Cadastral surveying and mapping optimization method and system based on low-altitude remote sensing

Abstract

The application discloses a cadastral surveying and mapping optimization method and system based on low-altitude remote sensing, relates to the related technical field of surveying and mapping optimization, and comprises the following steps: collecting satellite image data of a target survey area, identifying building density distribution, vegetation coverage density distribution and terrain elevation change rate; selecting the type of an unmanned aerial vehicle and the type of a sensor in a preset resource library, and then performing flight path planning based on flight altitude, heading overlap rate and lateral overlap rate; controlling the unmanned aerial vehicle to fly and collect low-altitude remote sensing data sets to establish a real scene three-dimensional model; calling a cadastral knowledge graph to perform reasoning, checking and conflict detection, and generating a target cadastral map. The technical problems that the existing cadastral surveying and mapping operation is difficult to perform accurate data collection in a load environment, and the accuracy and reliability of cadastral feature identification and the efficiency of cadastral surveying and mapping operation are low are solved, and the technical effects of improving the efficiency of cadastral surveying and mapping operation and enhancing the accuracy and reliability of cadastral feature identification in a complex environment are achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of surveying and mapping optimization, and particularly relates to a cadastral surveying and mapping optimization method and system based on low-altitude remote sensing. BACKGROUND

[0002] Traditional cadastral surveying and mapping mainly relies on total station, GPS-RTK and other ground surveying methods, which have limited operation range and are easily affected by factors such as terrain, vegetation and building obstruction. In addition, in areas with dense buildings, vegetation coverage or large terrain undulations, the data collection efficiency is significantly reduced, and it is difficult to meet the demand for high efficiency, real-time and comprehensive data of modern cadastral management. Unmanned aerial vehicles can carry multiple sensors for low-altitude data collection, but how to optimize the surveying and mapping path according to the specific characteristics of the survey area such as building density, vegetation coverage and terrain undulation becomes a key problem. In addition, it is still difficult to automatically extract cadastral elements such as boundary points, boundary lines and plot ranges from three-dimensional models, which may include model missing caused by ambiguous boundaries of ground objects, insufficient semantic information, etc., affecting the accuracy and reliability of cadastral surveying and mapping.

[0003] Therefore, in the related art, the technical problem of low accuracy and reliability of cadastral element identification and low efficiency of cadastral surveying and mapping operation exists because cadastral surveying and mapping operation cannot accurately collect data in a complex environment. SUMMARY

[0004] The present application provides a cadastral surveying and mapping optimization method and system based on low-altitude remote sensing, which solves the technical problem of low accuracy and reliability of cadastral element identification and low efficiency of cadastral surveying and mapping operation because cadastral surveying and mapping operation cannot accurately collect data in a complex environment, and achieves the technical effect of improving the efficiency of cadastral surveying and mapping operation and enhancing the accuracy and reliability of cadastral element identification in a complex environment.

[0005] The present application provides a cadastral surveying and mapping optimization method based on low-altitude remote sensing, which comprises: collecting satellite image data of a target survey area, identifying building density distribution, vegetation coverage density distribution and terrain elevation change rate of the target survey area, and generating an identification feature set; based on the identification feature set, selecting a type of unmanned aerial vehicle and a type of sensor in a pre-set resource library, and then performing flight path planning based on flight altitude, heading overlap rate and lateral overlap rate based on the selection result, and establishing an unmanned aerial vehicle-sensor-path mapping; controlling the unmanned aerial vehicle to fly based on the unmanned aerial vehicle-sensor-path mapping, collecting low-altitude remote sensing data set to establish a real scene three-dimensional model of the target survey area; calling a cadastral knowledge graph, reasoning, verifying and conflict detecting the real scene three-dimensional model, completing the correction of cadastral elements, and generating a target cadastral map.

[0006] In a possible implementation, the cadastral mapping optimization method based on low-altitude remote sensing further performs the following processing: identifying building contours from the satellite image data through semantic segmentation, calculating the number or coverage of buildings in a unit area to obtain a building density distribution; distinguishing vegetation and non-vegetation by calculating a vegetation index based on the satellite image data, and evaluating the density of the vegetation to obtain a vegetation coverage density distribution; performing spatial analysis on the satellite image data, calculating the difference between the elevation of each cell and the surrounding cells, and generating a slope map reflecting the change rate of the terrain elevation difference; and generating the set of identified features based on the building density distribution, the vegetation coverage density distribution, and the slope map.

[0007] In a possible implementation, the cadastral mapping optimization method based on low-altitude remote sensing further performs the following processing: performing multi-level consistent clustering based on the building density distribution, the vegetation coverage density distribution, and the change rate of the terrain elevation difference on the target survey area according to the set of identified features to establish various sub-areas; performing surveying and mapping accuracy optimization under corresponding environmental conditions based on the preset resource library for the various sub-areas, and completing the selection of the type of unmanned aerial vehicle and the type of sensor to obtain various selection results corresponding to the various sub-areas.

[0008] In a possible implementation, the cadastral mapping optimization method based on low-altitude remote sensing further performs the following processing: collecting a set of historical surveying and mapping records based on the preset resource library, each historical surveying and mapping record corresponding to a set of resource combinations, and each set of resource combinations including at least one type of unmanned aerial vehicle and one type of sensor; for the various sub-areas, extracting, from each set of reference surveying and mapping records, a resource combination with the highest cadastral modeling accuracy, to obtain the various selection results, where the set of reference surveying and mapping records match the regional environment similarity to a preset similarity threshold.

[0009] In a possible implementation, the cadastral mapping optimization method based on low-altitude remote sensing further performs the following processing: performing three-dimensional modeling complexity identification based on the building density and the vegetation coverage density of the various sub-areas to generate various modeling complexity coefficients; configuring a basic forward overlap rate threshold and a basic lateral overlap rate threshold based on a preset flat complexity threshold; calculating, based on the preset flat complexity threshold, an enhanced weight of each sub-area from the various modeling complexity coefficients, and performing weight enhancement processing on the basic forward overlap rate threshold and the basic lateral overlap rate threshold to generate a forward overlap rate of each sub-area and a lateral overlap rate of each sub-area; configuring a flight altitude based on the distance constraint of the sensor in each selection result, and planning a flight path based on the forward overlap rate of each sub-area and the lateral overlap rate of each sub-area to generate a planned path of each sub-area; and constructing a corresponding relationship between the various selection results and the planned paths of the various sub-areas to generate the unmanned aerial vehicle-sensor-path mapping.

[0010] In a possible implementation, the cadastral mapping optimization method based on low-altitude remote sensing further performs the following processing: the base heading overlap rate threshold and the base lateral overlap rate threshold are lower limits of a heading overlap rate and a lateral overlap rate of a preset flat area when three-dimensional modeling accuracy meets a preset accuracy threshold, and a three-dimensional modeling complexity of the preset flat area is less than a preset flat complexity threshold.

[0011] In a possible implementation, the cadastral mapping optimization method based on low-altitude remote sensing further performs the following processing: the enhancement weight of each sub-area is generated by calculating an enhancement ratio of each modeling complexity coefficient relative to the preset flat complexity threshold.

[0012] In a possible implementation, the cadastral mapping optimization method based on low-altitude remote sensing further performs the following processing: cadastral elements are extracted from the real-scene three-dimensional model; the extracted cadastral elements are compared with a predefined property boundary rule in the cadastral knowledge graph to determine whether there is ambiguity, and if so, an ambiguous area is labeled in the real-scene three-dimensional model to generate a prompt information; the prompt information is sent to an artificial verification end through a human-computer interaction interface and a correction instruction returned by the artificial verification end is received; cadastral elements are corrected according to the correction instruction, and a target cadastral map is generated based on the corrected real-scene three-dimensional model.

[0013] In a possible implementation, the cadastral mapping optimization method based on low-altitude remote sensing further performs the following processing: the cadastral knowledge graph is obtained based on collected multi-source cadastral rule data through identification and structured expression of cadastral entities and relationships between the cadastral entities.

[0014] The application further provides a cadastral mapping optimization system based on low-altitude remote sensing, which comprises: a feature recognition module, configured to collect satellite image data of a target survey area, recognize building density distribution, vegetation coverage density distribution, and terrain elevation change rate of the target survey area, and generate a recognition feature set; a path planning module, configured to perform selection of a type of unmanned aerial vehicle and a type of sensor in a preset resource library based on the recognition feature set, and perform flight path planning based on flight altitude, heading overlap rate and lateral overlap rate based on a selection result, and establish an unmanned aerial vehicle-sensor-path mapping; a model establishment module, configured to control an unmanned aerial vehicle to fly based on the unmanned aerial vehicle-sensor-path mapping, collect low-altitude remote sensing data sets, and establish a real-scene three-dimensional model of the target survey area; and a cadastral map generation module, configured to call a cadastral knowledge graph, perform reasoning, verification and conflict detection on the real-scene three-dimensional model, complete correction of cadastral elements, and generate a target cadastral map.

[0015] The cadastral surveying and mapping optimization method and system based on low-altitude remote sensing provided in the application can collect satellite image data of a target survey area, identify building density distribution, vegetation coverage density distribution, and terrain elevation change rate, select a type of unmanned aerial vehicle and a type of sensor in a preset resource library, then perform flight path planning based on flight altitude, heading overlap rate, and lateral overlap rate, control the unmanned aerial vehicle to fly and collect low-altitude remote sensing data sets to establish a real scene three-dimensional model, call a cadastral knowledge graph to perform reasoning, verification, and conflict detection, and generate a target cadastral map. The technical problems that the existing cadastral surveying and mapping operation cannot accurately collect data in a complex environment, resulting in low accuracy and reliability of cadastral feature identification and low efficiency of cadastral surveying and mapping operation are solved, and the technical effects of improving the efficiency of cadastral surveying and mapping operation and enhancing the accuracy and reliability of cadastral feature identification in a complex environment are achieved. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the drawings of the embodiments of the present disclosure will be briefly introduced below. In the present application, a flowchart is used to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the foregoing or the following operations are not necessarily performed in sequence. On the contrary, various steps can be processed in reverse order or simultaneously as needed. At the same time, other operations can be added to these processes, or one or more steps of operations can be removed from these processes.

[0017] Figure 1 The flowchart of the cadastral surveying and mapping optimization method based on low-altitude remote sensing provided by the embodiments of the present application is shown.

[0018] Figure 2 The structure schematic diagram of the cadastral surveying and mapping optimization system based on low-altitude remote sensing provided by the embodiments of the present application is shown.

[0019] Explanation of reference signs: feature identification module 10, path planning module 20, model establishment module 30, cadastral map generation module 40. DETAILED DESCRIPTION

[0020] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the specific embodiments, structures, features and effects according to the present application will be described in detail below with reference to the drawings and preferred embodiments.

[0021] The embodiments of the present application provide a cadastral surveying and mapping optimization method based on low-altitude remote sensing, as shown in Figure 1 The method comprises the following steps:

[0022] In step S100, satellite image data of a target survey area is collected, building density distribution, vegetation coverage density distribution, and terrain elevation change rate of the target survey area are identified, and an identification feature set is generated.

[0023] The step S100 further comprises a step S110 of identifying a building contour from the satellite image data by semantic segmentation, calculating the number of buildings or the coverage rate in a unit area, and obtaining a building density distribution; a step S120 of distinguishing vegetation and non-vegetation by calculating a vegetation index based on the satellite image data, and evaluating the density of the vegetation to obtain a vegetation coverage density distribution; a step S130 of performing spatial analysis on the satellite image data, calculating the difference between the elevation of each cell and the surrounding cells, and generating a slope map reflecting the change rate of the terrain elevation difference; and a step S140 of generating the set of identification features from the building density distribution, the vegetation coverage density distribution, and the slope map.

[0024] Preferably, according to the requirements of the surveying and mapping task, a remote sensing satellite equipped with an optical sensor is used to image the target survey area to obtain satellite image data, including a panchromatic band and a multi-spectral band, which are used to identify the fine contour features and attributes of ground objects, and then identify the building density distribution, vegetation coverage density distribution, and terrain elevation difference change rate of the target survey area. Specifically, semantic segmentation is performed on the satellite image data, i.e., a pre-trained convolutional neural network is used for pixel-level classification processing of the satellite image data to identify the building contour, including labeling each pixel point in the satellite image with a label indicating a building or a non-building, and then outputting a binary mask image, where white pixels represent buildings and black pixels represent non-buildings. Then, a fixed-size grid cell, such as 100m x 100m, is defined on the generated building mask image, and the number of building contours is identified and counted in each grid cell to obtain the number of buildings in a unit area, and the ratio of the total number of pixels classified as buildings in each grid cell to the total number of pixels in the grid cell is calculated as the coverage rate. Then, the number or coverage rate of each grid cell is assigned to the cell to generate a building density distribution covering the entire target survey area, where each grid has a corresponding density value. High-density areas represent severe occlusion and require oblique photography to obtain building facade information, while low-density areas only require vertical photography to meet the requirements.

[0025] Preferably, the vegetation index is calculated using the multi-spectral band of the satellite image through the normalized difference vegetation index formula, i.e., the ratio of the difference between the reflectance of the near-infrared band and the reflectance of the red band to the sum of the reflectance of the near-infrared band and the reflectance of the red band. Then, the image is binarized by setting a vegetation index threshold to distinguish vegetation and non-vegetation, where the vegetation index of the vegetation area is close to 1, and the vegetation index of the non-vegetation area is very low or even negative. The vegetation index value directly reflects the chlorophyll content or the density of the vegetation, and the higher the value, the denser the vegetation. Similarly, the average vegetation index value of all pixels in each grid cell is calculated and assigned to each grid cell to generate a vegetation coverage density distribution, which simultaneously reflects the presence or absence of vegetation and the density of the vegetation.

[0026] Preferably, the satellite image data is subjected to spatial analysis to calculate the slope, i.e. to calculate the difference in elevation of each cell and the surrounding cells, including traversing each pixel in the digital elevation model data and calculating the maximum rate of change between the elevation value of the pixel and the elevation values of the eight surrounding cells, specifically, by calculating the elevation change derivatives in the east-west direction and the north-south direction, and then combining the two to calculate the slope value of the point, and finally outputting a slope map reflecting the rate of change of the terrain elevation, wherein the value of each pixel represents the degree of inclination of the ground at that location, the larger the slope value, the more intense the change in terrain elevation, and the smaller the slope value, the flatter the terrain. Finally, the building density distribution, vegetation coverage density distribution and slope map are spatially registered to ensure that each grid cell is aligned, and a set of recognition features, i.e. a set of registered raster layers, is generated, so that for any geographic location within the target detection area, the building density value, vegetation coverage density value and slope value of the point can be queried simultaneously.

[0027] Step S200, based on the recognition features, selecting the type of unmanned aerial vehicle and sensor in the pre-set resource library, and then based on the selection results, performing flight path planning based on flight altitude, heading overlap rate and lateral overlap rate, and establishing an unmanned aerial vehicle-sensor-path mapping.

[0028] Step S200 further includes step S210, performing multi-level consistent clustering based on building density distribution, vegetation coverage density distribution and terrain elevation change rate on the target detection area according to the recognition features, and establishing various sub-areas; and step S220, based on the pre-set resource library, performing surveying and mapping accuracy optimization under corresponding environmental conditions for the various sub-areas, completing the selection of the type of unmanned aerial vehicle and sensor, and obtaining the various selection results corresponding to the various sub-areas.

[0029] Preferably, the K-Means or DBSCAN unsupervised clustering algorithm is used according to the recognition features to perform multi-level consistent clustering based on building density distribution, vegetation coverage density distribution and terrain elevation change rate on the target detection area, i.e. to classify pixel points close in feature space into the same class, specifically, considering the building density value, vegetation coverage density value and slope value simultaneously, and then generating a clustering distribution to ensure that the clustering division result comprehensively reflects the comprehensive influence of all environmental factors, wherein each pixel is assigned a class label, and all pixels with the same label and connected in space together form a sub-area, and finally various sub-areas are established, such as urban dense sub-area A with high building density, low vegetation density and low slope; forest and hilly sub-area B with low building density, high vegetation density and high slope; and plain farmland sub-area C with low building density, medium vegetation density and low slope.

[0030] Preferably, the preset resource library includes a historical mapping record table, each record of which at least includes building density, vegetation density, slope, etc. of a historical task area, a type of a used unmanned aerial vehicle and a type of a used sensor, and accuracy evaluation results of a finally generated real scene three-dimensional model, such as planar mean error and height mean error; and the mapping accuracy optimization under corresponding environment conditions of each sub-area is performed based on the preset resource library, that is, each sub-area is traversed, average environment characteristics such as average building density, average vegetation density, and average slope are used for searching and matching in the historical mapping record table, all records with similar historical environment characteristics and the current sub-area are identified and determined by calculating Euclidean distance or cosine similarity, the record with the highest result accuracy in all matched historical records is identified as the optimal historical record, and the resource combination of the type of the unmanned aerial vehicle and the type of the sensor included in the optimal historical record is determined as the selection result of the current sub-area, and finally the corresponding selection results of each sub-area are obtained, so that appropriate hardware resources are ensured to realize the optimization of overall efficiency under the premise of ensuring accuracy.

[0031] Further, step S220 further includes step S221 of collecting a historical mapping record set based on the preset resource library, each historical mapping record of which corresponds to a group of resource combinations, and each group of resource combinations at least includes a type of an unmanned aerial vehicle and a type of a sensor; step S222 of matching each reference mapping record set of the historical mapping record set with a region environment similarity reaching a preset similarity threshold for each sub-area; and step S223 of extracting a resource combination with the highest cadastral modeling accuracy from each reference mapping record set to obtain the selection result.

[0032] Preferably, the historical mapping record set of the preset resource library is obtained, each historical mapping record of which corresponds to a group of resource combinations, for clearly executing hardware resources used in the historical task, at least including a type of an unmanned aerial vehicle and a type of a sensor; further including average environment characteristics such as average building density, average vegetation density, and average slope, for recording environment parameters of the region where the historical task is located; and further including cadastral modeling accuracy, for measuring the accuracy of a finally generated real scene three-dimensional model on cadastral elements such as boundary points and boundary lines, such as planar mean error and height mean error, and the lower the error value, the higher the accuracy.

[0033] Preferably, for each sub-area, the Euclidean distance or cosine similarity is used to calculate the similarity between its environment feature vector and the area environment feature vector of each record in the historical record library. Specifically, a normalized multi-dimensional feature vector is constructed for each historical surveying task and the current sub-area, including building density, vegetation density, and terrain slope, where the value of each dimension is the result of maximum and minimum normalization, ranging from [0, 1], a preset building density feature weight of 0.5, a vegetation density feature weight of 0.3, and a terrain slope feature weight of 0.2 are used, and the feature weights can be updated according to experience optimization, the weighted cosine similarity formula is used to calculate the similarity between the two feature vectors, the preset similarity threshold is set to 0.85 according to experimental data and historical cadastral surveying task data, and can be adjusted, all historical records with a similarity calculation result reaching or exceeding the preset similarity threshold, i.e., a feature similarity greater than or equal to 0.85, are considered to have similar environments, and corresponding reference surveying record sets are formed, containing all historical task records similar to the current sub-area environment; then each reference surveying record set is sorted in ascending order according to cadastral modeling accuracy, wherein the unified evaluation standard of cadastral modeling accuracy adopts the accuracy indicators in the cadastral surveying regulations, including plane mean error and height mean error, the plane mean error is used to check the plane deviation between the model boundary points and the high-precision measured boundary points, and the root mean square error is calculated, the height mean error is used to check the deviation between the model elevation and the measured elevation, the modeling accuracy qualified threshold is set to a plane mean error less than or equal to 5 cm and a height mean error less than or equal to 10 cm, and can be adjusted according to the cadastral level, then the resource combination in the first historical record in each reference surveying record set sequence that meets the modeling accuracy qualified threshold and has the smallest plane mean error is extracted, including unmanned aerial vehicle type selection and sensor type selection, and finally it is determined as the selection result of the current sub-area.

[0034] Further, step S200 further comprises step S230, based on the building density and vegetation coverage density of each sub-area, a three-dimensional modeling complexity is identified, and each modeling complexity coefficient is generated; step S240, a basic forward overlap rate threshold and a basic lateral overlap rate threshold based on a preset flat complexity threshold are configured; step S250, based on the preset flat complexity threshold, each sub-area enhancement weight is calculated with the modeling complexity coefficients, and the basic forward overlap rate threshold and the basic lateral overlap rate threshold are subjected to weight enhancement processing, to generate each sub-area forward overlap rate and each sub-area lateral overlap rate; step S260, flight altitude configuration is performed with the distance constraint of the sensors in each selection result, flight path planning is performed in combination with each sub-area forward overlap rate and each sub-area lateral overlap rate, to generate each sub-area planning path; step S270, a corresponding relationship is constructed with the selection results and the sub-area planning paths, to generate the unmanned aerial vehicle-sensor-path mapping.

[0035] The step S250 further comprises generating the respective sub-region enhancement weight by calculating an enhancement ratio of the respective modeling complexity coefficient relative to the preset flat complexity threshold.

[0036] Preferably, based on the selection result, flight path planning is performed based on the flight altitude, the heading overlap rate and the lateral overlap rate. Specifically, three-dimensional modeling complexity is identified based on the building density and the vegetation coverage density of each sub-region, i.e. by calculating the weighted sum of the building density and the vegetation coverage density, the values are fused into a value representing complexity, and each modeling complexity coefficient is output, the greater the value, the more complex the three-dimensional modeling of the sub-region. The weight coefficient is set according to historical experience, the more dense the buildings, the more detailed the model, the higher the modeling complexity coefficient, and the more lush the vegetation, the more irregular its surface and the more it will sway with the wind, the higher the modeling complexity coefficient.

[0037] Preferably, the preset flat complexity threshold is used to define a flat simple region, and the basic heading overlap rate threshold and the basic lateral overlap rate threshold based on the preset flat complexity threshold are configured, i.e. the minimum heading overlap rate between adjacent photos and the lateral overlap rate between adjacent flight lines required to successfully complete three-dimensional modeling for a flat simple region, for example, the basic heading overlap rate threshold is 70% and the basic lateral overlap rate threshold is 60%. Then, based on the preset flat complexity threshold, the respective sub-region enhancement weight is calculated based on the respective modeling complexity coefficient, i.e. the enhancement ratio of the sub-region modeling complexity coefficient to the preset flat complexity threshold is calculated as the enhancement weight, wherein the weight of the flat sub-region is 1, the weight of the more complex sub-region is greater than 1, and the basic heading overlap rate threshold and the basic lateral overlap rate threshold are then subjected to weight enhancement processing, i.e. the basic heading overlap rate threshold and the basic lateral overlap rate threshold are respectively multiplied by the enhancement weight to obtain the respective sub-region heading overlap rate and the respective sub-region lateral overlap rate.

[0038] Preferably, the flight altitude configuration is performed according to the distance constraints of the selected sensors, wherein the distance constraints of the sensors refer to physical parameters of the selected sensors, mainly focal length and pixel size, and a preset ground resolution requirement, and a fixed flight relative height is calculated by a formula flight altitude = (focal length * preset ground resolution) / pixel size; based on the flight altitude, flight path planning is performed in combination with the heading overlap rate of each subzone and the lateral overlap rate of each subzone, specifically, a set of parallel flight routes are automatically generated according to the boundaries of the subzones, the set flight altitude and the overlap rate by using a standard track planning algorithm, so as to ensure that the entire subzone is completely covered and the customized overlap rate requirement is met, and then the planning paths of each subzone are output; finally, a corresponding relationship is constructed according to each selected result and each planning path of the subzone, that is, the unmanned aerial vehicle type, the sensor type and the planning path of the corresponding subzone and the flight altitude are mapped and associated to generate an unmanned aerial vehicle-sensor-path mapping, which is used to guide different types of unmanned aerial vehicles to carry different sensors to perform cadastral surveying and mapping tasks in different subzones according to different planning paths.

[0039] Further, the step S240 further includes that the basic heading overlap rate threshold and the basic lateral overlap rate threshold are a heading overlap rate lower limit and a lateral overlap rate lower limit of a preset flat zone when a three-dimensional modeling accuracy of the preset flat zone meets a preset accuracy threshold, wherein a three-dimensional modeling complexity of the preset flat zone is less than a preset flat complexity threshold.

[0040] Preferably, the preset flat zone is a pre-defined ideal region with extremely simple environment, and the three-dimensional modeling complexity of the preset flat zone is less than the preset flat complexity threshold, that is, the building density is extremely low, almost zero; the vegetation coverage density is extremely low, such as bare soil, hardened ground and the like; the terrain elevation change rate is extremely low, and the terrain is flat; when unmanned aerial vehicle aerial surveying is performed in the preset flat zone and three-dimensional modeling is completed, the plane and height errors of the generated three-dimensional modeling meet the preset accuracy threshold, that is, the standard value required by cadastral surveying and mapping specifications, such as the plane accuracy is better than 5 cm; then the heading overlap rate lower limit and the lateral overlap rate lower limit of the preset flat zone when the three-dimensional modeling accuracy meets the preset accuracy threshold are determined through experimental analysis, as the basic heading overlap rate threshold and the basic lateral overlap rate threshold, for example, it is verified that, on an absolutely flat and open site, using a specific camera, setting the basic heading overlap rate to 70% and the basic lateral overlap rate to 60% is the lowest condition for successfully generating a three-dimensional model meeting the accuracy requirement.

[0041] The step S300 controls the unmanned aerial vehicle to fly according to the unmanned aerial vehicle-sensor-path mapping, and collects low-altitude remote sensing data sets to establish a real scene three-dimensional model of the target surveying area.

[0042] Preferably, the UAV is controlled by the UAV-sensor-path mapping, specifically, for each sub-area task of the mapping, the equipment scheduling and loading are prompted, the operator is prompted to mount the corresponding sensor on the designated UAV, and the corresponding flight path planning including waypoints, altitudes, speeds, and flight parameters such as set heading overlap rate and lateral overlap rate are uploaded to the corresponding UAV through the ground station control center. The UAV flies along the flight path planning, and controls the sensor to expose and shoot at the designated position according to the set heading overlap rate and lateral overlap rate parameters, so as to ensure that each sub-area uses the most suitable equipment and the most optimized parameters to collect data, and then obtain a low-altitude remote sensing data set, mainly including a large number of sequence aerial images with high overlap and accurate position and attitude data, and corresponding flight trajectory, attitude record and other flight auxiliary data; then the collected low-altitude remote sensing data set is imported into the oblique photography three-dimensional modeling software for three-dimensional modeling, including identifying the same feature points in each photo and calculating the sparse point cloud by combining the accurate position and attitude data, then performing high-precision dense matching to generate high-density three-dimensional point cloud data, then connecting the three-dimensional point cloud data to construct a triangular mesh model expressing the geometric shape of the surface feature, and finally mapping the color information of the sequence aerial image to the triangular mesh model to generate a real scene three-dimensional model of the target survey area.

[0043] Step S400, calling the cadastral knowledge graph, reasoning, verifying and conflict detecting the real scene three-dimensional model, completing the correction of cadastral elements, and generating a target cadastral map.

[0044] Step S400 further includes the following steps: S410, extracting cadastral elements from the real scene three-dimensional model; S420, comparing the extracted cadastral elements with the pre-defined ownership definition rules in the cadastral knowledge graph to determine whether there is ambiguity, if so, marking the ambiguous area in the real scene three-dimensional model to generate a prompt information; S430, sending the prompt information to the artificial verification end through the human-computer interaction interface and receiving the returned correction instruction; S440, correcting the cadastral elements according to the correction instruction, and generating a target cadastral map with the corrected real scene three-dimensional model.

[0045] Preferably, the cadastral knowledge graph is called to reason, verify and conflict detect the real scene three-dimensional model, wherein the cadastral knowledge graph is a graph storing the rules and relationships in the cadastral field, for example, the house boundary should keep a distance of not less than 0.5 meters from the land boundary, the boundaries of adjacent lands must coincide, there must be no gaps or overlaps, and the fence is usually regarded as the land boundary. Specifically, cadastral features are extracted from the real scene three-dimensional model based on semantic segmentation and edge detection, including building contour lines, land boundary lines, fence barriers and other linear features, and road boundary lines; then the extracted cadastral features are compared with the pre-defined property boundary rules in the cadastral knowledge graph, that is, the pre-defined property boundary rules in the cadastral knowledge graph are called to check the extracted cadastral features one by one, and when a violation of the property boundary rules is detected, it is determined to be ambiguous, wherein the pre-defined property boundary rules in the cadastral knowledge graph include but are not limited to land closure, house-land relationship, boundary point consistency and fence-line, the land closure means that the land polygon must be closed and not self-intersecting; the house-land relationship means that the house contour must be completely located inside a certain land and keep a minimum distance D, such as D = 0.5 meters, from the land boundary; the boundary point consistency means that the common boundary line of adjacent lands must be defined by the same sequence of boundary points; the fence-line means that if the fence is continuous and not controversial, it can be regarded as the land boundary; for example, the contour line of the automatically extracted house M overlaps with the boundary line of the land N, and then the ambiguous area in the real scene three-dimensional model is labeled, such as highlighted, circled or generated with an error mark and a prompt information, for example, "Warning: a house feature and a land boundary conflict is found at coordinate xx, involving a property boundary error".

[0046] Preferably, the three-dimensional model with the labeled ambiguous area and detailed prompt information are pushed to the artificial verification end through the human-computer interaction interface for verification, including mapping the vector features such as house surface, land surface and boundary line extracted from the real scene three-dimensional model to the entity instances in the knowledge graph in combination with the field professional knowledge, historical archives and field investigation records, then importing the instantiated data into the rule engine, the rule engine traverses the rule set in the knowledge graph and performs spatial operations and logical judgments such as buffer analysis and overlay analysis to detect whether the rules are violated, if a violation of the house-land relationship rule is detected, the house and land boundary of the house are automatically highlighted in the three-dimensional model, triggering the labeling and prompting and generating a report "the boundary distance between the conflict house ID H001 and the land ID P005 is 0.2 meters <0.5 meters, suspected to be a boundary compression", then the problem is manually analyzed and a correction instruction is generated, and then returned through the human-computer interaction interface; finally, the cadastral features are corrected according to the correction instruction, for example, the boundary line of the house is moved according to the dragging of the auditor, so that it keeps a compliant distance from the land boundary, and finally a corrected real scene three-dimensional model is generated and a target cadastral map is output, thereby significantly improving the work efficiency of cadastral surveying and updating.

[0047] Further, step S400 further includes that the cadastral knowledge graph is obtained based on the collected multi-source cadastral rule data by identifying and structurally expressing cadastral entities and the relationships between the cadastral entities.

[0048] Preferably, based on the collected multi-source cadastral rule data, the cadastral knowledge graph is obtained by identifying and structurally expressing the cadastral entities and the relationships between the cadastral entities, that is, converting the scattered and unstructured cadastral rules, expert experience, field standards, local customs and the like into a structured and semantic knowledge network that can be understood and reasoned by a computer, so as to perform reasoning, checking, conflict detection and correction of cadastral elements on the real scene three-dimensional model, wherein the cadastral entities include a parcel, a boundary point, a boundary line, a building, a fence, a fence, a road and the like, and the spatial logical association and constraint rule relationship between the entities are used as the connection edges of the knowledge graph.

[0049] In the foregoing, with reference to Figure 1 The cadastral surveying and mapping optimization method based on low-altitude remote sensing according to the embodiments of the present application is described in detail. Next, with reference to Figure 2 The cadastral surveying and mapping optimization system based on low-altitude remote sensing according to the embodiments of the present application is described.

[0050] The cadastral surveying and mapping optimization system based on low-altitude remote sensing according to the embodiments of the present application is used to solve the technical problem that the cadastral surveying and mapping operation in the prior art is difficult to perform accurate data collection in a load environment, resulting in low accuracy and reliability of cadastral element identification and low efficiency of cadastral surveying and mapping operation, so as to improve the efficiency of cadastral surveying and mapping operation and enhance the accuracy and reliability of cadastral element identification in a complex environment. As shown in Figure 2 The cadastral surveying and mapping optimization system based on low-altitude remote sensing includes a feature recognition module 10, a path planning module 20, a model establishment module 30 and a cadastral map generation module 40.

[0051] The feature recognition module 10 is used to collect satellite image data of a target survey area, identify the building density distribution, vegetation coverage density distribution and terrain elevation change rate of the target survey area, and generate a recognition feature set. The path planning module 20 is used to select the type of unmanned aerial vehicle and sensor in a preset resource library based on the recognition feature set, and then perform flight path planning based on flight altitude, heading overlap rate and lateral overlap rate based on the selection result, and establish an unmanned aerial vehicle-sensor-path mapping. The model establishment module 30 is used to control the flight of the unmanned aerial vehicle based on the unmanned aerial vehicle-sensor-path mapping, collect low-altitude remote sensing data set and establish a real scene three-dimensional model of the target survey area. The cadastral map generation module 40 is used to call a cadastral knowledge graph, perform reasoning, checking and conflict detection on the real scene three-dimensional model, complete the correction of cadastral elements, and generate a target cadastral map.

[0052] Below, the specific configuration of the feature recognition module 10 will be described in detail. The feature recognition module 10 further comprises: identifying the building contour from the satellite image data through semantic segmentation, calculating the number or coverage of buildings in a unit area to obtain the building density distribution; distinguishing vegetation and non-vegetation based on the satellite image data by calculating the vegetation index, and evaluating the density of vegetation to obtain the vegetation coverage density distribution; performing spatial analysis on the satellite image data, calculating the difference between the elevation of each cell and the surrounding cells to generate a slope map reflecting the terrain elevation change rate; and generating the identified feature set based on the building density distribution, the vegetation coverage density distribution, and the slope map.

[0053] Below, the specific configuration of the path planning module 20 will be described in detail. The path planning module 20 further comprises: performing multi-level consistent clustering based on building density distribution, vegetation coverage density distribution, and terrain elevation change rate on the target survey area according to the identified feature set to establish various sub-areas; performing surveying and mapping accuracy optimization under corresponding environmental conditions based on the preset resource library for each sub-area to complete the selection of UAV type and sensor type, and obtaining each selection result corresponding to each sub-area.

[0054] Below, the specific configuration of the path planning module 20 will be described in detail. The path planning module 20 further comprises: collecting a set of historical surveying and mapping records based on the preset resource library, each historical surveying and mapping record corresponding to a set of resource combinations, each set of resource combinations including at least one UAV type and one sensor type; for each sub-area, each reference surveying and mapping record set in the set of historical surveying and mapping records has a regional environment similarity that meets a preset similarity threshold; extracting the resource combination with the highest cadastral modeling accuracy from each reference surveying and mapping record set to obtain the each selection result.

[0055] Below, the specific configuration of the path planning module 20 will be described in detail. The path planning module 20 further comprises: identifying the three-dimensional modeling complexity based on the building density and vegetation coverage density of each sub-area to generate each modeling complexity coefficient; configuring a basic forward overlap rate threshold and a basic lateral overlap rate threshold based on a preset flat complexity threshold; calculating the enhanced weight of each sub-area based on the preset flat complexity threshold and the each modeling complexity coefficient, and performing weight enhancement processing on the basic forward overlap rate threshold and the basic lateral overlap rate threshold to generate the forward overlap rate of each sub-area and the lateral overlap rate of each sub-area; configuring the flight altitude based on the distance constraint of the sensor in each selection result, and combining the forward overlap rate of each sub-area and the lateral overlap rate of each sub-area to plan the flight path to generate the planning path of each sub-area; and constructing the corresponding relationship between the each selection result and the planning path of each sub-area to generate the UAV-sensor-path mapping.

[0056] The specific configuration of the path planning module 20 will be described in detail below. The path planning module 20 further comprises that the basic heading overlap rate threshold and the basic lateral overlap rate threshold are lower limits of a heading overlap rate and a lateral overlap rate of a preset flat area when a three-dimensional modeling accuracy meets a preset accuracy threshold, wherein a three-dimensional modeling complexity of the preset flat area is less than a preset flat complexity threshold.

[0057] The specific configuration of the path planning module 20 will be described in detail below. The path planning module 20 further comprises that the respective sub-area enhancement weights are generated by calculating enhancement ratios of the respective modeling complexity coefficients relative to the preset flat complexity threshold.

[0058] The specific configuration of the cadastral map generation module 40 will be described in detail below. The cadastral map generation module 40 further comprises that cadastral elements are extracted from the real scene three-dimensional model; the extracted cadastral elements are compared with predefined property boundary rules in the cadastral knowledge graph to determine whether there is ambiguity, if so, the ambiguous area is marked in the real scene three-dimensional model to generate a prompt information; the prompt information is sent to an artificial verification end through a human-computer interaction interface and a correction instruction returned by the artificial verification end is received; the cadastral elements are corrected according to the correction instruction, and a target cadastral map is generated based on the corrected real scene three-dimensional model.

[0059] The specific configuration of the cadastral map generation module 40 will be described in detail below. The cadastral map generation module 40 further comprises that the cadastral knowledge graph is obtained based on collected multi-source cadastral rule data through identification and structured expression of cadastral entities and relationships between the cadastral entities.

[0060] The cadastral surveying and mapping optimization system based on low-altitude remote sensing provided by the embodiment of the present application can execute the cadastral surveying and mapping optimization method based on low-altitude remote sensing provided by any embodiment of the present application, and has the corresponding function modules and beneficial effects of executing the method.

[0061] The above is only a preferred embodiment of the present application, and does not limit the present application in any form. Although the present application has been disclosed as above with reference to the preferred embodiment, it is not intended to limit the present application. Any person skilled in the art can make some changes or modifications to the above disclosed technical content to obtain equivalent embodiments with equivalent changes, without departing from the scope of the technical solution of the present application. Any modification, equivalent change and modification made to the above embodiments based on the technical essence of the present application are still within the scope of the technical solution of the present application.

Claims

1. A cadastral surveying and mapping optimization method based on low-altitude remote sensing, characterized in that, The method comprises the following steps: Satellite image data of the target survey area is collected, and the building density distribution, vegetation coverage density distribution, and terrain elevation change rate of the target survey area are identified to generate an identification feature set; Based on the identification feature set, the type of unmanned aerial vehicle and the type of sensor are selected in the preset resource library, and then the flight path is planned based on the flight altitude, the heading overlap rate, and the lateral overlap rate based on the selection results, and the unmanned aerial vehicle-sensor-path mapping is established, specifically including: According to the identification feature set, the target survey area is subjected to multi-level consistent clustering based on the building density distribution, vegetation coverage density distribution, and terrain elevation change rate, and each sub-area is established; Based on the preset resource library, the surveying and mapping accuracy under the corresponding environmental conditions of each sub-area is optimized, the type of unmanned aerial vehicle and the type of sensor are selected, and each selection result corresponding to each sub-area is obtained; Based on the building density and vegetation coverage density of each sub-area, the three-dimensional modeling complexity is identified, and each modeling complexity coefficient is generated; The basic heading overlap rate threshold and the basic lateral overlap rate threshold based on the preset flat complexity threshold are configured; Based on the preset flat complexity threshold, the enhanced weight of each sub-area is calculated based on the modeling complexity coefficient, and the basic heading overlap rate threshold and the basic lateral overlap rate threshold are subjected to weight enhancement processing to generate the heading overlap rate of each sub-area and the lateral overlap rate of each sub-area; The flight altitude is configured based on the distance constraint of the sensor in each selection result, the flight path is planned in combination with the heading overlap rate of each sub-area and the lateral overlap rate of each sub-area, and the planning path of each sub-area is generated; The corresponding relationship is constructed based on the selection results and the planning paths of each sub-area, and the unmanned aerial vehicle-sensor-path mapping is generated; The unmanned aerial vehicle is controlled based on the unmanned aerial vehicle-sensor-path mapping, low-altitude remote sensing data set is collected, and a real scene three-dimensional model of the target survey area is established; The cadastral knowledge graph is called to reason, verify, and detect conflicts of the real scene three-dimensional model, the cadastral elements are modified, and a target cadastral map is generated; The basic heading overlap rate threshold and the basic lateral overlap rate threshold are configured, including: The basic heading overlap rate threshold and the basic lateral overlap rate threshold are the lower limit of the heading overlap rate and the lower limit of the lateral overlap rate of the preset flat area when the three-dimensional modeling accuracy meets the preset accuracy threshold, wherein the three-dimensional modeling complexity of the preset flat area is less than the preset flat complexity threshold; Based on the preset flat complexity threshold, the enhanced weight of each sub-area is calculated based on the modeling complexity coefficient, including: The enhanced weight of each sub-area is generated by calculating the enhancement ratio of the modeling complexity coefficient with respect to the preset flat complexity threshold.

2. The low-altitude remote sensing-based cadastral mapping optimization method according to claim 1, characterized in that, Satellite image data of the target survey area is collected, and the building density, vegetation coverage density, and terrain elevation change rate of the target survey area are identified to generate an identification feature set, including: The building contour is identified by semantic segmentation from the satellite image data, the number or coverage of buildings per unit area is calculated, and the building density distribution is obtained. Based on the satellite image data, the vegetation and non-vegetation are distinguished by calculating the vegetation index, and the density of vegetation coverage is evaluated to obtain the density distribution of vegetation coverage; The satellite image data is subjected to spatial analysis, the difference between the elevation of each cell and the surrounding cells is calculated, and a slope map reflecting the change rate of terrain elevation difference is generated; The building density distribution, the density distribution of vegetation coverage, and the slope map are used to generate the set of identification features.

3. The low-altitude remote sensing-based cadastral mapping optimization method according to claim 1, wherein, The selected results corresponding to the respective sub-regions are obtained, including: A set of historical surveying and mapping records based on the preset resource library is collected, each historical surveying and mapping record corresponding to a group of resource combinations, and each group of resource combinations including at least one type of unmanned aerial vehicle and one type of sensor; For each sub-region, each reference surveying and mapping record set in the set of historical surveying and mapping records has a region environment similarity that reaches a preset similarity threshold; The resource combination with the highest cadastral modeling accuracy is extracted from each reference surveying and mapping record set, and the selected results are obtained.

4. The low-altitude remote sensing based cadastral mapping optimization method of claim 1, wherein, The cadastral knowledge graph is called to reason, verify, and detect conflicts on the real scene three-dimensional model, complete the correction of cadastral elements, and generate a target cadastral map, including: Cadastral elements are extracted from the real scene three-dimensional model; The extracted cadastral elements are compared with the pre-defined ownership definition rules in the cadastral knowledge graph to determine whether there is ambiguity, and if so, the ambiguous area is marked in the real scene three-dimensional model to generate a prompt information; The prompt information is sent to the artificial verification end through the human-computer interaction interface and the returned correction instruction is received; The cadastral elements are corrected according to the correction instruction, and a target cadastral map is generated based on the corrected real scene three-dimensional model.

5. The low-altitude remote sensing based cadastral mapping optimization method according to claim 4, characterized in that, The cadastral knowledge graph is based on the collected multi-source cadastral rule data, and is obtained by identifying and structurally expressing the relationship between cadastral entities.

6. A cadastral mapping optimization system based on low-altitude remote sensing, characterized by, The system is used to implement the cadastral surveying and mapping optimization method based on low-altitude remote sensing according to any one of claims 1 to 5, and the system includes: A feature identification module is configured to collect satellite image data of a target surveying area, identify the building density distribution, vegetation coverage density distribution, and terrain elevation difference change rate of the target surveying area, and generate a set of identification features; A path planning module is configured to select a type of unmanned aerial vehicle and a type of sensor in a preset resource library based on the set of identification features, and perform flight path planning based on the selected results, flight altitude, heading overlap rate, and lateral overlap rate to establish an unmanned aerial vehicle-sensor-path mapping; A model establishment module is configured to control the flight of an unmanned aerial vehicle based on the unmanned aerial vehicle-sensor-path mapping, collect a set of low-altitude remote sensing data, and establish a real scene three-dimensional model of the target surveying area; A cadastral map generation module is configured to call a cadastral knowledge graph, reason, verify, and detect conflicts on the real scene three-dimensional model, complete the correction of cadastral elements, and generate a target cadastral map.

Images