Land investigation, analysis and comparison method and system

Abstract

The invention relates to a land investigation, analysis and comparison method and system, and belongs to the technical field of land investigation and data analysis, and the method comprises the steps: receiving an investigation and analysis instruction, and obtaining multi-source data of an investigated object, the multi-source data at least comprising a remote sensing image and land utilization attribute information; preprocessing the multi-source data to obtain processed data; based on a preset comparison logic, comparing the processed data with specified reference data, and generating and outputting a comparison result for an investigator to know; wherein the specified reference data refers to multi-source data of the investigated object with the same description as the processed data; the comparison result is at least used for representing the deviation between the processed data and the specified reference data, and the deviation is used for representing the change condition of the investigated object. According to the invention, the accuracy and high efficiency of land investigation and data analysis can be improved.

Description

Technical Field

[0001] This application relates to the field of land survey and data analysis technology, and in particular to a land survey analysis and comparison method and system. Background Technology

[0002] Land surveys are fundamental for understanding the state of land resources, serving national spatial planning, and ensuring national food and ecological security. Traditional land survey techniques mainly rely on manual ground surveys and single-source remote sensing image sets, such as traditional ground surveys or single remote sensing images. Furthermore, manual analysis methods are prone to issues of accuracy and efficiency due to subjective human factors, thus requiring improvement. Summary of the Invention

[0003] To optimize the accuracy and efficiency of land surveys and data analysis, this application provides a land survey analysis and comparison method and system.

[0004] Firstly, this application provides a land survey analysis and comparison method, which adopts the following technical solution: Receive survey and analysis instructions and acquire multi-source data of the surveyed object, wherein the multi-source data includes at least remote sensing images and land use attribute information; The multi-source data is preprocessed to obtain processed data; Based on the preset comparison logic, the processed data is compared with the specified benchmark data, and the comparison results are generated and output for investigators to know. Wherein, the designated benchmark data refers to multi-source data of the surveyed subjects that are identical to the processed data; and the comparison result is used at least to characterize the deviation between the processed data and the designated benchmark data, the deviation being used to characterize the changes in the surveyed subjects.

[0005] By adopting the above technical solution, and through the automatic integration of multi-source data and the standardized preprocessing process of multi-source data, the problem of incomplete and inaccurate analysis caused by single survey data is solved. Through the preset comparison logic, data analysis and comparison are automatically realized, reducing the misjudgment problem caused by human subjective judgment, improving the accuracy and efficiency of land change identification, realizing the transformation from "human judgment" to "machine judgment", and establishing an automated pipeline from data collection, processing, comparison to result update. It can respond quickly to land changes and realize real-time or near-real-time dynamic updates of survey data, greatly improving the timeliness of data.

[0006] Optionally, the processed data and the specified baseline data refer to the data obtained after processing multi-source data acquired by the survey subjects at different times; The process, based on preset comparison logic, compares the processed data with specified benchmark data, generates and outputs the comparison results for investigators to access, including: Spatial comparison and association are performed between the processed data and the patches contained in the specified reference data. If associated patches are determined, the overlap and positional offset of the associated patches are calculated to generate a spatial consistency judgment result. The associated patches refer to the patches in the processed data that are spatially associated with the patches in the specified reference data. Based on the association relationship in the spatial consistency judgment result, the fuzzy comprehensive evaluation method is used to calculate the similarity of attribute information between associated patches to generate attribute consistency comparison results and realize attribute comparison. Based on the spatial consistency determination results and attribute consistency determination results, the regions where the investigated object changes over time are identified, and the identification results are output for the investigators to know.

[0007] By adopting the above technical solutions and using a multi-level intelligent comparison logic that combines spatial, attribute, and temporal factors, the subjective errors of human judgment are significantly reduced, further improving the accuracy and efficiency of land use change identification.

[0008] Optionally, the method further includes: Analyze the identification results output from historical periods, mine the multi-source data of the changed areas before and after the changes, and generate bundled change events based on the preset bundled change event mining algorithm. The bundled change events contain multi-source datasets obtained from two different monitoring times, and are limited to the following: the investigated object described by the multi-source dataset of the earlier monitoring time is likely to change in the future, and the multi-source data of the investigated object obtained after the change will be consistent with the multi-source dataset of the later monitoring time. The processed data of the current surveyed object is obtained and matched with the bundled change event. When the processed data contains target data and successfully matches the multi-source dataset of the earlier monitoring time included in the bundled change event, the multi-source dataset of the later monitoring time in the bundled change event that successfully matches the target data is used as the future change prediction result of the patch corresponding to the target data. The future change prediction result is output for the investigators to know.

[0009] By adopting the above-mentioned technical solutions, the traditional land survey can be changed from a passive mode that can only record changes that have already occurred. It can proactively predict areas that may change, providing valuable lead time for land supervision. Planning departments can assess regional development trends in advance based on the prediction results, making land spatial planning more forward-looking and avoiding blind and passive approaches.

[0010] Optionally, the bundled change event also includes the bundled conditions and the bundled strength corresponding to the bundled conditions. The bundled strength is used to characterize the probability that the surveyed object described by the multi-source data of the earlier time period will change in the future, and that the multi-source data of the surveyed object monitored after the change will be consistent with the multi-source data of the later time period. The step of finding target data in the processed data and successfully matching it with the multi-source dataset of the earliest monitoring time included in the bundled change event includes: When the processed data contains target data that is consistent with the multi-source data of the earlier time period included in the bundled change event, and the patch described by the target data meets the corresponding bundling conditions, the matching is considered successful. The method further includes: When outputting the prediction results of future changes, the corresponding probability value of change represented by the binding strength is also output.

[0011] By adopting the above technical solution, the change probability value quantifies the likelihood of change, so that law enforcement agencies can focus on monitoring high-risk patches (i.e. patches with high change probability values), transforming "ordinary patrols" into "precise targeted patrols," which greatly improves the pertinence of supervision.

[0012] Optionally, the method further includes: Whenever an identification result is generated, multi-source features are extracted from the processed data corresponding to the changed area, and the multi-source features are analyzed and the cause type is output based on a preset cause analysis model; wherein, the cause type includes changes caused by human activities and changes caused by natural factors.

[0013] By adopting the above technical solutions and analyzing the causes of changes in the affected areas, the system can not only identify "where has changed" but also answer "why it has changed," providing a scientific basis for responsibility definition and precise management (for example, distinguishing between illegal land occupation and natural disasters).

[0014] Optionally, the method further includes: For areas where changes have occurred, assess the change in land value before and after the changes, whereby the land value includes at least economic and ecological value; Based on the analysis results of the changes in land value and their causes, suggestions for optimizing land use are generated and output according to a pre-set suggestion rule base.

[0015] By adopting the above-mentioned technical solutions, land management is expanded from a single economic dimension to the dimensions of economic and ecological benefits, avoiding the short-sighted behavior of "GDP-only" thinking and supporting sustainable development decisions. The system can proactively assess the gains and losses of land changes and put forward forward-looking optimization suggestions, thereby assisting government departments to shift from passive response to proactive planning and refined management, and greatly improving the modernization level of land space governance.

[0016] Optionally, the method further includes: Acquire all recognition results generated in historical periods, calculate the stability index of each patch, and define the regions where patches with stability indices higher than a preset stability threshold are located as stable regions. Stable regions that are spatially adjacent and have the same land use type are merged to form an independent comparison unit; When any change is detected in any comparison unit, the comparison unit is split and restored to the region before merging; Before performing spatial comparison and correlation of the processed data and the patches contained in the specified reference data, the method further includes: If the geographical area covered by the surveyed object contains comparison units, then for each comparison unit, all the map patches in the area where each comparison unit is located are merged to form a complete map patch.

[0017] By adopting the above technical solution, a large number of stable small patches are merged into a few large comparison units, significantly reducing the number of entities requiring spatial calculations and attribute comparisons. For large-scale land surveys, this saves substantial computation time and resources, resulting in a significant increase in efficiency. It frees up computational resources from repetitive, unchanging comparisons in "stable areas," allowing resources to be concentrated on "sensitive areas" and "change areas" that have undergone real changes or are disputed, making the entire analysis process more focused and intelligent. This optimization is performed while maintaining accuracy. Because the objects to be merged are "stable blocks" identified through rigorous historical data analysis, the probability of changes within them is extremely low. Therefore, merging them ensures that meaningful change information is not missed, achieving a balance between efficiency and accuracy.

[0018] Secondly, this application provides a land survey analysis and comparison system, including, The multi-source data acquisition module is used to receive survey and analysis instructions and acquire multi-source data of the surveyed object, wherein the multi-source data includes at least remote sensing images and land use attribute information; The data preprocessing module is used to preprocess the multi-source data to obtain processed data; The land change analysis module is used to compare the processed data with the specified benchmark data based on the preset comparison logic, generate and output the comparison results for the investigators to know. Wherein, the designated benchmark data refers to multi-source data of the surveyed subjects that are identical to the processed data; and the comparison result is used at least to characterize the deviation between the processed data and the designated benchmark data, the deviation being used to characterize the changes in the surveyed subjects.

[0019] Thirdly, this application provides a land survey analysis and comparison device, including a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executed as described in any of the methods in the first aspect.

[0020] Fourthly, this application provides a computer-readable storage medium storing a computer program that can be loaded by a processor and executed as described in any of the first aspects.

[0021] In summary, this application includes at least one of the following beneficial technical effects: In this application, by automatically integrating multi-source data and performing standardized preprocessing on the multi-source data, the problem of incomplete and inaccurate analysis caused by single survey data is solved. By automatically realizing data analysis and comparison through preset comparison logic, the misjudgment problem caused by human subjective judgment is reduced, the accuracy and efficiency of land change identification are improved, and the transformation from "human judgment" to "machine judgment" is realized. An automated pipeline from data collection, processing, comparison to result update is established, which can respond quickly to land changes and realize real-time or near-real-time dynamic updates of survey data, greatly improving the timeliness of data. Furthermore, by combining spatial, attribute, and temporal intelligent comparison logic, the system significantly reduces human subjective misjudgment and further improves the accuracy and efficiency of land use change identification. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart illustrating the land survey analysis and comparison method disclosed in the embodiments of this application.

[0024] Figure 2 This is a structural block diagram of the land survey analysis and comparison system disclosed in the embodiments of this application.

[0025] Figure labeling: 201, Multi-source data acquisition module; 202, Data preprocessing module; 203, Land change analysis module. Detailed Implementation

[0026] The following is in conjunction with the appendix Figure 1-2 This application will be described in further detail.

[0027] This application discloses a land survey analysis and comparison method (hereinafter referred to as the analysis and comparison method). The execution subject of the analysis and comparison method is a land survey analysis and comparison system (hereinafter referred to as the analysis and comparison system). The following will be combined with Figure 1 The specific steps of the analysis and comparison system's analysis and execution method are described in detail.

[0028] S101, Receive survey and analysis instructions and obtain multi-source data of the surveyed object, wherein the multi-source data includes at least remote sensing images and land use attribute information.

[0029] S102, preprocess the multi-source data to obtain the processed data.

[0030] S103, based on preset comparison logic, compares the processed data with specified benchmark data, generates and outputs the comparison results for investigators to know. The specified benchmark data refers to multi-source data of the surveyed subjects that describes the same information as the processed data; and the comparison results are used to characterize at least the deviation between the processed data and the specified benchmark data, with the deviation characterizing the changes in the surveyed subjects.

[0031] Specifically, S103 includes the following steps: S1031, Spatial comparison and association are performed between the processed data and the polygons contained in the specified reference data. If associated polygons are determined, the overlap and position offset of the associated polygons are calculated to generate a spatial consistency judgment result. Among them, associated polygons refer to polygons that are spatially associated with the polygons contained in the processed data and the polygons contained in the specified reference data. S1032, Based on the association relationship in the spatial consistency judgment result, the fuzzy comprehensive evaluation method is used to calculate the similarity of attribute information between associated map patches to generate attribute consistency comparison results and realize attribute comparison; S1033, based on the spatial consistency determination results and attribute consistency determination results, identifies the areas where the surveyed object changes over time and outputs the identification results for the investigators to know.

[0032] In practice, in this invention, the investigated object can be specifically represented as an entity with a unique identifier (such as a plot number) and a clearly defined geographical scope. This scope can be an administrative division (such as a county), a planning area, or a naturally formed continuous plot of land. The analysis and comparison system of this application provides a platform accessible to investigators, displaying a map on a preset display interface. The map is divided into several plots, and investigators can trigger selection instructions based on preset trigger buttons to define a clear geographical scope on the map. This geographical scope consists of plots, which is the investigated object currently identified by the investigators, thereby generating investigation and analysis instructions with the investigated object in mind.

[0033] Next, the analysis and comparison system will acquire multi-source data of the surveyed subjects. This multi-source data specifically includes, but is not limited to, the following four categories: Remote sensing imagery data: Panchromatic or multispectral images with a resolution of 0.5 to 2 meters can be obtained by connecting to high-resolution remote sensing satellite data interfaces (such as the domestic Gaofen series, WorldView, etc.). This data provides macroscopic and periodic land cover information.

[0034] Land use attribute information: This information is manually entered by ground survey personnel after conducting on-site surveys using smart terminals (such as PDAs or mobile apps) equipped with GPS modules. Attribute information includes, but is not limited to: land use type (hereinafter referred to as land category, such as cultivated land, forest land, and construction land), land ownership (including owner and user), area, soil type, and crop type. The smart terminal also supports taking and uploading photos and videos on-site as supporting evidence.

[0035] UAV aerial survey data: By adapting to the data receiving units of various UAV models, centimeter-level precision orthophotos and 3D point cloud data are acquired. This data is used to compensate for the lack of detail in satellite imagery and to conduct detailed mapping of key areas.

[0036] IoT sensor data: Access environmental sensor network data pre-set at the geographical location corresponding to the surveyed object through IoT interface protocols (such as MQTT, HTTP) to obtain physical parameters such as soil moisture, vegetation index, and surface temperature in real time or periodically.

[0037] It should be noted here that multi-source data can be obtained through periodic detection, and whenever multi-source data for any geographic area is obtained (the multi-source data here includes at least the four types of data mentioned above), the corresponding area can be used as the subject of investigation, and the generation of investigation and analysis instructions can be automatically triggered. This enables the investigation and analysis operation to be automatically started and updated as the multi-source data is updated.

[0038] After receiving the aforementioned multi-source data, the analysis and comparison system will preprocess it to eliminate the heterogeneity of the multi-source data, so that comparison can be performed in a unified environment, i.e., to obtain the processed data. The corresponding preprocessing operations to obtain the processed data are as follows: Data denoising: Wavelet transform algorithm is used to remove coherent noise (such as strip noise) from remote sensing images, and mean filtering or median filtering algorithm is used to process speckle noise caused by sensor issues for UAV images.

[0039] Format conversion: Use format conversion libraries (such as GDAL / OGR) to convert data from different sources (such as TIFF format remote sensing images, JPEG format drone photos, and Excel format attribute tables) into GIS industry standard vector formats (such as SHP format) or GeoJSON format suitable for web applications.

[0040] Coordinate unification: The Gauss-Kruger projection transformation algorithm is used to convert and unify all data from different coordinate systems (such as WGS84, Beijing 54, and Xi'an 80) to the nationally recognized 2000 National Geodetic Coordinate System (CGCS2000), ensuring that the spatial location benchmark of all data is consistent.

[0041] Data Fusion: For image data, a deep learning-based feature fusion algorithm is employed. Specifically, a dual-branch convolutional neural network (CNN) is used to extract macroscopic contextual features from high-resolution remote sensing images and local detail features from UAV images, respectively. These features are then stitched together at a grid feature layer, and finally reconstructed through a deconvolution layer to create a fused image that simultaneously contains macroscopic information and detail features, laying the foundation for subsequent high-precision change detection.

[0042] Then, the analysis and comparison results will be based on the preset comparison logic to analyze the processed data. The preset comparison logic is actually a multi-level analysis process that coordinates spatial comparison, attribute comparison and temporal comparison.

[0043] The comparison requires at least two different datasets (processed data and baseline data). Baseline data refers to a highly authoritative and accurate land dataset used as a reference standard in a specific comparison task. It represents the correct state recognized by official or business rules at a certain point in time or under a certain standard. Specific sources include, but are not limited to, the following: a. Official authoritative data (such as the previous year's land change survey results database, land ownership registration and certification data that has passed acceptance inspection, legally approved land spatial planning maps, etc. This is the most common source of baseline data); b. Historical data at a specific point in time (e.g., when analyzing land changes, survey data at time point T1 (i.e., the time when the data was generated, such as 2020) can be set as the baseline and compared with data at time point T2 (i.e., the time when the data was generated, such as 2023) to identify changes between T1 and T2); c. Correct data verified by humans: In the quality control stage before data entry, a batch of data verified by experts can be used as a baseline to verify the newly entered automated identification results. In this embodiment of the application, the data source described in b above (i.e., historical data at a certain point in time) is used as the baseline data.

[0044] Processed data refers to newly acquired land datasets or datasets of target areas that need to be compared with baseline data to verify their consistency, accuracy, or identify changes. Specific sources include, but are not limited to, the following: d. Newly collected survey data (such as data newly acquired through remote sensing, drones, etc.); e. Data from suspected problem areas (such as suspected illegal land use patches discovered based on law enforcement clues or remote sensing monitoring, which need to be compared with legally mandated planning maps or ownership maps (baseline data) to confirm their legality); f. Supplementary data from different sources (such as data from a key area quickly collected by low-cost drones, which needs to be compared with high-precision official base maps (baseline data) to assess their accuracy or update the data). In this embodiment, the data source described in d above (i.e., the data obtained and preprocessed corresponding to the current survey and analysis instructions) is used.

[0045] The core of spatial comparison is calculating the spatial relationships between the geometric shapes (area patches) of the same geographic entity within the two datasets corresponding to the processed data and the baseline data. The specific implementation steps are as follows: 1. Unify the baseline data and processed data into a planar vector layer (such as the Polygon type in an SHP file). For raster images (such as TIFF and JPG), the land use boundaries need to be extracted into polygonal patches through image interpretation and vectorization. Each extracted patch has recorded content, which includes at least attribute information (such as land use codes).

[0046] 2. For each polygon in the processed data, find the corresponding or spatially most relevant polygon in the baseline data. This is usually achieved through spatial queries: using each polygon in the processed data as the "query subject" and its spatial location as the "reference point," query the baseline data for polygons within the same spatial range as the processed data, and then determine whether there is a relationship between the polygons. This can be done using spatial relationship functions in GIS, such as Intersects or Within a certain distance (where "certain distance" refers to a specific preset fixed value based on the polygon size, such as 5 meters as mentioned below; for example, if there are several preset polygon area ranges and distance values ​​corresponding to each polygon area range, the distance value is used as the specific value of "certain distance"). For polygon A in the processed dataset, search for polygon B in the baseline dataset that intersects its boundary or is closest to its center point (e.g., within a 5-meter tolerance range). If found, polygon A and polygon B are considered a pair of related polygons that need to be compared.

[0047] It should be noted that successful association scenarios include one-to-one matching, such as when patch A in the processed dataset and patch B in the baseline data largely overlap spatially, forming a unique association; in addition, it also includes one-to-many matching, such as when patch A in the processed dataset may cover two patches B1 and B2 in the baseline data. For example, two originally adjacent farmlands (B1 and B2) are merged into one construction land (patch A). Patch A is identified as intersecting with patches B1 and B2, and patch A is compared with the whole of patches B1 and B2 to confirm whether the above-mentioned judgment logic for associated patches is met. If it is met, the merging relationship of patches B1 and B2 is recorded.

[0048] In the case of association failure, that is, if there is a patch C in the processed dataset, but there is a blank space or no corresponding patch in the spatial location of the baseline data, it is very likely that patch C is newly added construction land, newly reclaimed farmland, etc. In this case, patch C is marked as "new patch" and its spatial location is recorded.

[0049] In addition, the system is also used for reverse queries, that is, using the baseline data as a reference to query the processed data. Specifically, the system iterates through each patch in the baseline data. If a patch D in the baseline dataset satisfies either the condition that patch D is not associated with any patch in the processed data to form an associated patch (i.e., loss scenario one), or that no new patch exists at the location of patch D (i.e., loss scenario two), then it is considered that a lost patch has appeared. For example, the land occupied by an illegal building that has been legally demolished is restored to its original state. If it is loss scenario one, then patch D is considered to be lost in physical space; if it is loss scenario two, then patch D is considered to be lost with specific attributes and identity (i.e., a new land type has replaced the original land type). Finally, the system determines the association status of all patches in the processed dataset: associated, newly added, or lost. For the loss status, the loss scenario can be further marked (i.e., loss of physical space, loss of specific attributes and identity as mentioned above).

[0050] 3. For each pair of associated polygons established in the previous step (polymorphism A and polymorphism B are used as examples below), calculate the following two indices (overlap and offset): 3.1 Using the Intersection algorithm in GIS overlay analysis, the overlap degree of related polygons is calculated as follows: Overlap degree = (Area of ​​the intersecting part of related polygons / Area of ​​polygons belonging to the processed data in the related polygons) * 100%. A higher overlap degree (e.g., ≥95%) indicates better spatial consistency between the two polygons. Low overlap degree may indicate boundary adjustments or changes in land use such as segmentation or merging.

[0051] 3.2 Calculate the Euclidean distance between the geometric centers (centroids) of associated patches: Calculate the centroid coordinates (Xa, Ya) and (Xb, Yb) of patches A and B respectively, and apply the distance formula to calculate the offset: Offset = sqrt((Xa - Xb)^2 + (Ya - Yb)^2). The offset directly reflects the overall positional deviation between the two patches. This is significant for evaluating the accuracy of UAV measurements or determining whether the boundary has shifted as a whole.

[0052] 4. Based on the calculated overlap of the associated patches and their geometric centers, spatial consistency is determined using a preset threshold: if the overlap is greater than the preset overlap threshold and the offset is less than the preset offset threshold, the spatial positions of the corresponding associated patches are considered to be consistent; otherwise, it is determined to be "spatial position change" or "boundary inconsistency," and the patches in the associated patch and the processed data to which it belongs (such as patch A) are marked as patches with changed meaning or patches that require manual verification. Finally, in the processed dataset, the spatial consistency determination results of all patches (including the overlap, offset value, and consistency determination conclusion of each patch) will be written into the corresponding record content in the form of attribute fields and used for subsequent attribute comparison and time series comparison.

[0053] Attribute comparison is used to compare the attributes of associated map features that have been successfully linked after spatial comparison. Specifically, it is used to compare whether the land use type and ownership type (such as state-owned land and collective land) of the associated map features are consistent. Specifically, the fuzzy comprehensive evaluation method is used to realize attribute comparison: First, the attribute factors participating in the comparison are determined, such as U={land use code, land ownership, area}. Then, a set of comments is established, such as V={highly consistent, basically consistent, inconsistent}. Finally, a membership function is defined for each attribute factor to each comment (i.e., highly consistent, basically consistent, inconsistent).

[0054] For example, a semantic similarity table is used for land category codes and land ownership. If the land category codes and land ownership are completely consistent, the comment is "highly consistent" with a membership degree of 1. If the land category codes belong to the same major category (e.g., "paddy field" and "irrigated land" belong to the same major category), the comment is "basically consistent" with a membership degree of 0.7. If the land category codes do not belong to the same major category or the land ownership is inconsistent, the comment is "inconsistent" with a membership degree of 0.3.

[0055] For the "area" factor, a difference rate function can be defined: for example, area difference rate δ = |Area A - Area B| / Area A. If δ = 0, the membership degree of "highly consistent" is 1; if δ < 5%, a descending ridge distribution function can be used to calculate its membership degree of "basically consistent" and "highly consistent".

[0056] Finally, weights are assigned according to the importance of each attribute, such as P = (0.6, 0.3, 0.1), indicating that the land type is the most important. The single-factor evaluation results are combined into a fuzzy relation matrix R, and then fuzzy transformed with the weight set A (e.g., using the M(^,v) operator or weighted average operator) to obtain a comprehensive evaluation vector B = P×R. The evaluation level with the highest membership degree in vector B is taken as the final attribute comparison result. Alternatively, a comprehensive similarity score S (e.g., the weighted average of vector B) is calculated, and a threshold is set (e.g., 90%). If S≥90%, it is judged as "attribute comparison consistent", otherwise it is judged as "attribute comparison inconsistent".

[0057] The attribute comparison results and spatial consistency judgment results are then combined to form a comprehensive assessment result. If the attributes remain unchanged (i.e., attribute comparison is consistent) but the boundaries are adjusted (i.e., boundaries are inconsistent); or if the attributes change (i.e., attribute comparison is inconsistent) but the spatial location is consistent, then the land category / use is considered to have changed. Finally, the comprehensive assessment result, the corresponding datasets (i.e., processed data and baseline data), the data generation time of the corresponding datasets (e.g., processed data generated in 2021, and baseline data generated in 2020), and the generation time of the comprehensive assessment result are stored as analysis records.

[0058] Time-series comparison is used to retrieve multiple analysis records and generate change trajectories to visually represent the changes in the spatial location and attributes of a geographical area over time. For example, a piece of land that was woodland in 2021 became a bulldozing area in 2022, with changes in its boundaries and an increase in area compared to 2021, and a new land parcel with a building type added in 2023. The specific implementation scheme is as follows: The temporal comparison is specifically implemented using Principal Component Analysis (PCA): For the retrieved analysis records, based on the data generation time and corresponding datasets contained in all analysis records, multi-band images from different data generation times and the same location are overlaid to form a data content containing 2 x N bands (N being the number of bands in a single image). The data content is then transformed using PCA, and its covariance matrix and eigenvectors are calculated. The resulting new components (principal components) are arranged in descending order of variance. The first principal component typically represents stable and unchanging information (such as topography and most land features) in images from different data generation times, while the second principal component focuses on the differences between images from different data generation times, i.e., change information. Thresholding is performed on the principal component images representing change (e.g., using the OTSU algorithm to automatically determine the threshold). Regions with pixel values ​​greater than the threshold are identified as change regions, thus generating a binarized change mask. This change mask is then vectorized to obtain the boundaries of the change patches. Finally, the spatial and attribute comparison results from the first and second steps are used to analyze each change patch. For example, if a changed land parcel is classified as "arable land" in period T1 and "construction land" in period T2, the system automatically determines the change type as "arable land converted to construction land". Finally, the system outputs a map showing the land parcel where the surveyed object is located through a GIS map, highlighting the area that has changed (i.e., the changed land parcel), and provides an interactive data query function so that when the queryer clicks on the corresponding changed area, the change trajectory of the changed area is displayed. The change trajectory includes the recorded content of the changed area before and after the change (i.e., the recognition result).

[0059] Optionally, the analysis and comparison method may also include the following steps: S104, analyze the identification results output in the historical period, mine the multi-source data of the changed area before and after the change, generate bundled change events based on the preset bundled change event mining algorithm, the bundled change event contains multi-source datasets obtained from two different monitoring times, and is limited to: the investigated object described by the multi-source dataset of the earlier monitoring time is likely to change in the future, and the multi-source data of the investigated object obtained after the change will be consistent with the multi-source dataset of the later monitoring time. S105. Obtain the processed data of the current surveyed object, match the processed data with the bundled change events, and if the processed data contains target data and successfully matches the multi-source dataset of the earlier monitoring time included in the bundled change event, then the multi-source dataset of the later monitoring time in the bundled change event that successfully matches the target data is used as the future change prediction result of the patch corresponding to the target data, and the future change prediction result is output for the investigators to know.

[0060] Among them, the bundled change event also includes the bundled conditions and the bundled strength corresponding to the bundled conditions. The bundled strength is used to characterize the probability that the surveyed object described by the multi-source data of the earlier time period will change in the future, and that the multi-source data of the surveyed object monitored after the change will be consistent with the multi-source data of the later time period. The step S105, "When the processed data contains target data and successfully matches the multi-source dataset of the earliest monitoring time included in the bundled change event," includes the following steps: When the processed data contains the target data, is consistent with the multi-source data of the earlier time period included in the bundled change event, and the patch described by the target data meets the corresponding bundling conditions, the match is considered successful. S106, when outputting the prediction results of future changes, also outputs the probability value of the change represented by the corresponding binding strength.

[0061] In practice, the analysis and comparison system is also used to periodically obtain a specified number of identification results before the current time. As mentioned above, the identification results include records of the area before and after the change, and these records are specifically reflected in preprocessed multi-source data (such as land type, land ownership, area, and geographical location).

[0062] The analysis and comparison system is used to extract the change trajectory of changed patches in the identification results. The change trajectory can be considered as the specific data that has changed, such as area, land type, land ownership, etc., as well as the data content before and after the change. For example, if the land type before the change was forest land, and the land type after the change is cultivated land, or if the area before the change is X, and the area after the change is Y, with the size relationship of Y relative to X, it can be used to characterize the state of the patch before and after the change. Based on this change trajectory, two change types are generated. The first change type has a clear state before and after the change, such as specific descriptions of changes in land type and land attribute before and after the change. The second change type refers to the type that can only describe the change trend, such as area increasing / decreasing. This application specifically analyzes, summarizes, and predicts the change trajectory of the first change type.

[0063] In other words, the analysis and comparison system inputs the change trajectory corresponding to the first change type in the identification results into a preset bundled change event mining algorithm. The underlying logic of this algorithm is to determine whether a fixed combination exists in the identification results. A fixed combination refers to the number of times that a certain pre-change patch state and a certain post-change patch state appear simultaneously in the same identification result exceeds a preset threshold. It should be noted that the specific value of the preset threshold can vary depending on the area where the change occurred; that is, the preset threshold value is different for different areas. For example, in area A, the land type before the change is "arable land," and the land type after the change is "construction land." This fixed combination occurs frequently in area A but rarely in area B. Therefore, the preset threshold value for this fixed combination in area A is greater than its value in area B. The preset threshold value can be manually set and stored for each area after periodic division.

[0064] For fixed combinations, corresponding bundled change events are generated. These bundled change events include the pre-change state (the multi-source datasets from earlier monitoring times mentioned above are multi-source data used to describe the pre-change state, i.e., the data corresponding to the first change type, specifically including land type and land attribute) and the post-change state (similarly, the multi-source datasets from later monitoring times mentioned above are multi-source data used to describe the post-change state). In addition, if fixed combinations exist, it is necessary to calculate the bundle strength and bundle conditions for the fixed combinations. For example, the bundle strength can be calculated as follows: calculate the difference between the number of occurrences of the fixed combination and a preset number threshold, and then determine the difference range in which the number of occurrences of the fixed combination falls based on multiple preset difference ranges and the bundle strength corresponding to each difference range (the bundle strength is specifically represented by a probability value, i.e., the change probability value mentioned above, such as 80%), thereby obtaining the corresponding change probability value.

[0065] The method for setting the binding conditions is as follows: cluster the identification results that contain the same fixed combination into a set, analyze and mine the common patterns of all identification results in the set other than the fixed combination, and the common patterns are the binding conditions. The specific methods for uncovering common patterns can be as follows: Based on the recorded content of the changing patches, examine the spatial characteristics (such as distance from the city center, distance from the nearest main road, altitude, slope, etc.), neighborhood characteristics (what are the main land types within a specified distance (such as 500 meters), whether there are any planned key projects nearby, etc.), and attribute characteristics (land ownership (collective / state-owned), whether it belongs to basic farmland, soil quality grade, etc.). Use clustering algorithms (such as K-Means) to analyze the above-mentioned features (i.e., spatial features, neighborhood features, attribute features) contained in the above identification results. If there are the same features in more than a preset number of identification results, then the corresponding same features are used as the binding condition. For example, if the vast majority of the identification results in the clustered set (e.g., more than 90%) are distributed within the spatial feature of "distance from the main road < 1 kilometer" and "slope < 15 degrees", then this spatial feature is used as the binding condition. The corresponding binding condition can be described as: [distance from the main road < 1000 meters and average slope < 15 degrees].

[0066] After receiving a survey and analysis command and generating identification results, the multi-source data (specifically, data corresponding to the first change type, such as land category and land attribute) of each patch (including changed and unchanged patches) in the identification results is compared with the pre-change states of all bundled change events stored in historical periods. If the comparison matches, the corresponding patch is designated as the target patch. The multi-source data used to describe the target patch, which participated in the aforementioned matching operation, is the target data. Further, it is determined whether other multi-source data of the corresponding patch meet the bundling conditions. If they do, the match is considered successful. The post-change state of the successfully matched bundled change event is used as the future change prediction result. Simultaneously, the corresponding future change prediction result and the change probability value representing the bundling strength are output. For example, the target patch, along with its future change prediction result and change probability value, are highlighted on a preset GIS map. The default GIS map covers the land parcel corresponding to the surveyed object.

[0067] Optionally, the comparative analysis method may also include the following steps: Whenever an identification result is generated, multi-source features are extracted from the processed data corresponding to the changed area, and the multi-source features are analyzed based on a preset causal analysis model to output the causal type; the causal type includes changes caused by human activities and changes caused by natural factors.

[0068] In practice, whenever the analysis and comparison system determines the changed patches based on the identification results, it extracts multi-source features from the multi-source data of the changed patches and generates a set of feature values. The corresponding multi-source features include: Geometric morphological characteristics: Based on the vector boundary data (i.e., the vertex coordinates of the polygons corresponding to the changing patches) from the multi-source data of the changing patches, the aspect ratio (by calculating the ratio of the long side to the short side of the minimum bounding rectangle of the changing patch; the aspect ratio of human-intervened plots is usually low or a specific value, such as 2:1, while the aspect ratio of naturally formed patches may be extremely high or irregular) and boundary tortuosity (using algorithms such as box-counting dimension to calculate the fractal dimension of the boundary line as the boundary tortuosity; the closer the boundary tortuosity value is to 1, the smoother the boundary, indicating that the changing patch is more likely to be formed by human intervention; if the boundary tortuosity value is closer to 2, it means that the boundary is more tortuous, indicating that the changing patch is more likely to be generated by natural environmental influences). Spectral and Texture Features: Based on multispectral or hyperspectral remote sensing images before and after the change in light spot intensity, the mean and standard deviation of brightness for all pixels within the changed light spot in each band (e.g., red, green, near-infrared) are calculated. For example, vegetation has extremely high reflectivity in the near-infrared band, while concrete has lower reflectivity. By comparing the difference in spectral mean before and after the change, material transformation can be determined. The gray-level co-occurrence matrix algorithm can also be used to analyze the spatial relationship of gray values ​​between pixels to quantify texture. For example, the gray-level co-occurrence matrix algorithm can be used to calculate multiple texture indices such as contrast, entropy, homogeneity, and correlation.

[0069] Temporal scale feature extraction: Based on remote sensing image sequences of multiple time phases (such as monthly or quarterly images) corresponding to the changed patches, the rate of change is calculated (rate of change = changed area / change time, where the changed area is the area difference of the changed patch before and after the change, and the change time is the time interval before and after the change). It is assumed that: a larger rate of change corresponds to the influence of the natural environment, and a smaller rate of change corresponds to the influence of human expansion.

[0070] Spatial correlation characteristics: Based on vector data from multi-source data of changing patches and a pre-defined background geographic database (which stores road networks, settlements, planning maps, and geological hazard zoning maps), the Euclidean distance from the centroid of the changing patch to the nearest road, town center, and river is calculated using proximity analysis in practical GIS. The changing patch is spatially overlaid with various pre-input planning maps and protection zoning maps to generate Boolean indicators, such as: whether it overlaps with the urban planning area, whether it is located within the ecological protection red line, and whether it is located in a high-risk area for geological hazards.

[0071] Finally, all calculation results are combined into a feature vector, which is then input into a pre-trained causal analysis model (such as a decision tree or random forest) so that the causal analysis model outputs the causal type as "change caused by human activities" or "change caused by natural factors." The analysis and comparison system pre-establishes a causal knowledge base, specifically divided into a human activity database and a natural factor database. The human activity database stores feature values ​​corresponding to the aforementioned multi-source features, indicating changes in the changed map features caused by human intervention, such as "aspect ratio of 2:1," "boundary curvature value less than 1 and the difference from 1 is less than a preset difference," "change rate less than a preset threshold," and "located within the planning area = yes." Similarly, the natural factor database stores feature values ​​corresponding to the aforementioned multi-source features, indicating changes in the changed map features caused by natural factors, such as "1 < boundary curvature < 2, and the difference between the boundary curvature and 2 is less than a preset difference," and "change rate greater than a preset threshold."

[0072] The causal analysis model is used to match the feature vector obtained from the current combination with the contents stored in the natural factor library. It should be noted that each feature value in the human activity library and each feature value in the natural factor library has a pre-defined feature weight value (a fixed numerical value) to quantify the importance of different feature values. That is, the larger the feature weight value, the higher the importance and the more critical the feature value; conversely, the lower the weight value, the less important it is. For example, "boundary tortuosity value" is a key feature in the human activity library. During the matching process, the causal analysis model is used to find the feature values ​​in the human factor library that match the feature vector (hereinafter referred to as the first feature value); and the feature values ​​in the natural factor library that match the feature vector (hereinafter referred to as the second feature value). Then, all first feature values ​​are summed, and all second feature values ​​are summed. If the sum of the first feature values ​​is not less than the sum of the second feature values, then... The output cause type is "change caused by human activities," otherwise it is "change caused by natural factors." Furthermore, the confidence level can be determined based on the number of matching feature values. For example, several ranges of values ​​and their corresponding confidence levels can be preset (confidence levels can be expressed as probability values). By determining the range of values ​​corresponding to the cause type and the number of currently matching feature values ​​(i.e., if the cause type is change caused by human activities, the matching feature value is the first feature value; if the cause type is change caused by natural factors, the matching feature value is the second feature value), the corresponding confidence level can be determined. The confidence level and cause type are then output and displayed together for investigators to know.

[0073] Optionally, the analysis and comparison method may also include the following steps: For areas where changes have occurred, assess the change in land value before and after the changes. Land value includes at least economic and ecological value. Based on the analysis results of changes in land value and their causes, and according to the pre-set suggestion rule base, suggestions for optimizing land use are generated and output.

[0074] In implementation, the analysis and comparison system is used to calculate the change in land value before and after the change in the area where changes have occurred. The change in land value = land value before the change - land value after the change. The land value specifically includes economic value and ecological value. The method for calculating the economic value is as follows: First, the benchmark land price (the average price of the region published by the government, which can be stored in advance for different locations) is determined based on the location of the area that has changed (such as commercial center, residential area, industrial area, agricultural area). Here, it is assumed that the location of different areas in the GIS map is also stored in the background geographic database.

[0075] Then, based on the preset first correction factor, the benchmark land price is adjusted. The first correction factor includes: accessibility (distance from main roads, highway entrances, and subway stations. The closer the distance, the higher the corresponding correction coefficient K1, which can range from 0.7 to 1.5), infrastructure completeness (whether there is complete infrastructure such as water supply, power supply, communication, schools, and hospitals in the surrounding area. The more infrastructure there is, the higher the completeness and the higher the corresponding correction coefficient K2, which can range from 0.8 to 1.4), and planning conditions (plot ratio, and the higher the allowed plot ratio, the higher the corresponding correction coefficient K3, which can range from 0.9 to 2). A corresponding relationship table is set up in advance for each first correction factor. The relationship table stores different numerical ranges of the corresponding first correction factor (such as distance range, infrastructure quantity range, plot ratio range) and the corresponding correction coefficient K. It should also be noted that the default GIS map pre-marks transportation facilities (i.e., main roads, highways, subway stations) and infrastructure so that the analysis and comparison system can calculate the distance between the changed areas and transportation facilities, and the amount of infrastructure covered by the changed areas.

[0076] Economic value = Benchmark land price × K1 × K2 × K3 × Area of ​​the area where changes have occurred; where K1 is the correction coefficient corresponding to traffic accessibility, K2 is the correction coefficient corresponding to infrastructure completeness, and K3 is the correction coefficient corresponding to planning conditions.

[0077] The method for calculating ecological value is as follows: The system pre-sets basic ecological value equivalents for different land types (such as forests, wetlands, cultivated land, and grasslands). For example, the economic value of the annual natural grain yield of 1 hectare of standard farmland can be defined as 1 equivalent. The basic ecological value equivalents are then corrected according to a pre-set second correction factor S to obtain the ecological value. The second correction factor includes: ecological function importance (i.e., determining whether it is located in a water conservation area, a key water and soil conservation area, or a biodiversity conservation hotspot; the default background geographic database also stores water conservation areas, key water and soil conservation areas, and biodiversity conservation hotspots pre-delineated by investigators in the GIS map. Therefore, by analyzing whether the changed area overlaps with the aforementioned areas, it can be determined whether it is within the aforementioned areas. The larger the area of ​​the overlapping area, the higher the corresponding correction coefficient S1. For example, the correction coefficient S1 can range from 1.0 to 3.0), and ecological vulnerability (determining the distance between the changed area and the desert). If the changed area overlaps with the desert area, it indicates the highest vulnerability. If there is no overlap, the closer the changed area is to the desert area, the higher the vulnerability. Higher vulnerability means greater cost for ecological function loss and restoration once damaged, thus increasing the existing, undisturbed ecological value and consequently, the higher the corresponding correction coefficient S2 (which can range from 1.0 to 3.0). In other words, the correction coefficient S2 is positively correlated with ecological vulnerability. Similarly, the default GIS map pre-marks the desert area. Furthermore, the analysis and comparison system pre-stores a correspondence table for each second correction factor, containing different judgment results for each second correction factor and the correction coefficient for each numerical range (such as the area of ​​the overlapping region and the distance between the changed area and the desert area).

[0078] The final calculated ecological value = basic ecological value equivalent × s1 × s2 × area of ​​the changed region; where s1 is the correction coefficient corresponding to the importance of ecological function and s2 is the correction coefficient corresponding to the ecological vulnerability.

[0079] Ultimately, based on pre-set weights for economic and ecological value, the final land value is obtained through a weighted summation: Economic Value × w1 + Ecological Value × w2; where w1 and w2 are pre-set weight values, and the sum of the weights of w1 and w2 is 1.0. Furthermore, an interactive interface is provided for investigators to adjust the specific values ​​of w1 and w2.

[0080] In summary, the land value before and after the change is calculated using the above method, and then the change in land value is calculated. Finally, the change in land value is output. If the land value after the change is less than the land value before the change, the corresponding change in land value is negative, so that investigators can intuitively understand the change in land value before and after the change (i.e., increased / decreased).

[0081] In addition, the analysis and comparison system has a built-in suggestion rule base, which contains several suggestion trigger conditions and their corresponding land use suggestions, such as: The recommended trigger condition one is: "The cause type is 'change due to human activities' and the change in land value is < 0." Furthermore, the land use recommendation corresponding to trigger condition one includes several sub-recommendations, each with its own sub-conditions. For example, the sub-condition "The corresponding ecological value after the change is < the ecological value before the change, and the area where the change occurred is located in an ecologically sensitive area (i.e., a water conservation area, a key area for soil and water conservation, or a biodiversity conservation hotspot)" corresponds to the sub-recommendation: "This construction activity may have a negative impact on the ecology; it is recommended to review its planning permit and consider ecological restoration or adjusting the project layout." The second suggested trigger condition is: "The cause type is 'change caused by natural factors' and the change in land value is <0". Furthermore, the land use recommendation corresponding to trigger condition two includes several sub-recommendations, each with its own sub-condition. For example, the sub-condition "The area where the change occurred is located in a high-risk area for geological disasters" corresponds to the sub-recommendation: "This area is a high-risk area for geological disasters; reconstruction on the original site is not recommended. It is recommended to include it in the disaster prevention and control plan and implement relocation or engineering remediation."

[0082] The analysis and comparison system is also used to identify whether there are any suggested triggering conditions based on the causal type and land value change of the area that has changed. If so, it further determines whether the corresponding sub-conditions are met. If there are sub-conditions that are met, the sub-suggestions of the corresponding sub-conditions are output as land use suggestions for investigators to know.

[0083] Optionally, the analysis and comparison method may also include the following steps: Acquire all recognition results generated in historical periods, calculate the stability index of each patch, and define the regions where patches with stability indices higher than a preset stability threshold are located as stable regions. Stable regions that are spatially adjacent and have the same land use type are merged to form an independent comparison unit; When any change is detected in the alignment unit, the alignment unit is split and restored to the region before merging; The following steps are included prior to S1031: If the geographical area covered by the surveyed object contains comparison units, then for each comparison unit, all the map patches in the area where each comparison unit is located will be merged to form a complete map patch.

[0084] In implementation, the analysis and comparison system is used to collect the identification results within a specified period at regular intervals and establish the correspondence between the identification results and their corresponding survey subjects. Then, for all survey subjects collected in the current specified period, it determines all the identification results corresponding to each survey subject in the current specified period and before, clusters all the identification results corresponding to the same survey subject into a set, and then analyzes the stability index of each patch in the region contained by each survey subject.

[0085] The stability index is a weighted sum of boundary change frequency and land use stability. Boundary change frequency characterizes the changes in map patch boundaries, calculating the proportion of times a map patch boundary changed out of all historical surveys. Correspondingly, several boundary change frequency ranges and corresponding first stability scores can be preset. This correspondence determines the first stability score for the current boundary change frequency within its specified range. It should be noted that the lower the boundary change frequency range, the smaller the corresponding first stability score. Furthermore, second stability scores are preset for different land use types. By finding the corresponding relationships, the second stability score for the land use type corresponding to the map patch can be obtained. The second stability score quantifies land use stability; a higher second stability score indicates stronger stability for the corresponding land use type. By using the preset weights z1 and z2 (z1 and z2 are positive numbers) for the relevant boundary change frequency and land type stability, as well as the first stability score and the second stability score obtained above, the final stability index is obtained as follows: first stability score × z1 + second stability score × z2.

[0086] If any of the map features contained in the surveyed object have a stability index higher than the preset stability threshold, then the area where that map feature is located within the area covered by the surveyed object is considered a stable area. If there are spatially adjacent stable areas with the same land use type, then the stable areas are merged to form an independent area (i.e., a comparison unit). When the survey instruction for the surveyed object is received again, the map features corresponding to the comparison unit are merged to form a new map feature. Then, the processed data and the specified benchmark data are used to compare the map features (including spatial comparison, attribute comparison, and temporal comparison). That is, the map features corresponding to the comparison unit are treated as a whole, and the data describing the whole map feature is searched from the processed data and the specified benchmark data. Spatial comparison, attribute comparison, and temporal comparison are then performed to monitor whether the comparison unit has changed. If inconsistencies occur during the comparison of the overall patch corresponding to the comparison unit, i.e., the overall patch is identified as a changed patch in the identification result, then the overall patch is unmerged, all patches contained before the merger are restored, and the aforementioned comparison operation is re-executed on the restored patches extracted from the overall patch to analyze which patches before the merger the specific changed areas originated from.

[0087] This application also discloses a land survey analysis and comparison system. (Refer to...) Figure 2 ,include: The multi-source data acquisition module 201 is used to receive survey and analysis instructions and acquire multi-source data of the surveyed object, wherein the multi-source data includes at least remote sensing images and land use attribute information. Data preprocessing module 202 is used to preprocess multi-source data to obtain processed data; The land change analysis module 203 is used to compare the processed data with the specified baseline data based on the preset comparison logic, generate and output the comparison results for the surveyors to know. The designated baseline data refers to multi-source data of the surveyed subjects that are identical to the processed data; and the comparison results are used to characterize the deviation between the processed data and the designated baseline data, with the deviation used to characterize the changes in the surveyed subjects.

[0088] Optionally, the land change analysis module 203 is also used to spatially compare and correlate the processed data with the map features contained in the specified benchmark data. If correlated map features are determined, the overlap and positional offset of the correlated map features are calculated to generate a spatial consistency judgment result. Here, correlated map features refer to map features that are spatially correlated with the map features contained in the processed data and the map features contained in the specified benchmark data. Based on the correlation relationship in the spatial consistency judgment result, the fuzzy comprehensive evaluation method is used to calculate the similarity of the attribute information between the correlated map features to generate an attribute consistency comparison result and realize attribute comparison. It is also used to identify the areas where the surveyed object has changed over time based on the spatial consistency judgment result and the attribute consistency judgment result, and output the identification result for the surveyors to know.

[0089] Optionally, it also includes a future change prediction module, which is used to analyze the identification results output in historical periods, mine multi-source data of the changed area before and after the change, and generate bundled change events based on a preset bundled change event mining algorithm. The bundled change events contain multi-source datasets obtained from two different monitoring times, and are limited to: the surveyed object described by the multi-source dataset of the earlier monitoring time is likely to change in the future, and the multi-source data of the surveyed object obtained after the change will be consistent with the multi-source dataset of the later monitoring time. It is also used to obtain the processed data of the current survey subjects, match the processed data with the bundled change events, and when the processed data contains the target data and successfully matches the multi-source dataset of the earlier monitoring time contained in the bundled change event, the multi-source dataset of the later monitoring time in the bundled change event that successfully matches the target data is used as the future change prediction result of the patch corresponding to the target data, and the future change prediction result is output for the investigators to know.

[0090] Optionally, the future change prediction module is also used to consider a successful match when the processed data contains target data, is consistent with the multi-source data of the earlier time period included in the bundled change event, and the patch described by the target data meets the corresponding bundling conditions; when outputting the future change prediction result, it also outputs the change probability value represented by the corresponding bundling strength.

[0091] Optionally, it also includes a change cause analysis module, which is used to extract multi-source features from the processed data corresponding to the changed area whenever an identification result is generated, and analyze the multi-source features based on a preset cause analysis model and output the cause type; wherein the cause type includes changes caused by human activities and changes caused by natural factors.

[0092] Optionally, it also includes a value change assessment module, which is used to assess the change in land value before and after the change for areas where changes have occurred. The land value includes at least economic value and ecological value. Based on the land value change and the analysis results of its causes, it generates and outputs land optimization and utilization suggestions according to a preset suggestion rule base.

[0093] Optionally, it also includes a patch adjustment module, which is used to acquire all recognition results generated in historical periods, calculate the stability index of each patch, and identify the area where the patch with the stability index is higher than the preset stability threshold is located as a stable area; merge spatially adjacent stable areas with the same land use type to form an independent comparison unit; when any change is detected in any comparison unit, the comparison unit is split and restored to the area before merging; The land change analysis module 203 is also used to merge all the map patches in the area where each comparison unit is located into a complete map patch if the geographical area covered by the surveyed object contains comparison units.

[0094] This application also discloses a land survey analysis and comparison device, which includes a memory and a processor. The memory stores a computer program that can be loaded by the processor and executed as described above for the land survey analysis and comparison method.

[0095] This application also discloses a computer-readable storage medium that stores a computer program that can be loaded by a processor and executed as described above for land survey analysis and comparison. The computer-readable storage medium includes, for example, various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0096] It should be noted that in this paper, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations.

[0097] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit the scope of protection of the application. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on these embodiments, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

Claims

1. A method of land survey analysis comparison, characterized by, The method comprises the steps of: receiving survey analysis instructions, obtaining multi-source data of a surveyed object, wherein the multi-source data at least includes remote sensing images and land use attribute information; preprocessing the multi-source data to obtain processed data; comparing the processed data with specified reference data based on a preset comparison logic, generating and outputting a comparison result for survey personnel to know; wherein the specified reference data refers to multi-source data of the same surveyed object as described by the processed data; and the comparison result at least represents the deviation of the processed data from the specified reference data, which represents the change of the surveyed object.

2. The land survey analysis comparison method of claim 1, wherein, The processed data and the specified reference data refer to data obtained by processing multi-source data of the surveyed object obtained at different times; comparing the processed data with the specified reference data based on a preset comparison logic, generating and outputting a comparison result for survey personnel to know, comprising: spatially comparing and associating the polygons contained in the processed data and the specified reference data, if an associated polygon is determined, calculating the overlap and position offset of the associated polygon to generate a spatial consistency determination result; wherein the associated polygon refers to a polygon in the processed data that is spatially associated with a polygon in the specified reference data; based on the association relationship in the spatial consistency determination result, using fuzzy comprehensive evaluation method to calculate the similarity of the attribute information between the associated polygons to generate an attribute consistency comparison result, realizing attribute comparison; based on the spatial consistency determination result and the attribute consistency determination result, identifying the area of the surveyed object that changes over time, outputting the identification result for survey personnel to know.

3. The land investigation analysis comparison method of claim 2, wherein, The method further comprises: analyzing the identification result output in the historical period, mining the multi-source data of the changed area before and after the change, generating a bundled change event based on a preset bundled change event mining algorithm, the bundled change event contains two multi-source data sets obtained at different monitoring times, and is limited: the surveyed object described by the multi-source data set at the earlier monitoring time will change in the future with a high probability, and the multi-source data of the surveyed object obtained after the change will be consistent with the multi-source data set at the later monitoring time; obtaining the processed data of the current surveyed object, matching the processed data with the bundled change event, when there is target data in the processed data that matches the multi-source data set at the earlier monitoring time contained in the bundled change event successfully, taking the multi-source data set at the later monitoring time in the bundled change event that matches the target data successfully as the future change prediction result of the polygon corresponding to the target data, outputting the future change prediction result for survey personnel to know.

4. The land investigation analysis comparison method of claim 3, wherein, The bundling change event further comprises a bundling condition, and a bundling strength corresponding to the bundling condition, the bundling strength being used to represent a probability of a change of the investigated object in the future and a consistency of multi-source data of the investigated object after the change with multi-source data of a later time period. The matching success comprises: When the target data in the processed data is consistent with the multi-source data of the earlier time period in the bundling change event, and a corresponding plot area described by the target data satisfies the corresponding bundling condition, it is considered that the matching is successful. The method further comprises: When outputting the future change prediction result, a change probability value represented by the corresponding bundling strength is simultaneously outputted.

5. The land investigation analysis comparison method of claim 2, wherein, The method further comprises: Whenever the identification result is generated, multi-source features are extracted from the processed data corresponding to the changed area, and a cause type is outputted by analyzing the multi-source features based on a preset cause analysis model, wherein the cause type comprises a change caused by human activities and a change caused by natural factors.

6. The land survey analysis comparison method of claim 5, wherein, The method further comprises: For the changed area, a land value change amount before and after the change is evaluated, and the land value at least comprises an economic value and an ecological value; Based on the land value change amount and the cause analysis result, a land optimization utilization suggestion is generated and outputted according to a preset suggestion rule library.

7. The land investigation analysis comparison method of claim 2, wherein, The method further comprises: All identification results generated in a historical period are acquired, a stability index of each plot area is calculated, and an area in which a plot area with a stability index higher than a preset stability threshold value is located is regarded as a stable area. Spatially adjacent and same-class stable areas are merged to form an independent comparison unit. When a change of any comparison unit is monitored, the comparison unit is split and restored to the area before the merging. The spatial comparison and correlation of the processed data and the plot areas included in the specified reference data further comprises: If the comparison unit is included in the geographical area covered by the investigated object, all plot areas in the area of each comparison unit are merged to form a complete plot area.

8. A land survey analysis comparison system, characterized by, The method further comprises: A multi-source data acquisition module (201) is configured to receive an investigation and analysis instruction, and acquire multi-source data of an investigated object, wherein the multi-source data at least comprises remote sensing images and land use attribute information. A data preprocessing module (202) is configured to preprocess the multi-source data to obtain processed data. A land change analysis module (203) is configured to compare the processed data and specified reference data based on a preset comparison logic, generate and output a comparison result, so that an investigator can know the comparison result. The specified reference data refers to multi-source data of the same investigated object as the processed data, and the comparison result at least represents a deviation between the processed data and the specified reference data, and the deviation is used to represent a change of the investigated object.

9. A land survey analysis comparison device, characterized by, A computer program product comprising a memory and a processor, said memory having stored thereon a computer program loadable and executable by the processor to perform the method of any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, A computer program product comprising a memory and a processor, said memory having stored thereon a computer program loadable and executable by the processor to perform the method of any one of claims 1 to 7.

Images