A geospatial business calibration method and system

Abstract

The present invention provides a geospatial service calibration method and system, the method comprising: dividing a map spot to be calibrated into a plurality of grids, recording each grid data, the grid data including the row and column number of the grid and the effective area of ​​the grid; inputting each grid data into a first logic unit, obtaining the first prediction data of each grid, including at least one service type of the grid and the confidence of each service type; inputting the first prediction data of all grids of the map spot to be calibrated into a second logic unit, obtaining the second prediction data of the map spot to be calibrated, including at least one service type of the map spot to be calibrated and the confidence of each service type. The present invention combines spatial grid and artificial intelligence technology, generates grid sample data by gridding the monitoring area, generates map sample data after dividing the historical service map spot according to the same spatial grid scale, obtains a model after multiple rounds of iterative training, and finally realizes automatic calibration of the map spot.

Description

Technical Field

[0001] The present invention relates to the field of geographic information, and more specifically, to a geographic space service calibration method and system. Background Art

[0002] With the rapid development of space remote sensing and aerial remote sensing, the use of various types of images for analysis and extraction of spots has become a normal practice. Correspondingly, the real-time and intelligent process and method of spot business classification have become urgent new demands. At present, the common method of combining AI instance segmentation with manual visual interpretation is used to extract the range of spots. Since there are many business departments and regulatory departments that may be involved in the spots, how to intelligently and accurately classify the spots and push them to the corresponding business departments is a core problem that needs to be solved.

[0003] The current technical solution is to use DOM and DLG as reference maps, combine manual experience, and manually label them before issuing them. However, this technical solution has the disadvantages of low calibration efficiency, large workload, reliance on professional talents, and inability to achieve unmanned operation. Summary of the invention

[0004] The present invention aims at solving the technical problems existing in the prior art and provides a geospatial service calibration method and system.

[0005] According to a first aspect of the present invention, a geospatial service calibration method is provided, comprising:

[0006] Obtain a spot to be calibrated, divide the spot to be calibrated into a plurality of grids, and record data of each grid, wherein the grid data includes row and column numbers of the grid and an effective area of ​​the grid, and the row and column numbers of the grid represent the spatial position of the grid;

[0007] Inputting each grid data into a first logic unit, obtaining first prediction data of each grid output by the first logic unit, wherein the first prediction data includes at least one service type of the grid and the confidence of each service type;

[0008] The first prediction data of all grids of the spot to be calibrated output by the first logic unit is input into the second logic unit, and the second prediction data of the spot to be calibrated output by the second logic unit is obtained, wherein the second prediction data includes at least one business type of the spot to be calibrated and the confidence of each business type.

[0009] According to a second aspect of the present invention, there is provided a geospatial service calibration system, comprising:

[0010] A first acquisition unit is used to acquire a to-be-calibrated image spot, divide the to-be-calibrated image spot into a plurality of grids, and record data of each grid, wherein the grid data includes row and column numbers of the grid and an effective area of ​​the grid, and the row and column numbers of the grid represent the spatial position of the grid;

[0011] A second acquisition unit, configured to input each grid data into the first logic unit, and acquire first prediction data of each grid output by the first logic unit, wherein the first prediction data includes at least one service type of the grid and a confidence level of each service type;

[0012] The third acquisition unit is used to input the first prediction data of all grids of the image spot to be calibrated output by the first logic unit into the second logic unit, and obtain the second prediction data of the image spot to be calibrated output by the second logic unit, wherein the second prediction data includes at least one business type of the image spot to be calibrated and the confidence of each business type.

[0013] The present invention provides a geographic space business calibration method and system, which combines spatial grid and artificial intelligence technology, generates grid sample data by gridding the monitoring area, generates spot sample data after dividing the historical business spots according to the same spatial grid scale, and trains two logical units based on the grid sample data and the spot sample data respectively. According to the combination of the two logical units, the automatic calibration of the business types included in the spots is finally achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 A flow chart of a geospatial service calibration method provided by the present invention;

[0015] Figure 2 A schematic diagram of the structure of a geospatial business calibration system provided by the present invention. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. In addition, the technical features in the various embodiments or single embodiments provided by the present invention can be arbitrarily combined with each other to form a feasible technical solution. This combination is not subject to the constraints of the sequence of steps and / or the structural composition mode, but must be based on the ability of ordinary technicians in this field to achieve. When the combination of technical solutions is contradictory or cannot be achieved, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0017] Figure 1 A flow chart of a geospatial service calibration method provided by the present invention, such as Figure 1 As shown, the method includes:

[0018] Step 1, obtain the image spot to be calibrated, divide the image spot to be calibrated into multiple grids, and record the data of each grid, wherein the grid data includes the row and column numbers of the grid and the effective area of ​​the grid, and the row and column numbers of the grid represent the spatial position of the grid.

[0019] It is understandable that the image spot to be calibrated is obtained, the image spot to be calibrated is divided into a plurality of grids according to the size of each grid, and the data information of each grid is recorded.

[0020] The size of each grid after division can be preliminarily determined according to the size of the historical spots. The size of the grid affects the training and operation effects and calibration accuracy. The grid size can also be adjusted according to the application effect in the later stage. Specifically, collect historical spot data, obtain the length and width of the bounding rectangle of each historical spot, and calculate the average length and width of the bounding rectangle of all historical spots; determine the size of each grid after division according to the average size of multiple historical spots.

[0021] The calculation formula for the grid size is:

[0022] ;

[0023] in, is the width or height of the pth patch, g is the grid size adjustment coefficient, and m is the total number of patches.

[0024] After dividing the map to be calibrated into multiple grids, record the data of each grid. The grid data mainly includes the spatial position information of the grid. In the present invention, the spatial position information of the grid is represented by the row and column numbers of the grid in the map, and the business type of each grid and the intersection area corresponding to each business type are recorded. The intersection area refers to the intersection area between the current grid and the geometric range of the map.

[0025] Step 2: input each grid data into the first logic unit, and obtain the first prediction data of each grid output by the first logic unit, wherein the first prediction data includes at least one business type of the grid and the confidence of each business type.

[0026] It is understandable that before the grid data of the map to be calibrated is input into the first logic unit, the first logic unit is trained. Collect historical map data, grid-divide the historical map to form grid samples. For example, a map with a land use status of cultivated land is divided into 4 blocks according to the grid model, and 4 grid samples can be formed. The row and column numbers, business types and intersection areas of the grids are recorded. The intersection area is the area of ​​the intersection of the current grid and the geometric range of the map. The grid sample data is as follows:

[0027] ;

[0028] Where i is the row number of the grid, j is the column number of the grid, k is the service type of the grid, and a is the intersection area of ​​the grid.

[0029] Since each grid contains multiple business types, each grid sample needs to record all business type information:

[0030] ;

[0031] Among them, g represents the grid, i is the row number of the grid, and j is the column number of the grid. is the nth business type, is the intersection area of ​​the nth business type.

[0032] A plurality of grid samples form a grid sample set, and a first logic unit is trained based on the grid sample set, wherein the first logic unit is a deep learning network model.

[0033] The step of training the first logic unit based on the grid sample set includes:

[0034] Input each grid sample in the grid sample set into the first logic unit, and perform a deep convolution operation on each level of neurons to obtain a forward propagation result of each level of neurons;

[0035] Obtaining the grid sample prediction value output by the last level of neurons, wherein the grid sample prediction value includes the business type prediction value included in the grid and the confidence prediction value of each business type;

[0036] Calculate the forward propagation loss value of the first logic unit based on each grid sample actual value and grid sample predicted value;

[0037] Based on the forward propagation loss value, the weight values ​​of neurons at each level are gradually updated through back propagation;

[0038] After updating the weight values ​​of neurons at all levels, the iteration cycle continues until the first logic unit is fitted to obtain the optimal solution.

[0039] The specific process of training the first logic unit includes:

[0040] The first logical unit is iteratively trained using convolution operation and sigmoid activation function.

[0041] ;

[0042] Where: Indicates The weight matrix of the neurons in the layer, Indicates The forward propagation result of the layer neurons, For the The bias matrix of the layer neurons, For the The result after convolution operation of layer neurons.

[0043] ;

[0044] Where: Represents the result of the activation function.

[0045] After the grid samples pass through the multi-layer neurons of the first logic unit, the final output is the confidence level that meets each type of business:

[0046] ;

[0047] Where: i represents the grid row number, j represents the grid column row, Indicates the business type. Indicates the confidence level of the business type.

[0048] Calculate the loss value after forward propagation of the first logistic unit:

[0049] ;

[0050] In the formula, represents the forward propagation loss value, represents the actual value of the wth sample, Represents the predicted value of the w-th sample.

[0051] The weight values ​​of each level of neural network are gradually updated through back propagation:

[0052] ;

[0053] In the formula, represents the forward propagation loss, Indicates The forward propagation calculation results of the neural network layer, Indicates Layer network weight values.

[0054] After updating the weight values ​​of each level of neural network, continue to iterate until the model is fitted to obtain the optimal solution.

[0055] After the first logic unit is obtained through training, multiple grid data of the image spot to be calibrated are input into the first logic unit, and the first logic unit outputs the business type included in each grid and the confidence of each business type.

[0056] Step 3: input the first prediction data of all grids of the image spot to be calibrated output by the first logic unit into the second logic unit, and obtain the second prediction data of the image spot to be calibrated output by the second logic unit, wherein the second prediction data includes at least one business type of the image spot to be calibrated and the confidence of each business type.

[0057] It is understandable that a second logic unit is trained to calibrate the service type of the image spot. The second logic unit is trained, wherein the second logic unit is a deep learning network model.

[0058] The training process of the second logic unit includes:

[0059] Get a spot sample set. The spot sample set includes multiple spot samples. Each spot sample includes each grid information contained in the spot. Each spot sample is represented as:

[0060] ;

[0061] in, Indicates the first Grid information of each grid;

[0062] The second logic unit is trained based on the pattern sample set.

[0063] It should be noted that the method for training the second logic unit based on the patch sample set is the same as the method for training the first logic unit based on the grid sample set, and will not be repeated here.

[0064] After the training of the second logic unit is completed, the grid data (including the row and column numbers of each grid) after the map to be calibrated is divided and the predicted value of the business type included in each grid and the predicted value of the confidence of each business type output by the first logic unit, that is, the first predicted data of each grid are input into the second logic unit, and the second logic unit outputs the predicted value of the business type included in the map to be calibrated and the predicted value of the confidence of each business type, and determines the business type included in the map to be calibrated according to the predicted value of the confidence of each business type. For example, the business type with a confidence greater than a preset threshold is regarded as the business type included in the map to be calibrated.

[0065] See also Figure 2 , is a structural diagram of a geospatial service calibration system provided by an embodiment of the present invention, the system includes a first acquisition module 201, a second acquisition module 202 and a third acquisition module 203, wherein:

[0066] The first acquisition module 201 is used to acquire a spot to be calibrated, divide the spot to be calibrated into a plurality of grids, and record data of each grid, wherein the grid data includes row and column numbers of the grid and an effective area of ​​the grid, and the row and column numbers of the grid represent the spatial position of the grid;

[0067] A second acquisition module 202 is used to input each grid data into the first logic unit, and acquire first prediction data of each grid output by the first logic unit, where the first prediction data includes at least one service type of the grid and the confidence of each service type;

[0068] The third acquisition module 203 is used to input the first prediction data of all grids of the image spot to be calibrated output by the first logic unit into the second logic unit, and obtain the second prediction data of the image spot to be calibrated output by the second logic unit, wherein the second prediction data includes at least one business type of the image spot to be calibrated and the confidence of each business type.

[0069] It can be understood that the geospatial business calibration system provided by the present invention corresponds to the geospatial business calibration method provided in the aforementioned embodiments. The relevant technical features of the geospatial business calibration system can refer to the relevant technical features of the geospatial business calibration method, which will not be repeated here.

[0070] A geospatial business calibration method and system provided in an embodiment of the present invention combines spatial grid and artificial intelligence technology, generates grid sample data by gridding the monitoring area, generates spot sample data after dividing the historical business spots according to the same spatial grid scale, obtains a model after multiple rounds of iterative training, and finally realizes automatic calibration of the spot business type.

[0071] It should be noted that in the above embodiments, the description of each embodiment has its own emphasis, and for parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0072] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0073] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0074] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0075] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0076] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0077] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A geospatial service calibration method, characterized in that: include: Obtain a spot to be calibrated, divide the spot to be calibrated into a plurality of grids, and record data of each grid, wherein the grid data includes row and column numbers of the grid and an effective area of ​​the grid, and the row and column numbers of the grid represent the spatial position of the grid; Inputting each grid data into a first logic unit, obtaining first prediction data of each grid output by the first logic unit, wherein the first prediction data includes at least one service type of the grid and the confidence of each service type; Inputting the first prediction data of all grids of the spot to be calibrated output by the first logic unit into the second logic unit, obtaining the second prediction data of the spot to be calibrated output by the second logic unit, wherein the second prediction data includes at least one service type of the spot to be calibrated and the confidence of each service type; The training process of the first logic unit includes: Acquire historical spot data, divide each historical spot into a plurality of grids, record information of each grid, and form a grid sample set, wherein the grid sample set includes a plurality of grid samples; The grid information includes the row and column numbers of the grid, the service types included in the grid, and the intersection area of ​​each service type. The intersection area refers to the area of ​​intersection of the grid and the geometric range of the patch. The grid information is expressed as: ； Among them, g represents the grid, i is the row number of the grid, and j is the column number of the grid. is the nth business type, is the intersection area of ​​the nth business type; Training the first logic unit based on the grid sample set; The training process of the second logic unit includes: A spot sample set is obtained, wherein the spot sample set includes a plurality of spot samples, each spot sample includes each grid information contained in the spot, and each spot information is represented as: ； in, Indicates the first Grid information of each grid; The second logic unit is trained based on the pattern sample set.

2. The geospatial service calibration method according to claim 1, characterized in that: Dividing the image patch to be calibrated into a plurality of grids, including: Collect historical spot data, obtain the length and width of the bounding rectangle of each historical spot, and calculate the average length and width of the bounding rectangle of all historical spots; Determine the size of each grid after division according to the average value; Wherein, the average value is: ； in, is the width or height of the pth patch, g is the grid size adjustment coefficient, and m is the total number of patches.

3. The geospatial service calibration method according to claim 1, characterized in that: Training the first logic unit based on the grid sample set includes: Input each grid sample in the grid sample set into the first logic unit, and perform a deep convolution operation on each level of neurons to obtain a forward propagation result of each level of neurons; Obtaining the grid sample prediction value output by the last level of neurons, wherein the grid sample prediction value includes the business type prediction value included in the grid and the confidence prediction value of each business type; Calculate the forward propagation loss value of the first logic unit based on each grid sample actual value and grid sample predicted value; Based on the forward propagation loss value, the weight values ​​of neurons at each level are gradually updated through back propagation; After updating the weight values ​​of neurons at all levels, the iteration cycle continues until the first logic unit is fitted to obtain the optimal solution.

4. The geospatial service calibration method according to claim 3, characterized in that: Training the second logic unit based on the spot sample set includes: The second logic unit is trained using the same training method as the first logic unit.

5. A geospatial business calibration system, characterized in that: include: The first acquisition module is used to acquire the image spot to be calibrated, divide the image spot to be calibrated into multiple grids, and record the data of each grid, wherein the grid data includes the row and column numbers of the grid and the effective area of ​​the grid, and the row and column numbers of the grid represent the spatial position of the grid; A second acquisition module, used for inputting each grid data into the first logic unit, and acquiring first prediction data of each grid output by the first logic unit, wherein the first prediction data includes at least one service type of the grid and the confidence of each service type; A third acquisition module is used to input the first prediction data of all grids of the spot to be calibrated output by the first logic unit into the second logic unit, and obtain the second prediction data of the spot to be calibrated output by the second logic unit, wherein the second prediction data includes at least one business type of the spot to be calibrated and the confidence of each business type; The training process of the first logic unit includes: Acquire historical spot data, divide each historical spot into a plurality of grids, record information of each grid, and form a grid sample set, wherein the grid sample set includes a plurality of grid samples; The grid information includes the row and column numbers of the grid, the service types included in the grid, and the intersection area of ​​each service type. The intersection area refers to the area of ​​intersection of the grid and the geometric range of the patch. The grid information is expressed as: ； Among them, g represents the grid, i is the row number of the grid, and j is the column number of the grid. is the nth business type, is the intersection area of ​​the nth business type; Training the first logic unit based on the grid sample set; The training process of the second logic unit includes: A spot sample set is obtained, wherein the spot sample set includes a plurality of spot samples, each spot sample includes each grid information contained in the spot, and each spot information is represented as: ； in, Indicates the first Grid information of each grid; The second logic unit is trained based on the pattern sample set.

Images