A satellite map geographic object desensitization method and system based on vector map

Abstract

The present invention provides a method and system for desensitizing satellite map geographic objects based on vector maps, which relates to the field of geographic information security technology, including obtaining first information including vector map data and satellite map data to be trained, and second information including vector map data and satellite map data to be desensitized. A judgment model is used to determine whether a sample set can be constructed. If so, the first information is matched and segmented to obtain a sample set, and a convolutional encoding and decoding neural network model is used to learn and train the sample set to generate a trained satellite map data desensitization model. The specific geographic elements in the second information vector map data are processed and sent to the satellite map data desensitization model to perform desensitization processing on the specific geographic objects in the satellite map. The present invention can not only hide sensitive geographic objects in satellite maps by deleting specific vector geographic elements, but also perform diversified desensitization on satellite map geographic objects, so as to achieve the purpose of controllable desensitization of geographic objects.

Description

Technical Field

[0001] The present invention relates to the field of geographic information security technology, and in particular to a method and system for desensitizing geographic objects in satellite maps based on vector maps. Background Art

[0002] Satellite image maps produced based on satellite remote sensing images can more realistically express the regional geographical environment than vector maps, and are convenient for assisting users to form intuitive and comprehensive spatial cognition. They have been widely used in various electronic map platforms. Satellite map production focuses on satellite remote sensing image radiation correction, geometric correction, and splicing processing, and often ignores the selective expression of geographical elements. Directly publishing them on the Internet poses a high risk of geographic information security.

[0003] The satellite images generated by the existing satellite image desensitization technology often ignore the desensitization of finer-grained geographic objects such as buildings and roads in satellite maps. In particular, the desensitization strategy for geographic objects in satellite images is relatively simple, and the controllability and flexibility of desensitization of geographic objects in satellite maps are insufficient. The desensitization results of satellite map geographic objects generated by current technology often have problems such as the local desensitization results are inconsistent with the context environment, cannot effectively express the regional geographical environment, and affect the use of satellite maps. Therefore, a satellite map geographic object desensitization method based on vector maps is needed to achieve more controllable and flexible satellite map geographic object desensitization processing, so that the satellite map desensitization results take into account both effectiveness and security, so as to meet the current and future needs of protecting the security of Internet geographic information. Taking advantage of the rich semantic information and flexible editing and processing of vector data, based on deep learning models, the purpose of desensitizing specific geographic objects in satellite maps is achieved. Summary of the invention

[0004] The purpose of the present invention is to provide a satellite map geographic object desensitization method and system based on vector map to improve the above problems. In order to achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0005] In a first aspect, the present application provides a method for desensitizing geographic objects in a satellite map based on a vector map, comprising:

[0006] Acquire first information and second information, wherein the first information includes vector map data and satellite map data to be trained, and the second information includes vector map data and satellite map data to be desensitized;

[0007] The first information is sent to a judgment model to judge whether the first information can construct a sample set. If the sample set can be constructed, the first information is matched and segmented to obtain a sample set corresponding to the first information, and the sample set is a training set and a test set for training a satellite map data desensitization model;

[0008] Sending the sample set corresponding to the first information to a preset convolutional encoding and decoding neural network model for training to obtain a trained satellite map data desensitization model;

[0009] The vector map data to be desensitized in the second information is processed, and the processed second information is sent to the trained satellite map data desensitization model to desensitize the satellite map data, so as to obtain a geographic desensitized object of the satellite map corresponding to the specific geographic elements of the vector map.

[0010] In the second aspect, the present application also provides a satellite map geographic object desensitization system based on a vector map, comprising:

[0011] an acquisition unit, configured to acquire first information and second information, wherein the first information includes vector map data and satellite map data to be trained, and the second information includes vector map data and satellite map data to be desensitized;

[0012] A judgment unit, configured to send the first information to a judgment model to judge whether the first information can construct a sample set, and if the sample set can be constructed, match and segment the first information to obtain a sample set corresponding to the first information, wherein the sample set is a training set and a test set for training a satellite map data desensitization model;

[0013] A training unit, configured to send a sample set corresponding to the first information to a preset convolutional encoding and decoding neural network model for training to obtain a trained satellite map data desensitization model;

[0014] A processing unit is used to process the vector map data to be desensitized in the second information, and send the processed second information to the trained satellite map data desensitization model to desensitize the satellite map data to obtain desensitized geographical objects of the satellite map corresponding to the specific geographical elements of the vector map.

[0015] The beneficial effects of the present invention are:

[0016] The present invention matches satellite image maps and vector map data, divides the two in combination with map scale and image resolution, and extracts a sample set suitable for deep neural network model learning; then, a deep generation model structure is designed based on a convolutional neural network, and a desensitization model that can fit the mapping from vector map to satellite map is obtained through sample set training; finally, specific geographic elements in the vector map data of the desensitized area are added, deleted, and modified, and the modified vector map data is input into the satellite map desensitization model to generate a desensitized satellite map that hides, adds, changes, and shifts specific geographic objects, so as to achieve the purpose of controllable desensitization of satellite map geographic objects. Among them, based on the deep neural network generation model fitting the mapping from vector map to satellite map, the expression of corresponding geographic objects of satellite map is controlled by adjusting the vector map geographic elements, so as to achieve controllable desensitization of satellite map geographic objects. The desensitization method can not only achieve the purpose of hiding sensitive geographic objects of satellite maps by deleting specific vector geographic elements, but also achieve diversified desensitization of satellite map geographic objects by shifting, modifying, and creating new vector geographic elements.

[0017] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or be understood by implementing the embodiments of the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description, claims, and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0019] Figure 1 A schematic diagram of a process flow of a satellite map geographic object desensitization method based on a vector map according to an embodiment of the present invention;

[0020] Figure 2 The present invention is a schematic diagram of the structure of a satellite map geographic object desensitization system based on a vector map according to an embodiment of the present invention.

[0021] In the figure: 701, acquisition unit; 702, judgment unit; 703, training unit; 704, processing unit; 7021, first processing subunit; 7022, second processing subunit; 7023, first judgment subunit; 7024, third processing subunit; 7025, fourth processing subunit; 7026, fifth processing subunit; 7027, sixth processing subunit; 7031, first training subunit; 7032, second training subunit; 7033, third training subunit; 7041, seventh processing subunit; 7042, eighth processing subunit; 7043, ninth processing subunit; 70421, second judgment subunit; 70422, third judgment subunit; 70423, fourth judgment subunit; 70424, fifth judgment subunit. DETAILED DESCRIPTION

[0022] 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, rather than all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0023] It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings. At the same time, in the description of the present invention, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance. Embodiment 1:

[0024] This embodiment provides a method for desensitizing geographic objects in a satellite map based on a vector map.

[0025] See also Figure 1 , the figure shows that the method includes step S1, step S2, step S3 and step S4.

[0026] Step S1, obtaining first information and second information, wherein the first information includes vector map data and satellite map data to be trained, and the second information includes vector map data and satellite map data to be desensitized;

[0027] It can be understood that in this step, by obtaining high-quality first information and second information, the accuracy and consistency of the data in the subsequent steps are ensured, thereby improving the effect of desensitization model training and the authenticity of the final desensitization result. Among them, the present invention can obtain the latest vector map data from the urban planning department, use high-resolution commercial satellites (such as WorldView-3) to produce satellite maps of the area, and can also obtain vector map data and satellite map data from open source geographic information platforms.

[0028] Step S2: Send the first information to the judgment model to judge whether the first information can construct a sample set. If the sample set can be constructed, match and segment the first information to obtain a sample set corresponding to the first information, and the sample set is a training set and a test set for training a satellite map data desensitization model.

[0029] It can be understood that in this step, the input data is evaluated and matched to ensure the rationality and consistency of the sample set. This ensures the effectiveness of subsequent training data and improves the accuracy of the desensitization model. Among them, step S2 includes step S21, step S22 and step S23.

[0030] Step S21: determining first sub-information based on the first information, wherein the first sub-information includes a minimum resolvable distance and a scale of the vector map data to be trained;

[0031] It is understandable that determining the first sub-information (minimum resolvable distance and scale) in this step helps improve the accuracy of the data and ensure that the training data contains enough details to reflect the real geographical features. By clarifying the scale and minimum resolvable distance of the data, a standard is provided for subsequent data processing and model training. Providing high-quality input data helps train a more accurate and reliable desensitization model.

[0032] Step S22: determining second sub-information based on the first information, wherein the second sub-information includes the spatial resolution of the satellite map data to be trained;

[0033] It is understandable that determining the second sub-information (spatial resolution) in this step helps to: ensure image quality. Satellite images with high spatial resolution can capture more geographic details and improve the training effect of the desensitization model. Clarifying the spatial resolution of the image provides a standard for subsequent data processing and model training. Providing high-quality input image data helps to train a more accurate and reliable desensitization model.

[0034] Step S23: determine whether the first sub-information and the second sub-information can construct a sample set based on a preset judgment formula; if the sample set can be constructed, perform projection transformation on the vector map data and satellite map data to be trained in the first information to obtain vector map data and satellite map data in the same coordinate system.

[0035] It can be understood that this step ensures the quality and consistency of the sample set data through reasonable evaluation and conversion. Through resolution matching and scale consistency check, it is ensured that the vector data and satellite image data can be reasonably combined. Through projection transformation, it is ensured that the data is represented in the same coordinate system to avoid spatial dislocation in subsequent processing. Providing high-quality and consistent sample data helps to train a more accurate and reliable desensitization model. In this step, the preset judgment formula is as follows:

[0036] λ·(svo / scale)≤pix <svo / scale,0<λ<1

[0037] Among them, λ represents the adjustment parameter, svo represents the minimum resolvable distance on the vector map, scale represents the scale of the vector map, and pix represents the spatial resolution of the satellite map data.

[0038] It can be understood that in this step, step S2 also includes step S24, step S25, step S26 and step S27.

[0039] Step S24: constructing a subdivision grid for the vector map data and satellite map data to be trained in the first information based on a regular grid method, obtaining the subdivision grid for the vector map data and satellite map data to be trained in the first information, and performing regional subdivision on the vector map data and satellite map data using the subdivision grid;

[0040] It can be understood that this step uses the regular grid method to divide large-scale geographic data into smaller units, simplifying data management and processing. The divided data can be processed in parallel to improve computing efficiency and model training speed. Ensure that vector map data and satellite image data are divided in the same coordinate system to maintain data consistency and accuracy.

[0041] Step S25, selecting a grid of intersection data of the vector map data and satellite map data to be trained in the first information;

[0042] It can be understood that the intersection data extracted in this step ensures the spatial alignment of the vector map data and the satellite image data, providing a reliable basis for subsequent data processing and model training. Eliminating incomplete data units ensures that each extracted grid unit contains complete and consistent geographic information, which improves the quality of model training data. By extracting intersection data, unnecessary data processing burden is reduced and the efficiency of data processing and model training is improved.

[0043] Step S26, performing rasterization and normalization processing on the intersection data based on the spatial resolution of the satellite map data as the grid resolution, to obtain a set of pixel value matrices of the vector map data to be trained and pixel value matrices of the satellite map data to be trained;

[0044] It can be understood that this step uses the spatial resolution of the satellite map data as the grid resolution and the grid size to rasterize and normalize the area within the intersection data grid, wherein a regular grid grid is generated in the area covered by the vector data according to the selected grid resolution. Assign pixel values: According to the spatial position of the vector data, the corresponding pixel values ​​are assigned to the grid grid. For example, for polygonal vector data, the pixels inside the polygon can be assigned to (255, 255, 255), and the pixels outside can be assigned to (255, 255, 0); for line data, the pixels where the line is located can be assigned to (255, 0, 0), and the other pixels can be assigned to (0, 0, 0); different colors can be used to distinguish different semantic geographic objects. Convert vector data and satellite image data into a unified pixel value matrix form to facilitate subsequent model training. Through normalization processing, the value ranges of different data sources are unified, the differences between data are reduced, and the effect of model training is improved.

[0045] Step S27: Divide all the pixel value matrix sets into a training set and a test set according to a preset ratio to obtain a sample set corresponding to the constructed first information.

[0046] It can be understood that in this step, all pixel value matrices are divided into training sets and test sets according to a preset ratio to construct the sample set required for model training. The training set is used to train the model, and the test set is used to evaluate the performance and generalization ability of the model.

[0047] Step S3: sending the sample set corresponding to the first information to a preset convolutional encoding and decoding neural network model for training to obtain a trained satellite map data desensitization model;

[0048] It can be understood that the encoder in this step is used to extract the features of the input data, and is usually composed of multiple convolutional layers and pooling layers. The decoder is used to convert the extracted features into output data, and is usually composed of multiple convolutional layers and upsampling layers (such as deconvolution layers). The convolutional layer can effectively extract the spatial features of the input data, the encoder extracts high-level features, and the decoder restores the detail information. Through the learning of a large number of samples, the model can generate realistic desensitized satellite map data, which improves the authenticity and accuracy of desensitization. In this step, step S3 includes step S31, step S32 and step S33.

[0049] Step S31, constructing a convolutional neural network model based on a preset encoder and decoder;

[0050] It can be understood that in this step, the encoder can effectively extract the features of the input data, and the decoder restores these features to output data, achieving data compression and restoration. The convolution kernel in the convolutional neural network has the characteristic of parameter sharing, which reduces the number of parameters of the model and improves the generalization ability of the model. By adjusting the number and width of the encoder and decoder, the complexity of the model can be controlled to adapt to different data characteristics.

[0051] Step S32: sending the training in the sample set corresponding to the first information to the convolutional neural network model for training, wherein the training set is input into the convolutional neural network model in batches, the loss function of each batch of data is repeatedly calculated, and the model parameters are adjusted using the gradient descent algorithm so that the model continuously approaches the preset objective function, thereby obtaining the trained convolutional neural network model;

[0052] It can be understood that in this step, the model parameters are adjusted through the gradient descent algorithm so that the model can gradually adapt to the characteristics of the training data and improve the performance of the model. When the loss function reaches a smaller value or the number of training rounds reaches the preset value, the model training ends, indicating that the model has converged and reached a certain performance level. The preset objective function in this step is as follows:

[0053] ;

[0054] Among them, DE(W,B) represents the parameter set of the input convolutional neural network model, Represents the pixel value matrix in the vector map data, Represents the pixel value matrix in the satellite map data, Represents the training set, and Loss refers to the loss function.

[0055] Step S33: Send the test set to the trained convolutional neural network model for testing, and perform similarity calculation based on the vector map data in the test data and the trained satellite map data. If the obtained similarity value is greater than a preset threshold, the trained convolutional neural network model is used as a satellite map data desensitization model.

[0056] It can be understood that by calculating the similarity index in this step, the similarity between the satellite map data generated by the model and the real satellite map data can be objectively evaluated, so as to select a model with better performance. The entire process can be executed automatically, reducing manual intervention and improving efficiency and reliability. By setting a suitable threshold, the time cost of manual evaluation can be reduced and the degree of automation of the process can be improved. The similarity calculation in this step is calculated by the structural similarity index. It should be noted that this model can also introduce generative adversarial mechanisms, attention mechanisms, jump connections, etc. to improve the accuracy of the model.

[0057] Step S4: Process the vector map data to be desensitized in the second information, and send the processed second information to the trained satellite map data desensitization model to desensitize the satellite map data to obtain desensitized geographic objects of the satellite map corresponding to the specific geographic elements of the vector map.

[0058] It can be understood that this step edits the specific geographic elements in the vector map data through processing, and converts the edited vector map data into a format that matches the trained model input to ensure that the model can process the data correctly. The processed data is desensitized using the trained satellite map data desensitization model to achieve the conversion of vector map data to satellite map data. By converting and processing different types of map data, the desensitization of geographic objects in the satellite map corresponding to the specific geographic elements of the vector map is achieved. In this step, step S4 includes step S41, step S42 and step S43.

[0059] Step S41, searching the geographic elements of the vector map data to be desensitized in the second information to obtain vector geographic element data, wherein the vector geographic element data includes coordinate points of at least two geographic elements;

[0060] It is understandable that in this step, we will retrieve the vector map data to be desensitized in the second information to obtain the geographic element data therein. These geographic element data may include roads, buildings, water systems, etc. Through the analysis and processing of map data, the coordinate points of geographic elements can be accurately extracted to ensure the accuracy and reliability of the data. The extracted geographic element coordinate points can be used for further data analysis and processing, providing more information and reference for desensitized satellite map data.

[0061] Step S42, indexing the vector geographic element data according to preset geometric and semantic attributes, and processing the retrieved specific vector geographic element data in the vector map to be desensitized according to a preset processing method to obtain a processed pixel value matrix set;

[0062] It can be understood that in this step, the required geographic elements are accurately selected according to the preset coordinate points to ensure the accuracy of the data. According to the preset processing method, the selected geographic element data is flexibly edited and processed to meet the needs in different situations. The corresponding pixel value matrix set is obtained by performing segmentation, rasterization, normalization and other processing on the edited data, which provides a basis for the subsequent satellite map data desensitization. In this step, step S42 includes step S421, step S422, step S423 and step S424.

[0063] Step S421: If the preset processing method is to hide a specific geographic object in the satellite map to be desensitized, then retrieve and obtain the vector geographic element corresponding to the object, construct a subdivided grid with the geographic element as the center, which intersects with the geographic element and has the same size and resolution as the subdivided grid, delete the geographic element from the vector map data to be desensitized, and perform rasterization and pixel value normalization on the area contained in the subdivided grid in the vector map after the geographic element is deleted, so as to obtain a corresponding area pixel value matrix set;

[0064] It can be understood that this step improves the privacy and security of geographic information by hiding specific geographic elements in the vector map data to be desensitized. Unnecessary geographic element data is deleted, the complexity of the data is reduced, and the efficiency of data processing and the clarity of geographic information expression are improved. Constructing a subdivided grid and extracting a pixel value matrix set centered on the geographic element to be deleted is conducive to reducing unnecessary data calculation and desensitization processing, maintaining the accuracy of satellite maps, and improving the efficiency of satellite map desensitization, wherein the rasterization processing step is consistent with the resolution setting of step S26, wherein the pixel value normalization processing is consistent with the normalization method of step S26.

[0065] Step S422: If the preset processing method is to add a geographic object in the satellite map data to be desensitized, then the vector geographic element data corresponding to the object is added to the vector map data to be desensitized, and a subdivision grid intersecting with the geographic element and consistent with the size and resolution of the subdivision grid is constructed with the newly added geographic element as the center, and the area contained in the subdivision grid in the vector map after the geographic element is added is rasterized and pixel value normalized to obtain a corresponding area pixel value matrix set;

[0066] It is understandable that this step increases the geographic elements in the vector map data to be desensitized, thereby increasing the diversity of desensitization. New geographic element data is added to enrich the content and information of the data, making the data more complete and comprehensive. Constructing a subdivided grid and extracting a pixel value matrix set centered on the newly added geographic elements is conducive to reducing unnecessary data calculations and desensitization processing, maintaining the accuracy of satellite maps, and improving the efficiency of satellite map desensitization, wherein the rasterization processing step is consistent with the resolution setting of step S26, wherein the pixel value normalization processing is consistent with the normalization method of step S26.

[0067] Step S423: If the preset processing method is to change a specific geographic object in the satellite map data to be desensitized, the vector geographic element data corresponding to the object is edited to change its geometric form in the vector map to be desensitized, and a subdivision grid intersecting with the geographic element and consistent with the size and resolution of the subdivision grid is constructed with the deformed geographic element as the center, and the area contained in the subdivision grid in the vector map after the deformation of the geographic element is rasterized and pixel value normalized to obtain a corresponding area pixel value matrix set;

[0068] It is understandable that this step deforms the geographic elements in the vector map data to be desensitized, thereby improving the diversity of desensitization. The changed morphology of geographic elements facilitates the adjustment of geographic information expression and improves the privacy and security of geographic information. Constructing a subdivision grid and extracting a pixel value matrix set centered on the deformed geographic elements is conducive to reducing unnecessary data calculations and desensitization processing, maintaining the accuracy of satellite maps, and improving the efficiency of satellite map desensitization. It is understandable that changing specific geographic objects in the satellite map data to be desensitized can also be regarded as a combination of hiding specific geographic objects and creating new ones in the original position, wherein the step of rasterization processing is consistent with the resolution setting of step S26, wherein the pixel value normalization processing is consistent with the normalization method of step S26.

[0069] Step S424: If the preset processing method is to shift a specific geographic object in the satellite map data to be desensitized, the vector geographic element data corresponding to the object is moved as a whole to change its position in the vector map to be desensitized. A grid is constructed with the geographic elements before and after the shift as the center, intersecting with the geographic elements before and after the shift and consistent with the size and resolution of the grid, and the area contained in the grid in the vector map after the geographic elements are shifted is rasterized and pixel value normalized to obtain a corresponding area pixel value matrix set.

[0070] It is understandable that this step shifts the geographic elements in the vector map data to be desensitized, thereby improving the diversity of desensitization. The changed position of the geographic elements facilitates the adjustment of the expression of geographic information and improves the privacy and security of geographic information. Constructing a subdivided grid and extracting a pixel value matrix set centered on the geographic elements before and after the shift is conducive to reducing unnecessary data calculations and desensitization processing, maintaining the accuracy of satellite maps, and improving the efficiency of satellite map desensitization. It is understandable that shifting specific geographic objects in the satellite map data to be desensitized can be regarded as a combination of hiding specific geographic objects in the original position in the satellite map data and creating new specific geographic objects in the new position, wherein the step of rasterization processing is consistent with the resolution setting of step S26, wherein the pixel value normalization processing is consistent with the normalization method of step S26.

[0071] It should be noted that the combination of S421~S424 can also produce more diverse satellite map geographic object desensitization strategies, which also fall within the scope of protection of this patent.

[0072] Step S43: input the processed vector map data pixel value matrix set into the satellite map data desensitization model to generate desensitized satellite map data, replace the desensitized satellite map data with the satellite map data to be desensitized, and obtain the desensitized satellite map geographic objects corresponding to the desensitized vector map geographic elements.

[0073] It can be understood that in this step, the pixel value matrices in the pixel value matrix set generated after the vector map processing are sequentially input into the satellite map data desensitization model, and the generated pixel value matrix is ​​denormalized to obtain the desensitized satellite map data, and the desensitized satellite map data is replaced with the area corresponding to the subdivision grid constructed in S42 in the satellite map to be desensitized, so as to obtain the desensitized satellite map data corresponding to the desensitized vector map data. In this step, the desensitized satellite map data is generated using the satellite map data desensitization model so that it corresponds to the original desensitized vector map data. Embodiment 2:

[0074] like Figure 2 As shown, this embodiment provides a satellite map geographic object desensitization system based on vector map, see Figure 2 The system includes an acquisition unit 701 , a judgment unit 702 , a training unit 703 and a processing unit 704 .

[0075] An acquisition unit 701 is used to acquire first information and second information, wherein the first information includes vector map data and satellite map data to be trained, and the second information includes vector map data and satellite map data to be desensitized;

[0076] A judgment unit 702 is used to send the first information to a judgment model to judge whether the first information can construct a sample set. If the sample set can be constructed, the first information is matched and segmented to obtain a sample set corresponding to the first information, and the sample set is a training set and a test set for training a satellite map data desensitization model;

[0077] The judgment unit 702 includes a first processing subunit 7021 , a second processing subunit 7022 and a first judgment subunit 7023 .

[0078] A first processing subunit 7021 is used to determine first sub-information based on the first information, where the first sub-information includes a minimum resolvable distance and a scale of the vector map data to be trained;

[0079] A second processing subunit 7022 is used to determine second sub-information based on the first information, where the second sub-information includes the spatial resolution of the satellite map data to be trained;

[0080] The first judgment subunit 7023 is used to judge whether the first sub-information and the second sub-information can construct a sample set based on a preset judgment formula. If the sample set can be constructed, the vector map data and satellite map data to be trained in the first information are projected and transformed to obtain vector map data and satellite map data in the same coordinate system.

[0081] The determining unit 702 further includes a third processing subunit 7024 , a fourth processing subunit 7025 , a fifth processing subunit 7026 and a sixth processing subunit 7027 .

[0082] The third processing subunit 7024 is used to construct a subdivision grid for the vector map data and satellite map data to be trained in the first information based on a regular grid method, obtain the subdivision grid for the vector map data and satellite map data to be trained in the first information, and perform regional subdivision on the vector map data and satellite map data using the subdivision grid;

[0083] The fourth processing subunit 7025 is used to select a grid of intersection data of the vector map data and the satellite map data to be trained in the first information;

[0084] The fifth processing subunit 7026 is used to perform rasterization and normalization processing on the intersection data based on the spatial resolution of the satellite map data as the raster resolution, so as to obtain a set of a pixel value matrix of the vector map data to be trained and a pixel value matrix of the satellite map data to be trained;

[0085] The sixth processing subunit 7027 is used to divide all the pixel value matrix sets into a training set and a test set according to a preset ratio, so as to obtain a sample set corresponding to the constructed first information.

[0086] A training unit 703 is used to send a sample set corresponding to the first information to a preset convolutional encoding and decoding neural network model for training to obtain a trained satellite map data desensitization model;

[0087] The training unit 703 includes a first training subunit 7031 , a second training subunit 7032 and a third training subunit 7033 .

[0088] The first training subunit 7031 is used to build a convolutional neural network model based on a preset encoder and decoder;

[0089] The second training subunit 7032 is used to send the training in the sample set corresponding to the first information to the convolutional neural network model for training, wherein the training set is input into the convolutional neural network model in batches, the loss function of each batch of data is repeatedly calculated, and the model parameters are adjusted by using the gradient descent algorithm so that the model continuously approaches the preset objective function, thereby obtaining the trained convolutional neural network model;

[0090] The third training subunit 7033 is used to send the test set to the trained convolutional neural network model for testing, and perform similarity calculation based on the vector map data in the test data and the trained satellite map data. If the obtained similarity value is greater than a preset threshold, the trained convolutional neural network model is used as a satellite map data desensitization model.

[0091] The processing unit 704 is used to perform data processing on the second information, and send the processed second information to the trained satellite map data desensitization model to desensitize the satellite map data, so as to obtain the desensitized geographic objects of the satellite map corresponding to the specific geographic elements of the vector map.

[0092] The processing unit 704 includes a seventh processing sub-unit 7041 , an eighth processing sub-unit 7042 and a ninth processing sub-unit 7043 .

[0093] The seventh processing subunit 7041 is used to retrieve the vector map data to be desensitized in the second information to obtain vector geographic element data, wherein the vector geographic element data includes coordinate points of at least two geographic elements;

[0094] The eighth processing sub-unit 7042 is used to retrieve the geographic elements of the vector map data to be desensitized in the second information to obtain vector geographic element data, wherein the vector geographic element data contains coordinate points of at least two geographic elements; wherein the eighth processing sub-unit 7042 includes a second judgment sub-unit 70421, a third judgment sub-unit 70422, a fourth judgment sub-unit 70423 and a fifth judgment sub-unit 70424.

[0095] The second judgment subunit 70421 is used for, if the preset processing method is to hide a specific geographic object in the satellite map to be desensitized, retrieving and obtaining the vector geographic element corresponding to the object, constructing a subdivided grid with the geographic element as the center, intersecting with the geographic element and consistent with the size and resolution of the subdivided grid, deleting the geographic element from the vector map data to be desensitized, and performing rasterization processing and pixel value normalization processing on the area contained in the subdivided grid in the vector map after the geographic element is deleted, to obtain a corresponding area pixel value matrix set;

[0096] The third judgment subunit 70422 is used for adding the vector geographic element data corresponding to the object to the vector map data to be desensitized if the preset processing method is to add the geographic object in the satellite map data to be desensitized, constructing a subdivision grid with the newly added geographic element as the center, intersecting with the geographic element and consistent with the size and resolution of the subdivision grid, rasterizing and normalizing the pixel values ​​of the area contained in the subdivision grid in the vector map after the geographic element is added, and obtaining a corresponding area pixel value matrix set.

[0097] The fourth judgment subunit 70423 is used to edit the vector geographic element data corresponding to the object to change its geometric shape in the vector map to be desensitized if the preset processing method is to change the specific geographic object in the satellite map data to be desensitized, and to construct a subdivision grid intersecting with the geographic element and consistent with the size and resolution of the subdivision grid with the deformed geographic element as the center, and to rasterize and normalize the pixel values ​​of the area contained in the subdivision grid in the vector map after the deformation of the geographic element to obtain a corresponding area pixel value matrix set.

[0098] The fifth judgment subunit 70424 is used to, if the preset processing method is to shift a specific geographic object in the satellite map data to be desensitized, move the vector geographic element data corresponding to the object as a whole to change its position in the vector map to be desensitized. Construct a grid that intersects with the geographic elements before and after the shift and is consistent with the size and resolution of the grid, with the geographic elements before and after the shift as the center, and perform rasterization and pixel value normalization on the area contained in the grid in the vector map after the geographic elements are shifted, to obtain a corresponding area pixel value matrix set.

[0099] The ninth processing sub-unit 7043 inputs the processed vector map data pixel value matrix set into the satellite map data desensitization model to generate desensitized satellite map data, replaces the desensitized satellite map data with the satellite map data to be desensitized, and obtains the desensitized satellite map geographic objects corresponding to the desensitized vector map geographic elements.

[0100] It should be noted that, regarding the system in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0101] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

[0102] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A satellite map geographic object desensitization method based on vector map, characterized in that: include: Acquire first information and second information, wherein the first information includes vector map data and satellite map data to be trained, and the second information includes vector map data and satellite map data to be desensitized; The first information is sent to a judgment model to judge whether the first information can construct a sample set. If the sample set can be constructed, the first information is matched and segmented to obtain a sample set corresponding to the first information, and the sample set is a training set and a test set for training a satellite map data desensitization model; Sending the sample set corresponding to the first information to a preset convolutional encoding and decoding neural network model for training to obtain a trained satellite map data desensitization model; Processing the vector map data to be desensitized in the second information, and sending the processed second information to the trained satellite map data desensitization model to desensitize the satellite map data, so as to obtain a geographic desensitization object of the satellite map corresponding to the specific geographic elements of the vector map; The method of processing the vector map to be desensitized in the second information and sending the processed second information to the trained satellite map data desensitization model for satellite map data desensitization includes: Retrieving the geographic elements of the vector map data to be desensitized in the second information to obtain vector geographic element data, wherein the vector geographic element data includes coordinate points of at least two geographic elements; Selecting the vector geographic element data according to the preset coordinate points, and processing the selected vector geographic element data and the vector map data to be desensitized according to the preset processing method to obtain a processed pixel value matrix set; The processed vector map data pixel value matrix set is input into the satellite map data desensitization model to generate desensitized satellite map data, and the desensitized satellite map data is replaced with the satellite map data to be desensitized to obtain the desensitized satellite map geographic objects corresponding to the desensitized vector map geographic elements.

2. The satellite map geographic object desensitization method based on vector map according to claim 1 is characterized in that , sending the first information to the judgment model to judge whether the first information can construct a sample set, including: Determine first sub-information based on the first information, where the first sub-information includes a minimum resolvable distance and a scale of the vector map data to be trained; Determine second sub-information based on the first information, where the second sub-information includes a spatial resolution of the satellite map data to be trained; Based on a preset judgment formula, it is determined whether the first sub-information and the second sub-information can construct a sample set. If the sample set can be constructed, the vector map data and satellite map data to be trained in the first information are projected and transformed to obtain vector map data and satellite map data in the same coordinate system.

3. The satellite map geographic object desensitization method based on vector map according to claim 1 is characterized in that If the sample set can be constructed, the first information is matched and segmented to obtain a sample set corresponding to the first information, including: constructing a subdivision grid for the vector map data and satellite map data to be trained in the first information based on a regular grid method, obtaining the subdivision grid for the vector map data and satellite map data to be trained in the first information, and performing regional subdivision on the vector map data and satellite map data using the subdivision grid; Selecting a grid of intersection data of the vector map data and the satellite map data to be trained in the first information; Performing rasterization and normalization processing on the intersection data based on the spatial resolution of the satellite map data as the grid resolution, to obtain a set of pixel value matrices of the vector map data to be trained and pixel value matrices of the satellite map data to be trained; The set of all the pixel value matrices is divided into a training set and a test set according to a preset ratio to obtain a sample set corresponding to the constructed first information.

4. The satellite map geographic object desensitization method based on vector map according to claim 1 is characterized in that , based on the sample set corresponding to the first information, sending it to a preset convolutional encoding and decoding neural network model for training, and obtaining a trained satellite map data desensitization model, including: Build a convolutional neural network model based on the preset encoder and decoder; Sending the training set in the sample set corresponding to the first information to the convolutional neural network model for training, wherein the training set is input into the convolutional neural network model in batches, the loss function of each batch of data is repeatedly calculated, and the model parameters are adjusted using the gradient descent algorithm so that the model continuously approaches the preset objective function, thereby obtaining a trained convolutional neural network model; The test set is sent to the trained convolutional neural network model for testing, and a similarity calculation is performed based on the vector map data in the test data and the trained satellite map data. If the obtained similarity value is greater than a preset threshold, the trained convolutional neural network model is used as a satellite map data desensitization model.

5. A satellite map geographic object desensitization system based on vector map, characterized in that: include: an acquisition unit, configured to acquire first information and second information, wherein the first information includes vector map data and satellite map data to be trained, and the second information includes vector map data and satellite map data to be desensitized; A judgment unit, configured to send the first information to a judgment model to judge whether the first information can construct a sample set, and if the sample set can be constructed, match and segment the first information to obtain a sample set corresponding to the first information, wherein the sample set is a training set and a test set for training a satellite map data desensitization model; A training unit, configured to send a sample set corresponding to the first information to a preset convolutional encoding and decoding neural network model for training to obtain a trained satellite map data desensitization model; a processing unit, configured to process the vector map data to be desensitized in the second information, and send the processed second information to the trained satellite map data desensitization model to desensitize the satellite map data, so as to obtain a geographic desensitization object of the satellite map corresponding to the specific geographic element of the vector map; Wherein, the processing unit includes: A seventh processing subunit is used to retrieve the geographic elements of the vector map data to be desensitized in the second information to obtain vector geographic element data, wherein the vector geographic element data includes coordinate points of at least two geographic elements; An eighth processing subunit is used to select elements of the vector geographic element data according to preset coordinate points, and process the selected vector geographic element data and the vector map data to be desensitized according to a preset processing method to obtain a processed pixel value matrix set; The ninth processing sub-unit is used to input the processed vector map data pixel value matrix set into the satellite map data desensitization model to generate desensitized satellite map data, replace the desensitized satellite map data with the satellite map data to be desensitized, and obtain the desensitized satellite map geographical objects corresponding to the desensitized vector map geographical elements.

6. The satellite map geographic object desensitization system based on vector map according to claim 5, characterized in that: The judging unit comprises: A first processing subunit, configured to determine first sub-information based on the first information, wherein the first sub-information includes a minimum resolvable distance and a scale of the vector map data to be trained; a second processing subunit, configured to determine second sub-information based on the first information, wherein the second sub-information includes a spatial resolution of the satellite map data to be trained; The first judgment subunit is used to judge whether the first sub-information and the second sub-information can construct a sample set based on a preset judgment formula. If the sample set can be constructed, the vector map data and satellite map data to be trained in the first information are projected and transformed to obtain the vector map data and satellite map data in the same coordinate system.

7. The satellite map geographic object desensitization system based on vector map according to claim 5, characterized in that: The judging unit further includes: a third processing subunit, configured to construct a subdivision grid for the vector map data and satellite map data to be trained in the first information based on a regular grid method, obtain the subdivision grid for the vector map data and satellite map data to be trained in the first information, and perform regional subdivision on the vector map data and satellite map data using the subdivision grid; A fourth processing subunit, used for selecting a grid of intersection data of the vector map data and the satellite map data to be trained in the first information; a fifth processing subunit, configured to perform rasterization and normalization processing on the intersection data based on the spatial resolution of the satellite map data as the raster resolution, to obtain a set of a pixel value matrix of the vector map data to be trained and a pixel value matrix of the satellite map data to be trained; The sixth processing subunit is used to divide the set of all the pixel value matrices into a training set and a test set according to a preset ratio to obtain a sample set corresponding to the constructed first information.

8. The satellite map geographic object desensitization system based on vector map according to claim 5, characterized in that: The training unit comprises: A first training subunit, used for building a convolutional neural network model based on a preset encoder and decoder; A second training subunit is used to send the training set in the sample set corresponding to the first information to the convolutional neural network model for training, wherein the training set is input into the convolutional neural network model in batches, the loss function of each batch of data is repeatedly calculated, and the model parameters are adjusted using the gradient descent algorithm so that the model continuously approaches a preset objective function, thereby obtaining a trained convolutional neural network model; The third training subunit is used to send the test set to the trained convolutional neural network model for testing, and perform similarity calculation based on the vector map data in the test data and the trained satellite map data. If the obtained similarity value is greater than a preset threshold, the trained convolutional neural network model is used as a satellite map data desensitization model.

Images