthermore, it integrates the results to generate a multi-dimensional visual progress management interface and provides decision-making intervention tools. Therefore, it overcomes the technical problems of insufficient image data accuracy, imbalance between feature distinguishability and matching efficiency, lack of hierarchical logic in change identification, difficulty in quantifying progress deviations, delayed risk warnings, and weak decision support in traditional urban renewal project progress management. As a result, it achieves the technical effects of standardized image data in urban renewal areas, accurate perception of changes in the status of engineering entities, clear quantification of progress deviations and overall trends, scientific determination of risk levels, and intuitive, efficient, and dynamically intervention-enabled management decisions. Attached Figure Description

[0061] Figure 1 This is a schematic diagram of an urban renewal project progress management and evaluation system based on UAV imagery provided by an embodiment of the present invention.

[0062] Figure 2 This is a flowchart illustrating the process of a progress management and evaluation system for urban renewal projects based on UAV imagery, provided by an embodiment of the present invention. The system compares and analyzes the actual progress against the planned progress based on change detection results and combined with the project's planned progress data to obtain a progress deviation report. Detailed Implementation

[0063] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0064] like Figure 1 As shown, embodiments of the present invention propose an urban renewal project progress management and evaluation system based on UAV imagery, comprising:

[0065] The acquisition module is used to periodically collect sequential images of urban renewal project areas using drones, and to preprocess the collected images to obtain standardized image data.

[0066] The computation module is used to extract features from standardized image data, combine the grayscale information and gradient information of the neighborhood of feature points in the image to generate binary feature descriptors, and obtain local differential binary feature data with the advantages of high discriminativeness and fast matching; and construct a hierarchical feature structure of the construction scene based on the local differential binary feature data.

[0067] The detection module is used to set the baseline analysis range in the hierarchical feature structure and divide the baseline analysis range into multiple levels to obtain multiple interrelated feature analysis regions; the extracted local differential binary feature data is mapped to the corresponding feature analysis regions, and the engineering entities and state changes represented by the features of each region are identified by fast matching of the local differential binary features of images in adjacent time periods, so as to obtain the change detection results containing hierarchical structure information.

[0068] The comparison module is used to compare and analyze the actual progress with the planned progress based on the change detection results and the project schedule data, and to generate a schedule deviation report.

[0069] The assessment module is used to evaluate the overall progress trend and risk level of the project based on schedule deviation reports and historical schedule data, in order to obtain the assessment results;

[0070] The processing module integrates the evaluation results with the schedule deviation report to obtain a visual schedule management interface, which supports project managers in making schedule interventions and decision adjustments.

[0071] In this embodiment of the invention, standardized image data is obtained by periodically collecting sequential images of urban renewal project areas using drones and preprocessing them. Local differential binary feature data is generated by extracting grayscale and gradient information of the neighborhood of feature points in the standardized images, and a hierarchical feature structure of the construction scene is constructed. In this structure, multi-level feature analysis areas are divided and feature data is mapped. By quickly matching the image features of adjacent time periods, engineering entities and status changes are identified. The actual and planned progress are compared and analyzed in conjunction with the project's planned progress data. The project's progress trend and risk level are evaluated based on progress deviation reports and historical progress data. Furthermore, the relevant results are integrated to generate a visual progress management interface to support decision-making intervention. Therefore, this invention overcomes the technical problems of non-standard image data, poor feature differentiation and matching effects, lack of systematic identification of status changes, inaccurate progress comparison, insufficient risk prediction, and weak decision support in traditional urban renewal project progress management. As a result, it achieves the technical effects of standardized image data for urban renewal projects, accurate identification of changes in the status of engineering entities, clear quantification of progress deviations, scientific assessment of project progress trends and risk levels, and intuitive, efficient, and timely management decisions.

[0072] In a preferred embodiment of the present invention, a sequence of images of the urban renewal project area is periodically collected by a drone, and the collected images are preprocessed to obtain standardized image data, including:

[0073] The system receives raw image sequences uploaded by drones at set intervals. Based on these sequences, each image undergoes dehazing and illumination normalization processing to obtain dehazed images. Specifically, this involves: receiving raw image sequences of urban renewal project areas uploaded by drones according to a preset acquisition cycle in real time; first, detecting the fog concentration distribution of each image using a dark channel prior algorithm to separate the fog layer from the clear image subject; then, removing residual fog effects from the images using adaptive threshold adjustment; subsequently, calculating the overall average brightness and pixel brightness variance of each image; and performing a linear transformation on the brightness value of each pixel to ensure that the brightness of images acquired under different lighting conditions at different times is within a uniform and reasonable range, completely eliminating the brightness imbalance caused by direct sunlight and shadow occlusion, ultimately resulting in dehazed images with clear image quality and consistent brightness.

[0074] By performing camera distortion correction and geometric correction on the dehazed image, a corrected image is obtained. Specifically, this includes: performing reverse mapping calculation on each pixel in the dehazed image based on Zhang Zhengyou calibration method to correct radial and tangential distortion caused by the camera lens and eliminate the problem of image edge graphic distortion; then obtaining the real geographic coordinates of multiple preset ground control points in the project area; using these control points as a reference, using a quadratic polynomial correction method to establish the mapping relationship between image pixel coordinates and real ground coordinates; adjusting the geometric deformation of the image to correct the geometric offset caused by terrain undulation and shooting angle deviation; and finally obtaining a corrected image that conforms to the real spatial position of the ground.

[0075] The corrected images are then subjected to feature point matching and image stitching to obtain a complete orthophoto of the project area. Specifically, this involves: extracting key feature points from each corrected image, calculating the scale-invariant feature descriptor for each feature point, calculating the similarity of feature point descriptors between adjacent images using the Euclidean distance formula, selecting feature point pairs with high matching degrees, and then using a random sampling consensus algorithm to remove mismatched point pairs to ensure the accuracy of feature matching. Based on the precisely matched feature point pairs, the homography matrix between adjacent images is solved to clarify the spatial relationship between images. Then, all corrected images are stitched together one by one in a left-to-right and top-to-bottom order. A multi-resolution fusion algorithm is used to smooth the pixel brightness transition at the image stitching points to eliminate stitching seams and brightness differences, ultimately forming an orthophoto that completely covers the urban renewal project area.

[0076] Standardized image data is obtained by resampling the complete orthophoto at a higher resolution and unifying the coordinate system. Specifically, this involves: setting a unified target ground sampling distance based on the accuracy requirements of urban renewal project progress analysis; resampling the complete orthophoto at a higher resolution to adjust the pixel density of the image and ensure that the actual ground area corresponding to each pixel in the image is consistent, thus avoiding the impact of inconsistent resolution on the accuracy of subsequent feature extraction and change detection; then converting the original coordinate system of the image to the National Geodetic Coordinate System 2000; and establishing a precise correspondence between image pixel coordinates and geographic coordinates to directly associate and match the standardized image data with the spatial coordinate information in the project schedule data, ultimately obtaining standardized image data with unified resolution and standardized coordinates.

[0077] In this embodiment of the invention, because drones are used to periodically collect sequential images of urban renewal project areas, and the original images are sequentially processed with defogging, illumination normalization, camera distortion correction, and geometric correction, and then image stitching is completed through feature point matching to obtain a complete orthophoto image, and finally resolution resampling and coordinate system unification are used, the technical problems of poor quality caused by fog and illumination differences, camera distortion and geometric deviation, image fragmentation, inconsistent resolution, and chaotic coordinate system, which make the original images unsuitable for subsequent accurate analysis are overcome. Thus, high-quality, standardized, complete and coordinate-unified orthophoto image data is obtained.

[0078] In a preferred embodiment of the present invention, feature extraction is performed on standardized image data, and a binary feature descriptor is generated by combining the grayscale information and gradient information of the neighborhood of feature points in the image, thereby obtaining local differential binary feature data with both high discriminative power and fast matching advantages; a hierarchical feature structure for the construction scene is constructed based on the local differential binary feature data, including:

[0079] Based on standardized image data, multiple feature points are extracted using a pre-defined feature point detection algorithm. Specifically, this involves: first, acquiring pre-processed standardized image data, which has undergone resolution resampling and coordinate system unification, with a unified resolution of 0.1 meters and using the WGS84 coordinate system. The image content covers the entire construction area of ​​the urban renewal project, including demolition areas, new residential construction areas, and supporting facility construction areas. The ORB feature point detection algorithm is selected as the pre-defined algorithm, as it is suitable for real-time processing of UAV imagery and has good noise resistance. During processing, the standardized image is first converted to grayscale to eliminate color interference. Then, the FAST corner detection function of the ORB algorithm is used to identify areas in the grayscale image where pixel grayscale values ​​change abruptly. These areas correspond to key features in the construction scene, such as building outlines, construction machinery corners, and site boundaries. Finally, the detected corner points are filtered to remove false feature points caused by image noise, ensuring that 2000 to 5000 effective feature points are extracted per square kilometer of construction area, and that the feature points are evenly distributed across each construction zone, avoiding excessively dense or sparse feature points in local areas.

[0080] By sampling each feature point within its pixel neighborhood according to a preset sampling mode, simultaneously calculating the gray-level difference and gradient direction difference, and encoding the comparison results into a binary string, a binary feature descriptor corresponding to each feature point is generated, resulting in a local difference binary feature dataset. Specifically, for each extracted feature point, a 32×32 pixel neighborhood region is first determined. This size covers the key texture information around the feature point while avoiding the introduction of too many irrelevant pixels. Within the pixel neighborhood, a 16×16 grid sampling mode is used, dividing the 32×32 neighborhood into 256 16×16 sub-grids. Four pairs of sampling points are selected within each sub-grid, each pair containing a first sampling point and a second sampling point. The sampling points are evenly distributed within the sub-grid to ensure coverage of different positions within the sub-grid. For each pair of sampling points, the gray-level difference and gradient direction difference are calculated: the gray-level difference is the gray value of the first sampling point minus... The grayscale value of the second sampling point is obtained; the gradient direction difference value is calculated by taking the difference between the gradient directions of the two sampling points, and the preset grayscale threshold is set to 15 and the preset gradient threshold is 30 degrees; the calculated grayscale difference value is compared with 15. If the grayscale difference value is greater than 15, the binary bit is recorded as 1, and if it is less than or equal to 15, the binary bit is recorded as 0; the gradient direction difference value is compared with 30 degrees. If the gradient direction difference value is greater than 30 degrees, the binary bit is recorded as 1, and if it is less than or equal to 30 degrees, the binary bit is recorded as 0. According to the arrangement order of the sub-grids, the 8 binary bits corresponding to the 4 pairs of sampling points in each sub-grid are arranged in sequence. A total of 2048-bit feature description binary strings are generated from 256 sub-grids. A corresponding 2048-bit binary string is assigned to each feature point as the binary feature descriptor of that feature point. The binary feature descriptors of all feature points are summarized to form a local differential binary feature dataset covering the entire construction area.

[0081] Using a local differential binary feature dataset, feature points with spatial proximity and descriptor similarity are clustered to form primary feature clusters for characterizing specific construction areas. Specifically, this involves: first, establishing a spatial location-descriptor association table for feature points based on the generated local differential binary feature dataset, recording the pixel coordinates of each feature point on the image and its corresponding binary feature descriptor; second, determining the spatial proximity of feature points by calculating the pixel distance between any two feature points on the image. If the pixel distance is less than 50 pixels, the two feature points are considered to be spatially close. This distance threshold is set based on the typical small construction area in an urban renewal project, such as the image pixel range corresponding to the area occupied by a single building. Feature point descriptor similarity: For feature points with spatial proximity, calculate the Hamming distance of their binary descriptors. If the Hamming distance is less than 30, the descriptors of the two feature points are considered similar. This threshold can effectively distinguish the feature differences of different construction sites. Feature points that simultaneously satisfy spatial proximity and descriptor similarity are clustered into one class to form a primary feature cluster. Each primary feature cluster corresponds to a specific small-scale construction area in the urban renewal project, such as the area of ​​a house to be demolished or the basic construction area of ​​a new residential unit. The number of feature points contained in each primary feature cluster is controlled between 50 and 200 to ensure that the cluster size can accurately represent the local construction area while avoiding feature confusion due to excessively large clusters.

[0082] Based on primary feature clusters and according to pre-defined semantic segmentation rules for construction scenarios, upper-level feature groups are formed for different construction stage zones by combining them according to spatial location and functional association. Specifically, this includes: firstly, formulating pre-defined semantic segmentation rules for construction scenarios, and combining them with the construction process of urban renewal projects, dividing the semantics of construction scenarios into four categories: demolition stage, foundation construction stage, main construction stage, and supporting facility construction stage. Each category corresponds to specific construction content and land features. For example, the demolition stage corresponds to building demolition remnants and site leveling areas, while the foundation construction stage corresponds to foundation pit excavation and foundation pouring areas. The spatial location association of primary feature clusters is determined, the geographic coordinate range of each primary feature cluster is obtained, and the geographic coordinate distance between different primary feature clusters is calculated. If the distance is less than 100 meters, these primary feature clusters are considered to have spatial location association. This distance corresponds to a medium-sized construction zone in urban renewal, such as a demolition cluster and a group of newly built residential buildings. The geographical scope of the cluster determines the functional association of primary feature clusters. Based on the construction area type corresponding to the primary feature clusters, if the construction areas of multiple primary feature clusters have a connection in the construction process, such as the site leveling area after demolition and the adjacent foundation excavation area, or have the same construction function, such as the foundation pouring area of ​​multiple residential units, then these primary feature clusters are determined to have a functional association. According to the principle of semantic + spatial location association + functional association of the same construction scenario, the primary feature clusters that meet the conditions are combined to form upper-level feature groups. For example, 3 to 10 primary feature clusters that are spatially adjacent in the demolition stage and whose function is the same as the site clearing after demolition are combined to form the upper-level feature group of the demolition area; 3 to 10 primary feature clusters that are spatially close in the foundation construction stage and whose function is the same as the residential foundation construction are combined to form the upper-level feature group of residential foundation construction. Each upper-level feature group corresponds to a medium-sized construction zone in the urban renewal project.

[0083] Based on primary feature clusters and upper-level feature groups, a hierarchical feature structure for the construction scene is constructed according to the inclusion-containing relationship. Specifically, this includes: first, clarifying the three-level hierarchical relationship of the hierarchical feature structure: the bottom layer consists of extracted individual feature points, the middle layer consists of formed primary feature clusters, and the top layer consists of formed upper-level feature groups. The inclusion-containing relationship between levels is determined as follows: For each primary feature cluster, all individual feature points within that cluster are categorized under that primary feature cluster, forming the first-level inclusion relationship of upper-level feature groups, primary feature clusters, and individual feature points; that is, each primary feature cluster contains several spatially adjacent and descriptor-similar individual feature points. For each upper-level feature group, all primary feature clusters within that group are categorized under that upper-level feature group, forming the second-level inclusion relationship; that is, each... The upper-level feature group contains several primary feature clusters that are spatially and functionally related. Combining the overall construction scope of the urban renewal project, all upper-level feature groups, primary feature clusters, and individual feature points are integrated to form a complete hierarchical feature structure. For example, in the hierarchical feature structure of a certain urban village renewal project, the top layer includes the upper-level feature group of the demolition area, the upper-level feature group of the residential foundation construction, the upper-level feature group of the residential main structure construction, and the upper-level feature group of the community supporting construction. Among them, the upper-level feature group of the demolition area contains 5 primary feature clusters, and each primary feature cluster contains 80 to 150 individual feature points. These feature points are distributed in key locations such as the outline of the corresponding demolished buildings, doors and windows, and wall remnants, ultimately forming a hierarchical feature structure of the construction scene that covers the entire construction area of ​​the project and has clear levels.

[0084] In this embodiment of the invention, the technical means of extracting feature points based on standardized image data using a preset algorithm, simultaneously calculating gray-level differences and gradient direction differences in the neighborhood of feature point pixels according to a preset sampling mode and encoding them into binary strings to generate a local difference binary feature dataset, then clustering feature points with spatial proximity and descriptor similarity to form primary feature clusters, and combining them with construction scene semantic segmentation rules according to spatial location and functional association to form upper-level feature groups, and finally constructing a hierarchical feature structure of the construction scene according to the inclusion and contained relationship, overcomes the technical problems of traditional feature extraction, such as difficulty in balancing descriptor discriminability and matching efficiency, lack of systematic organization of feature data, and inability to achieve accurate hierarchical analysis in accordance with the construction scene. Thus, it achieves the acquisition of feature data with both high discriminability and fast matching advantages, and constructs a hierarchical feature structure that fits the logic of the construction scene.

[0085] In a preferred embodiment of the present invention, by simultaneously calculating the gray-level difference and gradient direction difference within the pixel neighborhood of each feature point according to a preset sampling mode, and encoding the comparison results into a binary string, a binary feature descriptor corresponding to each feature point is generated, resulting in a local difference binary feature dataset, including:

[0086] Based on the extracted feature points, a pixel neighborhood region of a preset size is determined for each feature point. Specifically, for each extracted feature point, a pixel neighborhood region of 32×32 pixels is determined in combination with the features of the urban renewal construction scene. The size has been tested in actual construction images and can not only fully cover key information such as the building outline texture and construction machinery corner details around the feature point, but also effectively avoid including too many irrelevant pixels far away from the feature point.

[0087] Within the pixel neighborhood, multiple pairs of sampling points are selected according to a preset sampling mode. Each pair of sampling points includes a first sampling point and a second sampling point. Specifically, within the defined 32×32 pixel neighborhood, a 16×16 grid sampling mode is used to evenly divide the entire neighborhood into 256 16×16 sub-grids. Within each sub-grid, four pairs of sampling points are selected according to the principle of uniform distribution. Each pair of sampling points includes a first sampling point and a second sampling point. The sampling points are located at the four corners and near the center of the sub-grid, respectively, to ensure that each pair of sampling points can cover the texture information of different areas within the sub-grid and avoid the omission of feature information due to the concentration of sampling points.

[0088] Based on the selected sampling point pairs, the gray-level difference value and gradient direction difference value between each pair of sampling points are calculated. Specifically, for each pair of sampling points in the sub-grid, the gray-level values ​​of the first and second sampling points are read first, and the gray-level value of the second sampling point is subtracted from the gray-level value of the first sampling point to obtain the gray-level difference value of the pair of sampling points. Then, the gradient direction of the two sampling points is calculated using the Sobel operator. The Sobel operator can effectively capture the edge features of ground objects in construction images and is suitable for the contour detection needs of buildings, construction facilities, etc. After that, the difference between the gradient directions of the two sampling points is taken to obtain the gradient direction difference value of the sampling points. The Sobel operator is an image edge detection tool based on discrete differential operators. Its core function is to capture the abrupt change regions of pixel gray-level values ​​in the image, and then identify the edge features of ground objects. Through specific calculation logic, the horizontal and vertical directions of the image are scanned and analyzed to detect the intensity of pixel gray-level changes in these two directions. Then, the gradient intensity and gradient direction of each pixel are accurately calculated by combining the detection results of the two directions.

[0089] The calculated grayscale difference value is compared with a preset grayscale threshold, and the gradient direction difference value is compared with a preset gradient threshold to obtain the comparison results. Specifically, the preset grayscale threshold is set to 15 and the preset gradient threshold is set to 30 degrees. The calculated grayscale difference value for each sampling point is compared with 15 to determine the significance of the grayscale difference. At the same time, the gradient direction difference value is compared with 30 degrees to determine the significance of the gradient direction change. Two sets of comparison results are obtained for each pair of sampling points in terms of grayscale and gradient direction.

[0090] The comparison results are converted into binary bits according to a preset encoding rule. All binary bits are then combined to form a feature description binary string. Specifically, the conversion is performed according to the preset encoding rule: if the gray-level difference value is greater than 15, the comparison result is converted into binary bit 1; if it is less than or equal to 15, it is converted into binary bit 0. If the gradient direction difference value is greater than 30 degrees, it is converted into binary bit 1; if it is less than or equal to 30 degrees, it is converted into binary bit 0. According to the arrangement order of the sub-grids in the 32x32 neighborhood from left to right and from top to bottom, the 8 binary bits corresponding to the 4 pairs of sampling points in each sub-grid are arranged continuously. A total of 2048 bits of feature description binary string are generated from 256 sub-grids, ensuring that the binary string can completely represent the gray-level and gradient features of the feature points.

[0091] Each feature point is assigned a binary string describing its features as a binary feature descriptor. The binary feature descriptors of all feature points are then aggregated to obtain a local differential binary feature dataset. Specifically, this involves assigning a separate 2048-bit binary string describing each feature point as its own binary feature descriptor, ensuring that the feature information of each feature point can be uniquely and accurately represented. Subsequently, the binary feature descriptors of all feature points within the construction area of ​​the urban renewal project are systematically aggregated to form a local differential binary feature dataset covering all construction areas, including demolition areas, new residential construction areas, and supporting facility construction areas.

[0092] In this embodiment of the invention, because a preset-sized pixel neighborhood region is determined for each feature point, multiple pairs of sampling points are selected within this region according to a preset sampling mode, and the gray-level difference value and gradient direction difference value of each pair of sampling points are calculated respectively, the two types of difference results are compared with the corresponding preset thresholds and converted into binary bits according to rules and combined to form a feature description binary string, and then the binary string is assigned to each feature point as a descriptor and summarized to form a local difference binary feature dataset, the technical problem of traditional feature descriptors relying only on single gray-level or gradient information, resulting in incomplete representation, insufficient distinguishability, inconsistent encoding methods, low matching efficiency, and difficulty in accurately capturing subtle feature differences in construction scene images is overcome. Thus, the generated binary feature descriptor has the advantages of high distinguishability and fast matching, and the feature dataset is standardized and unified, which can comprehensively and accurately represent image feature information.

[0093] In a preferred embodiment of the present invention, a baseline analysis range is set in the hierarchical feature structure, and the baseline analysis range is divided into multiple levels to obtain multiple interrelated feature analysis regions; the extracted local differential binary feature data is mapped to the corresponding feature analysis regions, and the engineering entities and state changes represented by the features of each region are identified by rapid matching of the local differential binary features of images in adjacent time periods, thereby obtaining change detection results containing hierarchical structure information, including:

[0094] Based on the constructed hierarchical feature structure, a benchmark analysis scope is set within the hierarchical feature structure according to the project planning area. Specifically, this includes: first, clarifying the construction area information covered by the constructed hierarchical feature structure, which contains the characteristic associations and spatial distribution of various construction areas in the urban renewal project; then, retrieving the clearly defined scope from the project planning documents, including the geographical coordinate boundaries of all planned construction areas such as resettlement areas, new residential construction areas, and supporting facility construction areas; and finally, delineating the corresponding benchmark analysis scope within the hierarchical feature structure based on the geographical coordinate boundaries, ensuring that this scope fully covers all construction areas in the project plan, neither exceeding the planning boundaries to include irrelevant areas nor omitting any planned construction zones, thus providing an accurate scope basis for subsequent hierarchical analysis.

[0095] By dividing the benchmark analysis scope into multiple levels according to spatial scale and construction stage, we obtain primary, secondary, and tertiary analysis areas. Specifically, the benchmark analysis scope is divided into three levels according to the dual logic of spatial scale from large to small and construction stage progression. The primary analysis area takes the entire urban renewal project as a unit, corresponding to the overall construction stage of the project, covering all planned construction content, and is the highest-level analysis unit. The secondary analysis area is divided according to spatial concentration and construction function relevance, breaking down the primary analysis area into multiple zone units. Each zone corresponds to a major construction stage, such as demolition zone, new residential zone, and supporting facilities zone. Each zone contains several functionally related small-scale construction areas. The tertiary analysis area is the most refined analysis unit. Within each secondary analysis area, it is divided according to a single construction object or specific construction task, such as a single building demolition area in a demolition zone, a single residential unit construction area in a new residential zone, and a section of pipeline laying area in a supporting facilities zone. Finally, we form a series of interconnected and clearly hierarchical characteristic analysis areas at the primary, secondary, and tertiary levels.

[0096] The local differential binary feature data is mapped to the corresponding tertiary analysis region based on spatial coordinates, and then aggregated upwards to form a correspondence with the multi-level analysis regions. Specifically, this involves: extracting the geospatial coordinates corresponding to each feature point in the local differential binary feature data; accurately mapping each feature point to the corresponding tertiary analysis region based on these coordinates, ensuring that each tertiary analysis region has its own set of feature data; after completing the mapping of the tertiary regions, the feature data in each tertiary analysis region is aggregated upwards to its corresponding secondary analysis region, and then the feature data in each secondary analysis region is further aggregated to the corresponding primary analysis region, forming a correspondence between the feature data of the tertiary analysis region, secondary analysis region, and primary analysis region.

[0097] Based on the correspondence, local differential binary feature data at the same spatial location in images from adjacent time periods are quickly matched to identify the spatial transformation relationship between feature point pairs. Specifically, this includes: based on the correspondence between the analysis area at each level and the feature data, selecting feature data located at the same spatial location in two adjacent acquisition time periods, which come from the same analysis area in the two time periods respectively; quickly matching local differential binary feature data at the same spatial location, and finding corresponding feature point pairs between the two time periods by comparing the binary descriptors of the feature points; and identifying the spatial transformation relationship between the feature point pairs based on the positional changes of the feature point pairs in the images of the two time periods, including translation, rotation, scaling, etc., which directly reflect the positional or morphological changes of the corresponding features in the area.

[0098] Based on spatial transformation relationships, the feature similarity of each tertiary analysis region is calculated. When the similarity is lower than a preset threshold, a state change is determined. The system then uses a construction knowledge base to identify the engineering entity type and state change type, resulting in a state change outcome. Specifically, based on the identified spatial transformation relationships, a feature matching algorithm is used to calculate the feature similarity of feature data from different time periods within each tertiary analysis region. This similarity reflects the consistency of land feature characteristics within the region. A preset similarity threshold of 85 is set. When the feature similarity of a tertiary analysis region is lower than 85, a state change is determined for that region. Simultaneously, the construction knowledge base is invoked. This knowledge base contains common engineering entity types in urban renewal projects, such as buildings to be demolished, excavated foundation pits, main structural frames, and pipeline laying sections, as well as typical state change characteristics corresponding to each type of entity, such as feature differences before and after building demolition, and feature changes after foundation pit pouring is completed. Combining the information in the knowledge base, the tertiary analysis regions experiencing state changes are identified, determining the engineering entity type and specific state change type within the region, such as completed building demolition, foundation pit pouring progress exceeding halfway, and deviations in pipeline laying, thus forming the state change outcome for each tertiary analysis region.

[0099] The state change results are aggregated upwards according to hierarchical relationships to obtain hierarchical structural change detection results. Specifically, this includes: firstly, aggregating the state change results of each tertiary analysis area to the corresponding secondary analysis area according to their relationship; combining the construction stage attributes of the secondary analysis area, statistically analyzing the changes of each tertiary area within that secondary analysis area to form the overall state change results of the secondary analysis area, such as how many buildings have been demolished in a demolition area or how many residential units have completed foundation construction in a residential construction area; then, further aggregating the state change results of all secondary analysis areas upwards to the primary analysis area, and combining the changes of all secondary areas to form the overall state change results of the entire urban renewal project, ultimately obtaining hierarchical structural change detection results that include the state change information of the primary, secondary, and tertiary analysis areas.

[0100] In this embodiment of the invention, because the benchmark analysis range is set according to the constructed hierarchical feature structure, and the analysis area is divided into first-level, second-level, and third-level analysis areas according to spatial scale and construction stage, the local differential binary feature data is mapped to the third-level analysis area according to spatial coordinates and aggregated upward to form a multi-level correspondence. The spatial transformation relationship is identified by quickly matching the same spatial location features of images in adjacent time periods, the feature similarity of the third-level areas is calculated to determine the state change, and the engineering entity and change type are identified by combining the construction knowledge base. Then, the change detection results are obtained by hierarchical aggregation. Therefore, this invention overcomes the technical problems of traditional change detection, such as lack of hierarchical division, inaccurate correspondence between features and analysis areas, low matching efficiency, fuzzy identification of engineering entity and state change, and lack of systematic hierarchical information to support accurate progress analysis. Thus, it achieves multi-level accurate change detection of urban renewal project areas from the whole to the part, improves feature matching efficiency and change identification accuracy, clarifies the engineering entity type and state change type, and outputs standardized detection results containing hierarchical structure information.

[0101] like Figure 2 As shown, in another preferred embodiment of the present invention, based on the change detection results and combined with the project schedule data, a comparison analysis of the actual progress and the planned progress is performed to obtain a schedule deviation report, including:

[0102] Based on the hierarchical structure change detection results, the change type, change area, and change time nodes of each level of analysis area are extracted. Specifically, this includes: extracting key information one by one according to the hierarchical order of the first-level, second-level, and third-level analysis areas, based on the obtained hierarchical structure change detection results; clarifying the change type for each analysis area, ensuring that the change type aligns with the urban renewal construction scenario, including specific construction-related changes such as building demolition completion, foundation pit excavation, main structure construction, pipeline laying, and site leveling; measuring and recording the change area, which is calculated through the geographical range mapped from the corresponding area feature data to accurately reflect the actual spatial scale of construction; and determining the change time nodes, which are based on the time of the corresponding image collected by the drone, accurate to the day, to ensure that the change type, change area, and change time node information of each level of analysis area are complete and consistent.

[0103] Retrieve planned progress data for corresponding time nodes from the project management database. This data includes the planned completion time, planned workload, and planned construction content for each construction area. Specifically, this involves: accessing the project management database, which stores planned progress information for the entire lifecycle of the urban renewal project; filtering and retrieving planned progress data for each time node based on the extracted change time nodes for each level of analysis area. The planned progress data needs to be detailed down to each analysis area. Specifically, the planned completion time should clearly specify the expected completion date for each construction task in each construction area, accurate to the day; the planned workload corresponds to the expected completion scale of each construction task, using area as the primary unit of measurement, maintaining consistency with the actual changed area; and the planned construction content should clearly specify the expected construction procedures or deliverables for each time node, such as the planned completion of foundation pouring in a third-level area at a certain time node, or the planned completion of 50% building demolition in a second-level area at a certain time node, ensuring that the retrieved data can be accurately compared with the actual change information.

[0104] The extracted change time points are compared with the retrieved planned completion times to calculate the time deviation value for each level of analysis area. Simultaneously, the actual changed area is quantitatively compared with the planned workload to calculate the workload completion rate. Specifically, for each level of analysis area, the extracted actual change time points are compared with the retrieved planned completion times. The actual change time point is subtracted from the planned completion time to obtain the time deviation value for each area. A positive result indicates a delay, while a negative result indicates an overshoot. At the same time, the actual changed area of ​​each area is quantitatively compared with the corresponding planned workload. The actual changed area is divided by the planned workload and then multiplied by 100 to obtain the workload completion rate, presented as a percentage to intuitively reflect the degree of alignment between the actual construction progress and the planned workload. During the calculation process, it is ensured that the calculation objects for each level and each area correspond one-to-one to avoid data confusion and guarantee the accuracy of the calculation results.

[0105] Based on the time deviation value and the completion rate of the project, combined with preset deviation threshold standards, the progress status of each construction area is determined, including ahead of schedule, normal, behind schedule, and severely behind schedule. Specifically, the preset deviation threshold standards are: time deviation value thresholds are set to -3 days, 3 days, and 10 days, and the completion rate thresholds are set to 70%, 90%, and 110%. A comprehensive judgment is made based on the time deviation value and the completion rate of the project: when the time deviation value is less than -3 days and the completion rate of the project is greater than 110%, the progress status is determined to be ahead of schedule; when the time deviation value is between -3 days and 3 days and the completion rate of the project is between 90% and 110%, the progress status is determined to be normal; when the time deviation value is between 3 days and 10 days and the completion rate of the project is between 70% and 90%, the progress status is determined to be behind schedule; when the time deviation value is greater than 10 days and the completion rate of the project is less than 70%, the progress status is determined to be severely behind schedule.

[0106] The system integrates the assessed progress status, time deviation values, and project completion rates to generate a progress deviation report that includes progress deviation analysis, problem area location, and preliminary judgment of deviation causes. Specifically, it includes: integrating the assessment results from each level of analysis area, including progress status, specific time deviation values, and precise project completion rates; in the progress deviation analysis section, detailing the differences between actual and planned progress in each area, including the specific degree of being ahead of schedule or behind schedule, and the achievement of project completion targets; in the problem area location section, clearly identifying areas with a delayed or severely delayed progress status, down to the third-level analysis area, such as a construction area for a specific residential unit within a newly built residential area, or a section of pipeline laying within a supporting facilities area; and in the preliminary judgment of deviation causes section, analyzing factors that may lead to progress deviations based on construction common sense and project realities, such as delayed demolition due to resident coordination issues, pipeline laying deviations due to improper construction techniques, and insufficient project completion rates due to insufficient construction personnel or machinery, to obtain a clear and comprehensive progress deviation report.

[0107] In this embodiment of the invention, by extracting the change type, change area, and change time node of each hierarchical analysis area from the hierarchical structure change detection results, retrieving the planned progress data of the corresponding time node, comparing the actual change time node with the planned completion time, and the actual change area with the planned workload, calculating the time deviation value and workload completion rate, and combining the preset deviation threshold standard to determine the progress status of each construction area, and integrating relevant data to generate a progress deviation report that includes progress deviation analysis, problem area location, and preliminary judgment of deviation causes, this invention overcomes the technical problems in traditional urban renewal project progress comparison, such as lack of quantitative analysis basis, inaccurate calculation of time and workload deviation, vague progress status judgment, difficulty in locating problem areas, and incomplete deviation report information, which cannot provide effective support for subsequent progress evaluation. This achieves accurate quantitative comparison between actual progress and planned progress, clearly defines the progress status of each construction area, accurately locates progress problem areas and preliminarily analyzes the causes of deviation, and generates a comprehensive and targeted progress deviation report.

[0108] In a preferred embodiment of the present invention, the overall progress trend and risk level of the project are assessed based on the schedule deviation report and historical schedule data to obtain the assessment result, including:

[0109] The schedule deviation report extracts schedule deviation values, work completion rates, and schedule status information for each analysis area. Specifically, this includes: extracting key data from the generated schedule deviation report by classifying analysis areas into first-level, second-level, and third-level categories; accurately extracting specific schedule deviation values ​​for each analysis area, with the values ​​directly reflecting the time difference between the actual progress and the planned progress; extracting the work completion rate accurate to a percentage, reflecting the degree of fit between the actual construction scale and the planned work volume; and clearly marking the schedule status, i.e., ahead of schedule, normal delay, or serious delay.

[0110] Historical progress data for similar construction areas will be retrieved from the project's historical database. This includes historical completion times, historical deviation trends, and historical risk management records. Specifically, this involves: accessing the project's historical database, which stores complete progress records for similar past urban renewal projects; filtering and retrieving historical progress data for similar construction areas based on the type of the current project's analysis areas, such as demolition areas, new residential construction areas, and supporting facility construction areas. The historical data must cover three core dimensions: historical completion time (the actual completion date of each construction task in similar past areas); historical deviation trends (including changes in progress deviations at different stages of construction in similar areas); and historical risk management records (detailed records of measures taken, implementation processes, and final effects when similar progress deviations occurred in similar past areas).

[0111] By performing time-series comparative analysis of progress status information and historical progress data, the development trend and changing patterns of current progress deviations are identified. Specifically, this includes: comparing the extracted current progress status information of each analysis area with the retrieved historical progress data one by one in chronological order; tracing the change trajectory of the current progress deviation from the project's inception to the present, and comparing it with the deviation change trajectory of similar areas in the same period in the past; analyzing the rate of change of the current progress deviation, such as whether the lag is continuously expanding or gradually shrinking, and whether the completion rate of the work is steadily increasing or fluctuating; identifying the patterns of deviation changes, such as whether deviation peaks are prone to occur in specific construction stages, and whether they are related to factors such as the connection of construction procedures and the influence of the external environment; and clarifying the development trend and core changing characteristics of the current progress deviation through comparison.

[0112] Based on the changing patterns and combined with pre-set progress prediction rules, the overall project completion time deviation and the probability of achieving key milestones are calculated to obtain the overall project progress trend assessment result. Specifically, this includes: formulating pre-set progress prediction rules based on the identified changing patterns; the prediction rules need to take into account the construction cycle characteristics of urban renewal projects, the connection logic of each construction stage, and the progress rate of similar historical projects to clarify the progress extrapolation standards under different deviation trends; based on the progress prediction rules, combined with the current progress deviation values ​​and work completion rates of each level of analysis area, the overall project completion time deviation is calculated, i.e., the difference between the expected actual completion time and the planned completion time; at the same time, the probability of achieving key milestones is calculated. Key milestones include core stages such as the completion of all demolition, the finishing of foundation construction, the topping out of the main structure, and the completion of supporting facilities. The probability of achievement is presented as a percentage, such as a 90% probability of achieving the demolition completion milestone and a 75% probability of achieving the main structure topping out milestone. Finally, an overall project progress trend assessment result is formed, which includes the completion time deviation prediction and the probability of achieving key milestones.

[0113] Based on the progress trend assessment results, combined with historical risk management effectiveness records, and in accordance with the risk level quantification standards, the current risk index of the project is calculated and the risk level is determined. Specifically, this includes: referring to retrieved historical risk management effectiveness records, analyzing the probability, scope of impact, and difficulty of handling risks under similar deviations and trends in the past, establishing risk level quantification standards, and dividing the risk index into three intervals: 0 to 30 for low risk, 31 to 60 for medium risk, and 61 to 100 for high risk; based on the obtained overall project progress trend assessment results, such as the magnitude of completion time deviations and the probability of achieving key milestones, combined with historical risk management effectiveness, the current project risk index is quantitatively calculated. Based on the risk index's interval, the current risk level of the project is determined: a risk index of 25 is considered low risk, 45 for medium risk, and 70 for high risk, ensuring that the risk level determination is objective and quantifiable.

[0114] By integrating the overall project progress trend assessment results with the determined risk levels, a comprehensive assessment result is obtained, including completion time prediction, risk warning levels, and targeted remedial suggestions. Specifically, this includes: integrating the overall project progress trend assessment results with the determined risk levels to form a structurally complete comprehensive assessment result. The completion time prediction clearly indicates the difference between the expected actual completion date and the planned completion date, such as expecting completion 5 days ahead or 12 days behind; the risk warning level directly corresponds to the risk level, i.e., low-risk warning, medium-risk warning, or high-risk warning; targeted remedial suggestions need to combine the current progress deviation trend risk level and historical handling experience to accurately target the core of the problem. For example, for a situation where the demolition area is severely lagging behind and at high risk, it is recommended to increase the number of demolition work teams and optimize the resident communication and coordination mechanism; for a situation where the construction progress of supporting facilities is ahead of schedule and at low risk, it is recommended to maintain the existing construction configuration and simultaneously advance the preparation of subsequent procedures; for a situation where the construction of new residential buildings is lagging behind and at medium risk, it is recommended to increase the number of construction machinery and optimize the connection of construction procedures, ultimately forming the comprehensive assessment result.

[0115] In this embodiment of the invention, the technical means of extracting progress deviation values, work completion rates, and progress status information from progress deviation reports for each analysis area, retrieving historical progress data from construction areas of the same type, identifying the current progress deviation development trend and change pattern through time-series comparative analysis, calculating the overall project completion time deviation and the probability of achieving key nodes by combining preset rules, calculating the risk index and determining the risk level by comparing with the risk level quantification standard and combining with the historical risk handling effect, and finally integrating to form a comprehensive evaluation result that includes completion time prediction, risk warning level, and targeted handling suggestions, overcomes the technical problems of traditional project evaluation that rely solely on current progress data, lack historical data support, have inaccurate progress trend predictions, subjective and vague risk level determination, and lack targeted handling suggestions, making it difficult to achieve forward-looking risk prevention and decision-making guidance. Thus, it achieves the technical effect of scientifically predicting the overall progress trend of the project, accurately quantifying and determining the risk level, providing project managers with a comprehensive evaluation result that is predictive, early warning, and practical, and effectively supporting forward-looking risk prevention and scientific decision-making.

[0116] In a preferred embodiment of the present invention, the evaluation results are integrated with the schedule deviation report to obtain a visual schedule management interface, supporting project managers to intervene in the schedule and make adjustments to decisions, including:

[0117] Based on the comprehensive assessment results and schedule deviation reports, completion time forecasts, risk warning levels, handling suggestions, and progress status information for each area are extracted. Specifically, this includes: accurately extracting completion time forecasts, risk warning levels, and targeted handling suggestions from the comprehensive assessment results; clearly indicating the difference between the overall estimated actual completion date and the planned completion date of the project, accurate to the day, and including the estimated achievement time of each key milestone, such as the estimated time for complete demolition, the estimated time for foundation construction completion, and the estimated time for main structure topping out; and classifying risk warning levels into three categories: low risk, medium risk, and high risk. Clearly label the risk status of the overall project and each secondary analysis area. The handling recommendations should be detailed to each tertiary analysis area, and specific recommendations should be given for different progress statuses and risk levels. For example, personnel allocation recommendations when a demolition area is severely lagging and at high risk, and maintenance measures when a residential construction area is at normal progress. Extract the progress status of each level of analysis area from the progress deviation report, including the advanced, normal, lagging and severe lagging status of the first, second and third level analysis areas respectively. At the same time, link the time deviation value and the completion rate of the work volume of each area to ensure that all extracted information is complete and consistent, without omissions or misalignments.

[0118] Based on the acquired status information and combined with the hierarchical feature structure of the corresponding spatial coordinates, a progress status dataset containing both time and spatial dimensions is constructed. Based on this dataset, multi-dimensional visualization components are generated, specifically including: Associating the corresponding spatial coordinates of each level of analysis region with all extracted core information. The spatial coordinates accurately correspond to the geographical range in the hierarchical feature structure, ensuring that each information point can be located at a specific spatial position. This constructs a progress status dataset containing both time and spatial dimensions. The time dimension covers all time periods of UAV image acquisition, divided by day, recording the progress changes in each region during each time period. The spatial dimension is divided into first-, second-, and third-level analysis regions, clearly defining the geographical coordinate boundaries and corresponding progress information of each region, forming... A complete dataset with spatiotemporal correlation; based on the complete dataset, multi-dimensional visualization components are generated, including a spatial distribution component that displays the geographical location of each analysis area in map form, using different colors to indicate progress status: red indicates severe lag, orange indicates lag, blue indicates normal, and green indicates ahead of schedule, with color intensity corresponding to the degree of deviation; a time trend component that presents changes in the completion rate and time deviation value of each area in line graph form, intuitively displaying the dynamic trend of progress; a risk warning component that uses eye-catching icons and colors to indicate the risk warning level of each area, with high-risk areas marked with flashing icons; and a progress details component that lists detailed information such as progress status, time deviation value, completion rate of work volume, and handling suggestions for each area in tabular form, facilitating quick access for management personnel.

[0119] The system integrates multi-dimensional visualization components into a unified progress management interface. Based on this interface, it provides a progress intervention toolset, receives adjustment instructions from project managers, generates progress adjustment plans, and feeds them back to the project plan database. Specifically, this includes: integrating the generated multi-dimensional visualization components into a unified progress management interface with a logical layout; a top navigation bar containing modules such as spatial view, time trend, risk warning, and progress details, which managers can freely switch between; a central area with a spatial distribution component as its core, overlaid with risk warning indicators, intuitively presenting the overall progress and risk distribution; a progress details component on the right, where clicking on any area within the spatial distribution component displays detailed progress information and handling suggestions for that area; and a bottom time trend component, supporting filtering and viewing by hierarchical analysis area and time range, based on progress. The management interface provides a comprehensive set of progress intervention tools, including a progress adjustment calculator. Managers can input plan adjustment parameters, such as increasing the number of construction teams or extending daily construction hours, and the toolset can automatically calculate the adjusted estimated completion time and changes in the project completion rate. A resource allocation suggestion generator automatically generates allocation plans for personnel, machinery, and materials based on the specific conditions of lagging areas and historical data. A key node adjustment module allows managers to directly modify the planned completion time of key nodes, and the system synchronously updates the overall progress forecast. Upon receiving adjustment instructions from managers through the toolset, the system automatically integrates the instruction information, generates a progress adjustment plan that includes adjustment objectives, adjustment measures, expected effects, and implementation steps, and synchronously feeds the progress adjustment plan back to the project planning database, updating the corresponding planned progress data to achieve closed-loop adjustment of progress management.

[0120] In this embodiment of the invention, by extracting completion time predictions, risk warning levels, handling suggestions, and progress status information of each region from comprehensive evaluation results and schedule deviation reports, and constructing a progress status dataset containing time and space dimensions by combining the hierarchical feature structure of corresponding spatial coordinates, a multi-dimensional visualization component is generated based on this dataset and integrated into a unified progress management interface. Simultaneously, a progress intervention toolset is provided to receive adjustment instructions from management personnel, generate progress adjustment plans, and feed them back to the project plan database. Therefore, this overcomes the technical problems in traditional project management, such as scattered evaluation results and schedule deviation information, lack of spatiotemporal integration, low visualization, difficulty for management personnel to intuitively grasp the overall project status, lack of convenient tools for progress intervention, and inability to effectively close the loop of adjustment plans. This achieves a spatiotemporally integrated visual presentation of project progress information, allowing management personnel to intuitively and clearly grasp project completion predictions, risk warnings, and progress status of each region. Convenient intervention tools enable rapid formulation and feedback of progress adjustment plans, providing intuitive and operable technical support for efficient decision-making and adjustments, and improving the closed-loop nature and operability of project progress management.

[0121] Embodiments of the present invention also provide a computing device, including: a processor and a memory storing a computer program, wherein the computer program, when executed by the processor, performs the system as described above. All implementations in the above system embodiments are applicable to this embodiment and can achieve the same technical effects.

[0122] Embodiments of the present invention also provide a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the system as described above. All implementations in the above system embodiments are applicable to this embodiment and can achieve the same technical effects.

[0123] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A system for managing and evaluating the progress of urban renewal projects based on UAV imagery, characterized in that: include: The acquisition module is used to periodically collect sequential images of urban renewal project areas using drones, and to preprocess the collected images to obtain standardized image data. The computation module is used to extract features from standardized image data and generate binary feature descriptors by combining the grayscale information and gradient information of the neighborhood of feature points in the image, so as to obtain local differential binary feature data with the advantages of high discriminativeness and fast matching. A hierarchical feature structure for construction scenarios is constructed based on locally differential binary feature data, including: Based on standardized image data, multiple feature points are extracted using a preset feature point detection algorithm. Within the neighborhood of each feature point pixel, gray-level difference and gradient direction difference are calculated simultaneously according to a preset sampling mode. The comparison results are encoded into binary strings to generate binary feature descriptors for each feature point, resulting in a local difference binary feature dataset, including: Based on the extracted multiple feature points, a pixel neighborhood region of a preset size is determined for each feature point; Within the pixel neighborhood, multiple pairs of sampling points are selected according to a preset sampling mode, and each pair of sampling points includes a first sampling point and a second sampling point; Based on the selected sampling point pairs, calculate the gray-level difference and gradient direction difference between each pair of sampling points; The calculated grayscale difference value is compared with the preset grayscale threshold, and the gradient direction difference value is compared with the preset gradient threshold to obtain the comparison result. The comparison results are converted into binary bits according to a preset encoding rule, and all binary bits are combined to form a binary string describing the features. Each feature point is assigned a binary string describing the feature as a binary feature descriptor. The binary feature descriptors of all feature points are aggregated to obtain the local difference binary feature dataset. By using the local differential binary feature dataset, feature points with spatial proximity and descriptor similarity are clustered to form primary feature clusters for characterizing specific construction areas. Based on the primary feature clusters, and according to the preset semantic segmentation rules of construction scenarios, they are combined according to spatial location and functional association to form upper-level feature groups for different construction stage partitions. Based on primary feature clusters and upper-level feature groups, a hierarchical feature structure for the construction scenario is constructed by organizing them according to the inclusion and contained relationship. The detection module is used to define the baseline analysis range in the hierarchical feature structure and divide the baseline analysis range into multiple levels to obtain multiple interrelated feature analysis regions. It maps the extracted local differential binary feature data to the corresponding feature analysis regions, and through rapid matching of local differential binary features of images from adjacent time periods, identifies the engineering entities and state changes represented by the features of each region, obtaining change detection results containing hierarchical structural information, including: Based on the constructed hierarchical feature structure, the benchmark analysis range is set within the hierarchical feature structure according to the project planning area; By dividing the benchmark analysis scope into multiple levels according to spatial scale and construction stage, we obtain the first-level analysis area, the second-level analysis area, and the third-level analysis area. The local differential binary feature data is mapped to the corresponding third-level analysis region according to the spatial coordinates, and then aggregated upwards to form a correspondence with the multi-level analysis region; Based on the correspondence, local differential binary feature data of the same spatial location in images of adjacent time periods are quickly matched to identify the spatial transformation relationship between feature point pairs. Based on the spatial transformation relationship, the feature similarity of each third-level analysis region is calculated. When the similarity is lower than the preset threshold, it is determined that a state change has occurred. The engineering entity type and state change type are identified by combining the construction knowledge base to obtain the state change result. The state change results are aggregated upwards according to the hierarchical relationship to obtain hierarchical structure change detection results; The comparison module is used to compare and analyze the actual progress with the planned progress based on the change detection results and the project schedule data, and to generate a schedule deviation report. The assessment module is used to evaluate the overall progress trend and risk level of the project based on schedule deviation reports and historical schedule data, in order to obtain the assessment results; The processing module integrates the evaluation results with the schedule deviation report to obtain a visual schedule management interface, which supports project managers in making schedule interventions and decision adjustments.

2. The urban renewal project progress management and evaluation system based on UAV imagery as described in claim 1, characterized in that, The urban renewal project area is regularly photographed using drones, and the collected images are preprocessed to obtain standardized image data, including: Receive the raw image sequence uploaded by the drone at regular intervals, and perform dehazing and illumination normalization processing on each image in turn to obtain the dehazed image. By performing camera distortion correction and geometric correction on the dehazed image, the corrected image is obtained; The corrected image is then subjected to feature point matching and image stitching to obtain a complete orthophoto of the project area. Standardized image data is obtained by resampling the resolution of the complete orthophoto and normalizing the coordinate system.

3. The urban renewal project progress management and evaluation system based on UAV imagery as described in claim 2, characterized in that, Based on the change detection results and combined with the project schedule data, a comparative analysis of the actual progress and the planned progress is conducted to obtain a schedule deviation report, including: Based on the hierarchical structure change detection results, extract the change type, change area and change time node of each hierarchical analysis area; Retrieve the planned progress data for the corresponding time node from the project management database. The planned progress data includes the planned completion time, planned workload, and planned construction content for each construction area. The extracted change time points are compared with the retrieved planned completion time to calculate the time deviation value of each level of analysis area; at the same time, the actual change area is quantitatively compared with the planned engineering quantity to calculate the engineering quantity completion rate. Based on the time deviation value and the completion rate of the project, combined with the preset deviation threshold standard, the progress status of each construction area is determined, including ahead of schedule, normal, behind schedule, and severely behind schedule. By integrating the determined progress status, time deviation value, and work completion rate, a progress deviation report is generated, which includes progress deviation analysis, problem area location, and preliminary judgment of the cause of deviation.

4. The urban renewal project progress management and evaluation system based on UAV imagery according to claim 3, characterized in that, Based on schedule deviation reports and historical schedule data, assess the overall project progress trend and risk level to obtain assessment results, including: The schedule deviation report is used to extract the schedule deviation value, work completion rate and schedule status information for each analysis area. Retrieve historical progress data for similar construction areas from the project history database, including historical completion time, historical deviation trends, and historical risk management records; By comparing and analyzing the progress status information with historical progress data over time, the development trend and changing pattern of the current progress deviation can be identified. Based on the changing patterns and combined with the preset progress prediction rules, the overall project completion time deviation and the probability of achieving key milestones are calculated to obtain the overall project progress trend assessment results. Based on the progress trend assessment results, combined with historical risk management effectiveness records, and in accordance with the risk level quantification standards, the current risk index of the project is calculated and the risk level is determined. By integrating the overall project progress trend assessment results with the determined risk level, a comprehensive assessment result is obtained, which includes completion time prediction, risk warning level, and targeted handling suggestions.

5. The urban renewal project progress management and evaluation system based on UAV imagery according to claim 4, characterized in that, The assessment results are integrated with the schedule deviation report to create a visual schedule management interface, supporting project managers in making schedule interventions and adjustments, including: Based on the comprehensive assessment results and schedule deviation reports, the completion time forecast, risk warning level, handling suggestions, and progress status information for each region are extracted. Based on the obtained status information and combined with the hierarchical feature structure of the corresponding spatial coordinates, a progress status dataset containing time and spatial dimensions is constructed; based on the progress status dataset, multi-dimensional visualization components are generated. The system integrates multi-dimensional visualization components into a unified schedule management interface. Based on this interface, it provides a schedule intervention toolset, receives adjustment instructions from project managers, generates schedule adjustment plans, and feeds them back to the project plan database.

6. A computing device, characterized in that, include: One or more processors; A storage device for storing one or more programs, which, when executed by the one or more processors, cause the one or more processors to perform the system as described in any one of claims 1 to 5.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program that, when executed by a processor, performs the system as described in any one of claims 1 to 5.

Images