GIS-based address data intelligent analysis and processing method and related device

Abstract

The present invention discloses a method and related devices for intelligent analysis and processing of address data based on GIS, and relates to the technical field of data processing. The method comprises: preprocessing the reported address information data; performing preliminary segmentation on the reported address information data, and performing segmentation and marking on the reported address information data; performing reduplication expansion on the reported address information data based on the segmentation and marking results, and performing address segmentation on the reported address information data; performing address type identification based on address segmentation information combined with address element index analysis; performing address matching based on address segmentation information and target address type information; extracting a real-life image of a building corresponding to a target matching address to perform building mask detection; performing contour extraction based on building mask information to generate a target building view, and performing address marking. The present invention improves the efficiency and accuracy of address data analysis and processing, and enables the analysis and processing of address data to achieve a more ideal effect.

Description

Technical Field

[0001] The present invention mainly relates to the technical field of data processing, and in particular to an intelligent analysis and processing method for address data based on GIS and related devices. Background Art

[0002] With the improvement of the level of smart city construction, the analysis and processing of address data has become an important support for various enterprises to carry out their business. The segmentation of address information data is a very important part of address data analysis and processing. At present, the segmentation of address information data is mainly achieved by manually constructing address dictionaries. However, this method relies too much on the professional quality and prior knowledge of address segmentation personnel, resulting in the inability to guarantee the accuracy of address segmentation, and the human cost invested is too high. At the same time, in the current address data analysis and processing, the recognition and processing of address types are rarely considered. The lack of recognition and processing of address types will lead to the inability to effectively improve the accuracy and efficiency of address matching, and when the user's address data input is slightly wrong, a satisfactory address matching result cannot be obtained. In the current address data processing, in order to locate the matched user location more quickly, the address is usually marked in the building view. Therefore, how to obtain an accurate building view image has also become a research focus. At present, the building outline is detected by selecting a feature threshold to generate a building view. However, this method has a large instability in the selection of feature thresholds, which easily leads to a large deviation between the obtained building outline and the actual situation, making the reliability of the final building view low. Summary of the invention

[0003] The purpose of the present invention is to overcome the deficiencies of the prior art. The present invention provides a method and related devices for intelligent analysis and processing of address data based on GIS, which improves the efficiency and accuracy of address data analysis and processing, and enables the analysis and processing of address data to achieve a more ideal effect.

[0004] In order to solve the above technical problems, the present invention provides a GIS-based address data intelligent analysis and processing method, which is applied to a client device and a geographic information system GIS server, wherein the client device is in communication connection with the GIS server; the method comprises:

[0005] The GIS server receives the registration address information data input by the user in the client device, and pre-processes the registration address information data to obtain the pre-processed registration address information data;

[0006] Preliminarily segmenting the preprocessed registered address information data to obtain the preliminarily segmented registered address information data, and performing word segmentation and tagging on the preliminarily segmented registered address information data to obtain the word segmentation and tagging results;

[0007] Based on the word segmentation and marking results, the preliminarily segmented registered address information data is expanded with reduplication to obtain the registered address information data after reduplication expansion, and the registered address information data after reduplication expansion is processed with address word segmentation based on the Bayesian probability model to obtain address word segmentation information;

[0008] Based on the address segmentation information combined with address element index analysis, address type identification is performed to obtain target address type information;

[0009] Performing address matching based on the address segmentation information and the target address type information to obtain a target matching address;

[0010] Calling the GIS map to extract the real-scene image of the building corresponding to the target matching address, and performing building mask detection on the real-scene image of the building to obtain building mask information;

[0011] Contour extraction is performed based on the building mask information to obtain building contour information, a target building view is generated based on the building contour information combined with a building label, and the target matching address is marked in a corresponding position of the target building view.

[0012] Optionally, preprocessing the installation address information data to obtain the preprocessed installation address information data includes:

[0013] Cleaning the reported address information data to obtain the cleaned reported address information data;

[0014] The cleaned address information data is standardized to obtain pre-processed address information data.

[0015] Optionally, the pre-processed registered address information data is preliminarily segmented to obtain the pre-segmented registered address information data, and the pre-segmented registered address information data is segmented and annotated to obtain the segmentation and annotated results, including:

[0016] Performing preliminary segmentation on the preprocessed application address information data based on a preset segmentation unit to obtain preliminary segmented application address information data;

[0017] Obtaining a preset reduplication annotation relationship table and segmentation training corpus, performing model statistical learning based on the preset reduplication annotation relationship table and segmentation training corpus, and obtaining a target reduplication annotation model;

[0018] The word segmentation and annotation are performed on the preliminarily segmented registered address information data based on the target reduplication annotation model to obtain a word segmentation and annotation result.

[0019] Optionally, the step of performing reduplication expansion on the preliminarily segmented registered address information data based on the word segmentation and marking result to obtain the reduplication-expanded registered address information data, and performing address segmentation processing on the reduplication-expanded registered address information data based on a Bayesian probability model to obtain address segmentation information includes:

[0020] Expanding the characters with the segmentation mark in the preliminary segmented address information data to obtain the expanded address information data;

[0021] Generate a bag-of-words frequency vector based on a preset stop word list, and construct a Bayesian probability model based on the bag-of-words frequency vector;

[0022] Performing topic-term distribution analysis based on the Bayesian probability model to obtain a feature word library;

[0023] Based on the word segmentation string, mapping identifiers are traversed for the segmentation training corpus to obtain a target mapping word segmentation identifier, and a word segmentation array is constructed based on the target mapping word segmentation identifier;

[0024] Based on the segmentation array and the feature word library, address segmentation processing is performed on the reported address information data after the reduplication expansion to obtain address segmentation information.

[0025] Optionally, the performing address type identification based on the address segmentation information in combination with address element index analysis to obtain target address type information includes:

[0026] Acquire a corresponding regional code based on the address segmentation information, compare index information based on the regional code with a pre-stored dictionary tree, and obtain a target dictionary tree;

[0027] Determine the point of interest information corresponding to the address segmentation information based on the target dictionary tree;

[0028] Extracting address elements from the address segmentation information based on an address element dictionary to obtain address element information, and performing spatial index analysis based on the address element information to obtain spatial index information;

[0029] Address type identification is performed based on the spatial index information in combination with the point of interest information to obtain target address type information.

[0030] Optionally, performing address matching based on the address segmentation information and the target address type information to obtain a target matching address includes:

[0031] Based on the address segmentation information and the target address type information, a syntax tree recursive operation is used to perform address similarity calculation in a standard address database to obtain a corresponding similarity calculation result, and based on the similarity calculation result, an address precise matching is performed to obtain an address precise matching result, and based on the address precise matching result, it is determined whether a target matching address is matched;

[0032] If it is determined that the target matching address is not matched, the address segmentation information and the target address type information are vectorized based on the word vector model to obtain an address segmentation information vector and a target address type information vector;

[0033] Inputting the address segmentation information vector and the target address type information into the address matching model to perform fuzzy matching processing based on edit distance to obtain a fuzzy matching result;

[0034] A target matching address is determined based on the fuzzy matching result and user feedback information obtained from the client device.

[0035] Optionally, performing building mask detection on the building real scene image to obtain building mask information includes:

[0036] Preprocessing the building real scene image to obtain a preprocessed building real scene image;

[0037] Binarizing the preprocessed building real scene image to obtain a binary building real scene image;

[0038] Performing corrosion processing on the binary building real scene image to obtain a binary building real scene image after corrosion processing, and performing inversion processing on the binary building real scene image after corrosion processing to obtain a binary building real scene image after inversion processing;

[0039] The binary building real scene image after inversion processing is subjected to connected domain filtering to obtain the building mask information.

[0040] Optionally, extracting contours based on the building mask information to obtain building contour information includes:

[0041] Performing triangular mesh processing based on the building real scene image to obtain a target triangular mesh, and performing semantic segmentation processing on the target triangular mesh to obtain a plurality of building clusters;

[0042] Extracting building contours using a plurality of building clusters based on the building mask information, performing merged line matching based on the building contours using a collinear equation to obtain a target merged line;

[0043] A two-dimensional outline of the building is generated based on the target merged line and the building outline to obtain building outline information.

[0044] In addition, the present invention also provides a GIS-based address data intelligent analysis and processing device, which is applied to a client device and a geographic information system GIS server, wherein the client device is communicatively connected with the GIS server; the device comprises:

[0045] Data preprocessing module: used for the GIS server to receive the registration address information data input by the user in the client device, and preprocess the registration address information data to obtain the preprocessed registration address information data;

[0046] Word segmentation and annotation module: used to perform preliminary segmentation on the preprocessed registered address information data, obtain the registered address information data after preliminary segmentation, and perform word segmentation and annotation on the registered address information data after preliminary segmentation to obtain the word segmentation and annotation results;

[0047] An address segmentation module is used to perform reduplication expansion on the reported address information data after preliminary segmentation based on the segmentation and marking results to obtain the reported address information data after reduplication expansion, and perform address segmentation processing on the reported address information data after reduplication expansion based on a Bayesian probability model to obtain address segmentation information;

[0048] Address type identification module: used to identify the address type based on the address segmentation information combined with address element index analysis to obtain target address type information;

[0049] Address matching module: used to perform address matching based on the address segmentation information and the target address type information to obtain a target matching address;

[0050] Building mask detection module: used to call GIS map to extract the building real scene image corresponding to the target matching address, and perform building mask detection on the building real scene image to obtain building mask information;

[0051] Building view generation module: used to extract the contour based on the building mask information, obtain the building contour information, generate the target building view based on the building contour information combined with the building label, and mark the target matching address in the corresponding position of the target building view.

[0052] In addition, the present invention also provides a GIS-based address data intelligent analysis and processing system, the system includes a client device and a geographic information system GIS server, the client device is communicatively connected to the GIS server; the system is configured to execute the above-mentioned GIS-based address data intelligent analysis and processing method.

[0053] In an embodiment of the present invention, the reduplication expansion of the registered address information data after preliminary segmentation is performed based on the segmentation and annotation results, which can avoid the ambiguity caused by the ambiguity of the segmentation; the address segmentation processing of the registered address information data after the reduplication expansion is performed based on the Bayesian probability model, which can improve the accuracy of the address segmentation, and reduce the human cost invested without affecting the reliability of the obtained address segmentation information. The address type identification is performed based on the address segmentation information combined with the address element index analysis, which can ensure the reliability of the address type identification and provide sufficient feature information for address matching. The address matching is performed based on the address segmentation information and the target address type information, respectively, and there are address precise matching and address fuzzy matching. Even if the user's address data input has an inaccurate description, efficient and accurate matching of the address can be achieved. The contour extraction is performed based on the obtained building mask information, and the building contour information is obtained to generate the target building view in combination with the building label, which can avoid the deviation of the obtained building view from the actual situation. The target matching address is marked in the corresponding position of the target building view, so that the matched user position can be located and viewed more quickly. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0055] Figure 1 It is a flowchart of the GIS-based address data intelligent analysis and processing method in an embodiment of the present invention;

[0056] Figure 2 It is a flowchart of an intelligent analysis and processing method of address data based on GIS in another embodiment of the present invention;

[0057] Figure 3 It is a schematic diagram of the structure of the GIS-based address data intelligent analysis and processing system in an embodiment of the present invention;

[0058] Figure 4 It is a schematic diagram of the structural composition of the GIS-based address data intelligent analysis and processing device in an embodiment of the present invention. DETAILED DESCRIPTION

[0059] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of 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.

[0060] Embodiment 1

[0061] See also Figure 1 , Figure 1 1 is a flow chart of a method for intelligent analysis and processing of address data based on GIS in an embodiment of the present invention. The method is applied to a client device and a geographic information system (GIS) server. The client device is in communication connection with the GIS server. The method includes:

[0062] S11: The GIS server receives the installation address information data input by the user in the client device, and pre-processes the installation address information data to obtain the pre-processed installation address information data;

[0063] In the specific implementation process of the present invention, the preprocessing of the reported address information data to obtain the preprocessed reported address information data includes: cleaning the reported address information data to obtain the cleaned reported address information data; normalizing the cleaned reported address information data to obtain the preprocessed reported address information data.

[0064] Specifically, this application generally involves the positioning of the appointment door-to-door registration service for equipment, such as broadband appointment registration or other appointment registration services that require the installation of equipment. The user selects the registration service in the corresponding application of the client device and fills in the registration address. The registration address data is filled in by the user according to the seven-level address standard input rules. The client device will transmit the registration address to the Geographic Information System (GIS) server. The GIS server receives the registration address information data entered by the user in the client device. The registration address information data is cleaned, and the registration address information data is cleaned of special characters and abnormal data to obtain the registration address information data after the cleaning process. The registration address information data after the cleaning process is normalized, that is, the registration address information data after the cleaning process is subjected to font unification and standardization to obtain the pre-processed registration address information data.

[0065] S12: Preliminarily segmenting the preprocessed reported address information data to obtain the preliminarily segmented reported address information data, and performing word segmentation and tagging on the preliminarily segmented reported address information data to obtain a word segmentation and tagging result;

[0066] In the specific implementation process of the present invention, the pre-processed registration address information data is initially segmented to obtain the registration address information data after the initial segmentation, and the registration address information data after the initial segmentation is segmented to obtain the word segmentation and marking results, including: based on a preset segmentation unit, the pre-processed registration address information data is initially segmented to obtain the registration address information data after the initial segmentation; obtaining a preset reduplication annotation relationship table and a segmentation training corpus, and performing model statistical learning based on the preset reduplication annotation relationship table and the segmentation training corpus to obtain a target reduplication annotation model; based on the target reduplication annotation model, the registration address information data after the initial segmentation is segmented to obtain the word segmentation and marking results.

[0067] Specifically, since the length of the registered address information data input by the user may be relatively long, the pre-processed registered address information data is first preliminarily segmented based on a preset segmentation unit. The preset segmentation unit contains preliminary segmentation rules for the registered address information data, such as segmenting the location of a certain district or a certain city in the pre-processed registered address information data, dividing it into two or three segments, etc., to obtain the registered address information data after preliminary segmentation, thereby improving the efficiency of subsequent word segmentation and tagging. A preset reduplication annotation relationship table and a segmentation training corpus are obtained, wherein the preset reduplication annotation relationship table contains the annotation relationship for the overlapping abbreviations, and the segmentation training corpus is a related address information corpus. Model statistical learning is performed based on the preset reduplication annotation relationship table and the segmentation training corpus, and the annotation relationship between each character and the overlapping abbreviations in the segmentation training corpus is statistically analyzed to obtain a target reduplication annotation model. The overlapping abbreviations in the segmentation training corpus are annotated according to the preset reduplication annotation relationship table, and feature extraction is performed on the annotated segmentation training corpus, that is, context features of the contexts in which the characters annotated by the overlapping abbreviations in the segmentation training corpus are extracted to obtain feature information, and model statistical learning is performed based on the feature information to obtain a final target reduplication annotation model. Based on the target reduplication annotation model, the preliminarily segmented registration address information data is segmented and annotated, that is, the overlapping abbreviations in the preliminarily segmented registration address information data are annotated to obtain a segmentation and annotating result.

[0068] S13: performing reduplication expansion on the preliminarily segmented registered address information data based on the word segmentation and marking results to obtain the reduplication-expanded registered address information data, and performing address word segmentation processing on the reduplication-expanded registered address information data based on a Bayesian probability model to obtain address word segmentation information;

[0069] In the specific implementation process of the present invention, the reduplication expansion is performed on the registered address information data after preliminary segmentation based on the word segmentation and annotation results to obtain the registered address information data after reduplication expansion, and the registered address information data after reduplication expansion is processed by address segmentation based on the Bayesian probability model to obtain the address segmentation information, including: reduplication expansion of the characters with word segmentation annotations in the registered address information data after preliminary segmentation to obtain the registered address information data after reduplication expansion; bag-of-words frequency vectors are generated based on a preset stop word list, and a Bayesian probability model is constructed based on the bag-of-words frequency vectors; topic-term distribution analysis is performed based on the Bayesian probability model to obtain a feature word library; mapping identifier traversal is performed on the segmentation training corpus based on the word segmentation string to obtain a target mapping word segmentation identifier, and a word segmentation array is constructed based on the target mapping word segmentation identifier; address segmentation is performed on the registered address information data after reduplication expansion based on the word segmentation array and the feature word library to obtain the address segmentation information.

[0070] Specifically, the characters with segmentation marks in the reported address information data after preliminary segmentation are expanded with reduplication, that is, the characters with segmentation marks are expanded at the position adjacent to the characters to obtain the reported address information data after reduplication expansion. After the reduplication expansion, ambiguity in subsequent segmentation processing can be avoided. If the reduplication expansion is not performed, it will cause ambiguity in the segmentation processing. If there is no word that needs reduplication expansion, no expansion is required. A bag-of-words frequency vector is generated based on a preset stop word list. The stop words in the preset stop word list specifically include the name of the province, city, and district to which the Chinese address belongs, directional words, quantifiers, etc. The stop words can be set accordingly according to the specific situation. The stop words are removed from the segmentation training corpus according to the preset stop word list to obtain the segmentation training corpus after the stop words are removed. A bag-of-words frequency vector is generated for the segmentation training corpus using natural language processing technology, and a Bayesian probability model is constructed based on the bag-of-words frequency vector. A vector space model is constructed based on the bag-of-words frequency vector. A Bayesian probability model is constructed based on the vector space model combined with a naive Bayes classifier. The Bayesian probability model is a three-layer Bayesian probability model, including three layers: words, topics, and documents. Based on the Bayesian probability model, a topic-term distribution analysis is performed to obtain a feature word library, and clustering is performed using a Bayesian probability model through a bag-of-words word frequency vector, wherein the clustering process is: a certain topic is selected with a certain probability, and a certain word in the bag-of-words word frequency vector is selected in this topic with a certain probability, the document to the topic obeys a multinomial distribution, and the topic to the word obeys a multinomial distribution, and a topic-term probability matrix and a document-topic probability matrix are obtained through clustering, and a topic-term distribution is obtained through the topic-term probability matrix and the document-topic probability matrix, and a topic-term set is obtained according to the topic-term distribution, and the relative term frequency of each topic term in the topic term set is calculated, and a feature word library is constructed according to the relative term frequency and the topic term set. Based on the word segmentation string, the segmentation training corpus is mapped and traversed, starting from the first character in the segmentation training corpus, a search and match is performed in the preset administrative area dictionary, the matched address information corpus is stored in the word segmentation string, and the identification code of the word segmentation string is obtained, the identification code includes the administrative level identification and the name identification, that is, the target mapping word segmentation identification is obtained, and a word segmentation array is constructed based on the target mapping word segmentation identification, and the word segmentation string and the corresponding target mapping word segmentation identification are stored in the array to construct the word segmentation array.Based on the segmentation array and the feature word library, the address segmentation processing is performed on the reported address information data after the reduplication expansion, and the address segmentation position is determined for the reported address information data after the reduplication expansion according to the corresponding segmentation identifier in the segmentation array and the feature word in the feature word library, that is, the first segmentation position is determined according to the segmentation array, and the second segmentation position is determined according to the feature word library, and the repeated positions in the first segmentation position and the second segmentation position are removed, which can better avoid the error of segmentation, and then the segmentation position between the two characters that appear repeatedly after the reduplication expansion is added to obtain the final segmentation position, and the address segmentation processing is performed on the reported address information data after the reduplication expansion according to the segmentation position to obtain the address segmentation information. The accuracy of address segmentation can be improved, and the human cost invested is reduced while the reliability of the obtained address segmentation information is not affected.

[0071] S14: performing address type identification based on the address segmentation information combined with address element index analysis to obtain target address type information;

[0072] In the specific implementation process of the present invention, the address type identification is performed based on the address segmentation information combined with the address element index analysis to obtain the target address type information, including: obtaining the corresponding regional code based on the address segmentation information, comparing the index information based on the regional code with the pre-stored dictionary tree, and obtaining the target dictionary tree; determining the point of interest information corresponding to the address segmentation information based on the target dictionary tree; extracting the address elements from the address segmentation information based on the address element dictionary to obtain the address element information, and performing spatial index analysis based on the address element information to obtain spatial index information; identifying the address type based on the spatial index information combined with the point of interest information to obtain the target address type information.

[0073] Specifically, the corresponding regional code is obtained based on the address segmentation information, and the address segmentation information is matched according to the pre-stored regional code dictionary. The regional code dictionary stores the regional name of the administrative division, the regional code of the administrative division, and the mapping relationship between the two to obtain the corresponding regional code. The regional code is compared with the index information of the pre-stored dictionary tree to obtain the target dictionary tree. Each pre-stored dictionary tree uses the regional code as index information. The regional code is compared with the index information of each dictionary tree to obtain the first dictionary tree and the second dictionary tree. The first dictionary tree includes the road information of the town and the second dictionary tree includes all the points of interest (Point of Point of Interest (POI) information, point of interest information includes longitude and latitude, category, name, etc., and the matched first dictionary tree and second dictionary tree are used as target dictionary trees. Based on the target dictionary tree, the point of interest information corresponding to the address segmentation information is determined, and the address segmentation information is input into the target dictionary tree. The first dictionary tree in the target dictionary tree performs road information matching, and the corresponding point of interest information is matched in the second dictionary tree according to the matched road information. The address element dictionary is used to extract the address element from the address segmentation information. The address element dictionary contains specific house numbers, building addresses, unit room numbers, etc. The address element matching of the address segmentation information is performed through the address element dictionary to obtain the address element information. Spatial index analysis is performed based on the address element information, and the address element information and several pre-stored spatial indexes are used to perform spatial index matching. Row matching, the spatial index contains several spatial areas, each of which is limited by different address elements. The target spatial area in the spatial index covering the address element information in the address segmentation information is determined, that is, the spatial index information is obtained. Based on the spatial index information combined with the point of interest information, the address type is identified. The spatial index information may contain several spatial areas. A geographic fence is generated according to the point of interest information. The area can be limited. According to the geographic fence, the spatial area outside the geographic fence is excluded from several spatial areas. The spatial area with the smallest spatial area in the remaining spatial area is taken as the target spatial area, and the address type corresponding to the target spatial area is taken as the target address type information. The address type information includes schools, office buildings, shopping malls, residences, etc. It can ensure the reliability of address type identification and provide sufficient and rich information for subsequent address matching processing.

[0074] S15: performing address matching based on the address segmentation information and the target address type information to obtain a target matching address;

[0075] In the specific implementation process of the present invention, the address matching is performed based on the address segmentation information and the target address type information to obtain the target matching address, including: based on the address segmentation information and the target address type information, using syntax tree recursive operation to perform address similarity calculation in a standard address database to obtain the corresponding similarity calculation result, and based on the similarity calculation result, perform address precise matching to obtain the address precise matching result, and determine whether the target matching address is matched based on the address precise matching result; if it is determined that the target matching address is not matched, the address segmentation information and the target address type information are vectorized based on a word vector model to obtain an address segmentation information vector and a target address type information vector; the address segmentation information vector and the target address type information are input into the address matching model to perform fuzzy matching processing based on the edit distance to obtain a fuzzy matching result; based on the fuzzy matching result and combined with user feedback information obtained from the client device, the target matching address is determined.

[0076] Specifically, based on the address segmentation information and the target address type information, a syntax tree recursive operation is used to calculate the similarity of the address in the standard address database, the address segmentation information is indexed, the address information in the standard address database is indexed, and the corresponding index information is obtained. A corresponding first syntax tree is generated according to the address segmentation information, and a second syntax tree is generated according to the address information in the standard address database. The end node of the syntax tree is a numerical expression of the information generated based on the index information. A recursive operation is performed from the end node of the first syntax tree to the top node to obtain a first numerical expression corresponding to the address segmentation. A recursive operation is performed from the end node of the second syntax tree to the top node to obtain a second numerical expression corresponding to the address segmentation. A first similarity is calculated based on the first numerical expression and the second numerical expression to improve the accuracy of the similarity calculation, that is, the similarity between the address segmentation information and each address in the standard address database is calculated, and a similarity calculation based on the Euclidean distance is performed on the target address type information and the address type in the standard address database to obtain a second similarity, and a corresponding similarity calculation result is obtained. Based on the similarity calculation result, the address is accurately matched, that is, the second similarity is compared with the preset type similarity threshold, and the address information that reaches the preset type similarity threshold is retained. In the retained address information, the first similarity is compared with the preset similarity threshold. If the similarity in the retained address information reaches the preset similarity threshold, the address information with the greatest similarity is used as the target matching address. If the similarities in the retained address information all fail to reach the preset similarity threshold, the target matching address is not matched. If it is determined that the target matching address is not matched, the address word segmentation information and the target address type information are vectorized based on the word vector model. The word vector model uses the Word2vec model in the natural language processing library. The Word2vec model is an unsupervised neural network language model, which is essentially a fully connected neural network. It is modeled according to the distribution of the current word and its local context to obtain the word vector of the current word. The model training is rich in negative sampling training to obtain the address word segmentation information vector and the target address type information vector. The address segmentation information vector and the target address type information are input into the address matching model for fuzzy matching based on edit distance. The address matching model adopts the Enhanced Sequential Inference Model (ESIM) model. The ESIM model is an interactive deep text matching model. It comprehensively utilizes the characteristics of convolutional neural networks and recurrent neural networks in model construction. It can perform local modeling according to the contextual dependencies of address elements to achieve accurate address matching and obtain fuzzy matching results.The target matching address is determined based on the fuzzy matching result combined with the user feedback information obtained from the user terminal device, and the fuzzy matching result is transmitted to the client device for display to the user. The user selects the corresponding address information in the fuzzy matching result and uses it as the user feedback information to determine the target matching address. Even if the user's address data input is inaccurately described, efficient and accurate matching of the address can be achieved.

[0077] S16: calling the GIS map to extract the building real scene image corresponding to the target matching address, and performing building mask detection on the building real scene image to obtain building mask information;

[0078] In the specific implementation process of the present invention, the building mask detection is performed on the real-scene image of the building to obtain the building mask information, including: preprocessing the real-scene image of the building to obtain the preprocessed real-scene image of the building; binarizing the preprocessed real-scene image of the building to obtain the binary real-scene image of the building; corroding the binary real-scene image of the building to obtain the corroded binary real-scene image, and inverting the corroded binary real-scene image to obtain the inverted binary real-scene image of the building; and performing connected domain filtering on the inverted binary real-scene image of the building to obtain the building mask information.

[0079] Specifically, the GIS server calls the GIS map to extract the real-scene image of the building corresponding to the target matching address, and the real-scene image of the building is the real-scene image of the community building within the preset range of the target matching address. The real-scene image of the building is preprocessed, and the real-scene image of the building is denoised and image enhanced to obtain the preprocessed real-scene image of the building. The preprocessed real-scene image of the building is binarized, and the threshold method can be used for binarization to obtain the binary real-scene image of the building. The binary real-scene image of the building is corroded, and each pixel point is corroded according to the convolution kernel in a preset traversal order, wherein the convolution kernel is defined as a circle with a diameter of several pixels, and the center point is set as an anchor point. The preset traversal order includes taking the pixel point at the set position as the starting point, traversing all pixels in the order of rows first and columns later, and obtaining the binary real-scene image of the building after corrosion. The binary real-scene image of the building after the corrosion processing is subjected to inversion processing, the maximum depth value and the minimum depth value of the area described by each pixel point in the binary real-scene image of the building after the corrosion processing are calculated, and the inversion correction value is calculated according to the maximum depth value and the minimum depth value and the pixel value of each pixel point, and the binary real-scene image of the building after the corrosion processing is subjected to inversion processing according to the inversion correction value to obtain the binary real-scene image of the building after the inversion processing. The binary real-scene image of the building after the inversion processing is subjected to connected domain filtering, and the binary real-scene image of the building after the inversion processing is projected onto the XOY plane for connected domain detection to obtain each binary patch, each binary patch represents a building, and each binary patch is subjected to morphological closing operation to obtain the mask information of the building.

[0080] S17: Perform contour extraction based on the building mask information to obtain building contour information, generate a target building view based on the building contour information combined with the building label, and mark the target matching address in a corresponding position of the target building view.

[0081] In the specific implementation process of the present invention, the contour extraction based on the building mask information to obtain the building contour information includes: performing triangular meshing processing based on the real-scene image of the building to obtain a target triangular mesh, and performing semantic segmentation processing on the target triangular mesh to obtain a number of building clusters; extracting building contour lines based on the building mask information using a number of building clusters, performing merged line matching based on the building contour lines using a collinear equation to obtain a target merged line; generating a two-dimensional contour of the building based on the target merged line and the building contour line to obtain the building contour information.

[0082] Specifically, a triangular mesh processing is performed based on the real-scene image of the building, and a polygon growth algorithm is used to extract the vertices of the polygon as the initial point set, and then a triangulation algorithm is used to generate a triangular mesh to obtain a target triangular mesh. The target triangular mesh is semantically segmented, and the target triangular mesh is three-dimensionally segmented according to the building category to obtain a triangular mesh of the building category. The triangular mesh of the building category is clustered according to the semantic instance label to obtain a number of building clusters, so as to ensure the accuracy and efficiency of the building information extraction and make the processing content of the two-dimensional building more targeted. Based on the building mask information, the building contour line is extracted using a number of building clusters, and a straight line segment detection operator is used to establish a buffer zone according to a preset size with the contour line of the building mask as the center. The straight line segment extracted from the building cluster intersecting with the buffer zone is the building contour line. Based on the building contour line, the collinear equation is used to match the merged line, and the points obtained by interpolating the building contour line are projected by the collinear equation, and the projected points are fitted with two-dimensional line segments to obtain the target merged line. Based on the target merged line and the building contour line, the two-dimensional contour of the building is generated, and the building contour line and the target merged line are extended on the two-dimensional plane to obtain the sub-area corresponding to the building cluster. The internal and external judgment processing is performed on each sub-area by the ray method, that is, whether the area belongs to the inside or outside of the building, so as to avoid that the sub-area of ​​the two-dimensional contour of the building cannot completely cover the building monomer. The two-dimensional contour information of the building is determined based on the determined internal area and external area, that is, the building contour information is obtained, and the accuracy and efficiency of the building contour generation are improved. When high and low buildings are staggered and densely distributed, the building boundary can be well detected, and the building with incomplete edge of the picture can also be detected. Based on the building contour information and the building label, the target building view is generated, and the building label includes the building value label and the coverage level label, and the target matching address is marked in the corresponding position of the target building view. The address mark of the building view is introduced, which can better realize the improvement of the address data, and the marked target building view is transmitted to the client device for viewing, so that the matched user position can be viewed more quickly and clearly.

[0083] In an embodiment of the present invention, the reduplication expansion of the registered address information data after preliminary segmentation is performed based on the segmentation and annotation results, which can avoid the ambiguity caused by the ambiguity of the segmentation; the address segmentation processing of the registered address information data after the reduplication expansion is performed based on the Bayesian probability model, which can improve the accuracy of the address segmentation, and reduce the human cost invested without affecting the reliability of the obtained address segmentation information. The address type identification is performed based on the address segmentation information combined with the address element index analysis, which can ensure the reliability of the address type identification and provide sufficient feature information for address matching. The address matching is performed based on the address segmentation information and the target address type information, respectively, and there are address precise matching and address fuzzy matching. Even if the user's address data input has an inaccurate description, efficient and accurate matching of the address can be achieved. The contour extraction is performed based on the obtained building mask information, and the building contour information is obtained to generate the target building view in combination with the building label, which can avoid the deviation of the obtained building view from the actual situation. The target matching address is marked in the corresponding position of the target building view, so that the matched user position can be located and viewed more quickly.

[0084] Embodiment 2

[0085] See also Figure 2 , Figure 2 1 is a flow chart of a method for intelligent analysis and processing of address data based on GIS in another embodiment of the present invention. The method is applied to a client device and a geographic information system (GIS) server. The client device is in communication connection with the GIS server. The method includes:

[0086] S201: The GIS server receives the installation address information data input by the user in the client device, and pre-processes the installation address information data to obtain the pre-processed installation address information data;

[0087] S202: Preliminarily segmenting the preprocessed registered address information data to obtain the preliminarily segmented registered address information data, and performing word segmentation and tagging on the preliminarily segmented registered address information data to obtain a word segmentation and tagging result;

[0088] S203: Expanding the characters with the segmentation mark in the preliminary segmented address information data to obtain the expanded address information data;

[0089] In the specific implementation process of the present invention, the characters with segmentation marks in the reported address information data after preliminary segmentation are expanded, that is, the characters with segmentation marks are expanded at the position adjacent to the characters to obtain the reported address information data after the expansion of the reduplication. After the expansion of the reduplication, ambiguity in the subsequent segmentation processing can be avoided. If the reduplication expansion is not performed, it will cause ambiguity in the segmentation processing. If there is no word that needs reduplication expansion, no expansion is required.

[0090] S204: Generate a bag-of-words frequency vector based on a preset stop word list, and construct a Bayesian probability model based on the bag-of-words frequency vector;

[0091] In the specific implementation process of the present invention, a bag-of-words frequency vector is generated based on a preset stop word list. The stop words in the preset stop word list specifically include the name of the province, city, and district to which the Chinese address belongs, directional words and quantifiers, etc. The stop words can be set accordingly according to the specific situation. The stop words are removed from the segmentation training corpus according to the preset stop word list to obtain the segmentation training corpus after the stop words are removed. A bag-of-words frequency vector is generated for the segmentation training corpus using natural language processing technology, and a Bayesian probability model is constructed based on the bag-of-words frequency vector, a vector space model is constructed based on the bag-of-words frequency vector, and a Bayesian probability model is constructed based on the vector space model combined with a naive Bayes classifier. The Bayesian probability model is a three-layer Bayesian probability model, including three layers: words, topics, and documents.

[0092] S205: Performing topic-term distribution analysis based on the Bayesian probability model to obtain a feature word library;

[0093] In the specific implementation process of the present invention, a topic-term distribution analysis is performed based on the Bayesian probability model to obtain a feature word library, and clustering is performed using a Bayesian probability model through a bag-of-words word frequency vector, wherein the clustering process is: a certain topic is selected with a certain probability, and a certain word in the bag-of-words word frequency vector is selected in this topic with a certain probability, the document to topic obeys a polynomial distribution, and the topic to word obeys a polynomial distribution, and a topic-term probability matrix and a document-topic probability matrix are obtained through clustering, and a topic-term distribution is obtained through the topic-term probability matrix and the document-topic probability matrix, and a topic-term set is obtained according to the topic-term distribution, and the relative term frequency of each topic term in the topic term set is calculated, and a feature word library is constructed according to the relative term frequency and the topic term set.

[0094] S206: traversing the segmentation training corpus based on the segmentation string to obtain a target mapping segmentation identifier, and constructing a segmentation array based on the target mapping segmentation identifier;

[0095] In the specific implementation process of the present invention, the segmentation training corpus is mapped and traversed based on the segmentation string, starting from the first character in the segmentation training corpus, a search and match is performed in the preset administrative area dictionary, the matched address information corpus is stored in the segmentation string, and the identification code of the segmentation string is obtained, the identification code includes the administrative level identification and the name identification, that is, the target mapping segmentation identification is obtained, and a segmentation array is constructed based on the target mapping segmentation identification, and the segmentation string and the corresponding target mapping segmentation identification are stored in the array to construct the segmentation array.

[0096] S207: performing address segmentation processing on the reported address information data after the reduplication expansion based on the segmentation array and the feature word library to obtain address segmentation information;

[0097] In the specific implementation process of the present invention, address segmentation processing is performed on the reported address information data after the reduplication expansion based on the segmentation array and the feature word library, and the address segmentation position is determined for the reported address information data after the reduplication expansion according to the corresponding segmentation identifier in the segmentation array and the feature word in the feature word library, that is, the first segmentation position is determined according to the segmentation array, and the second segmentation position is determined according to the feature word library, and the repeated positions in the first segmentation position and the second segmentation position are removed, which can better avoid the error of segmentation, and then the segmentation position between the two characters that appear repeatedly after the reduplication expansion is added to obtain the final segmentation position, and the address segmentation processing is performed on the reported address information data after the reduplication expansion according to the segmentation position to obtain the address segmentation information. The accuracy of address segmentation can be improved, and the human cost invested is reduced while the reliability of the obtained address segmentation information is not affected.

[0098] S208: Acquire a corresponding regional code based on the address segmentation information, compare index information based on the regional code with a pre-stored dictionary tree, and obtain a target dictionary tree;

[0099] In the specific implementation process of the present invention, the corresponding regional code is obtained based on the address segmentation information, and the address segmentation information is matched according to the pre-stored regional code dictionary. The regional code dictionary stores the regional name of the administrative division, the regional code of the administrative division, and the mapping relationship between the two, so as to obtain the corresponding regional code. Based on the regional code, the index information is compared with the pre-stored dictionary tree to obtain the target dictionary tree. Each pre-stored dictionary tree uses the regional code as the index information. The regional code is compared with the index information of each dictionary tree to obtain the first dictionary tree and the second dictionary tree. The first dictionary tree includes the road information of the town and the second dictionary tree includes all the points of interest (POI) information of the town and the street. The point of interest information includes latitude and longitude, category and name, etc. The matched first dictionary tree and second dictionary tree are used as the target dictionary tree.

[0100] S209: Determine the point of interest information corresponding to the address segmentation information based on the target dictionary tree;

[0101] In the specific implementation process of the present invention, the point of interest information corresponding to the address segmentation information is determined based on the target dictionary tree, the address segmentation information is input into the target dictionary tree, the first dictionary tree in the target dictionary tree matches the road information, and the corresponding point of interest information is matched in the second dictionary tree according to the matched road information.

[0102] S210: extracting address elements from the address segmentation information based on an address element dictionary to obtain address element information, and performing spatial index analysis based on the address element information to obtain spatial index information;

[0103] In the specific implementation process of the present invention, address elements are extracted from the address segmentation information based on the address element dictionary, which contains specific house numbers, building addresses, unit room numbers, etc. The address element dictionary is used to match the address element of the address segmentation information to obtain the address element information. Spatial index analysis is performed based on the address element information, and matching is performed based on the address element information and a plurality of pre-stored spatial indexes, the spatial index contains a plurality of spatial regions, each of which is defined by different address elements, and the target spatial region in the spatial index covering the address element information in the address segmentation information is determined, that is, the spatial index information is obtained.

[0104] S211: performing address type identification based on the spatial index information and the point of interest information to obtain target address type information;

[0105] In the specific implementation process of the present invention, address type identification is performed based on the spatial index information combined with the point of interest information. The spatial index information may include several spatial areas. A geo-fence is generated according to the point of interest information, and an area can be limited. The spatial area outside the geo-fence is excluded from several spatial areas according to the geo-fence, and the spatial area with the smallest spatial area in the remaining spatial areas is taken as the target spatial area, and the address type corresponding to the target spatial area is taken as the target address type information. The address type information includes school, office building, shopping mall, residence, etc. It can ensure the reliability of address type identification and provide sufficient and rich information for subsequent address matching processing.

[0106] S212: performing address matching based on the address segmentation information and the target address type information to obtain a target matching address;

[0107] S213: calling the GIS map to extract the building real scene image corresponding to the target matching address, and performing building mask detection on the building real scene image to obtain building mask information;

[0108] S214: Perform contour extraction based on the building mask information to obtain building contour information, generate a target building view based on the building contour information combined with the building label, and mark the target matching address in a corresponding position of the target building view.

[0109] In an embodiment of the present invention, the reduplication expansion of the registered address information data after preliminary segmentation is performed based on the segmentation and annotation results, which can avoid the ambiguity caused by the ambiguity of the segmentation; the address segmentation processing of the registered address information data after the reduplication expansion is performed based on the Bayesian probability model, which can improve the accuracy of the address segmentation, and reduce the human cost invested without affecting the reliability of the obtained address segmentation information. The address type identification is performed based on the address segmentation information combined with the address element index analysis, which can ensure the reliability of the address type identification and provide sufficient feature information for address matching. The address matching is performed based on the address segmentation information and the target address type information, respectively, and there are address precise matching and address fuzzy matching. Even if the user's address data input has an inaccurate description, efficient and accurate matching of the address can be achieved. The contour extraction is performed based on the obtained building mask information, and the building contour information is obtained to generate the target building view in combination with the building label, which can avoid the deviation of the obtained building view from the actual situation. The target matching address is marked in the corresponding position of the target building view, so that the matched user position can be located and viewed more quickly.

[0110] Embodiment 3

[0111] See also Figure 3 , Figure 3 It is a schematic diagram of the structural composition of the GIS-based address data intelligent analysis and processing system in an embodiment of the present invention, the system includes a client device 31 and a geographic information system GIS server 32, and the client device 31 is communicatively connected with the GIS server 32; the system is configured to execute the GIS-based address data intelligent analysis and processing method described in the above embodiment.

[0112] In the specific implementation process of the present invention, the client device 31 can be a smart terminal. The user selects the application service in the corresponding application of the client device 31, fills in the application address information data in the application service interface, and then the client device 31 transmits the application address information data to the GIS server 32. The GIS server 32 is used to analyze and process the received application address information data, and then transmits the analysis and processing results to the client device 31 for display. Figure 3 The system shown does not constitute a limitation of all components, and may include more or less components than shown, or combine certain components.

[0113] In the specific implementation process of the present invention, the specific implementation methods of the system items can be referred to the above embodiments, which will not be repeated here.

[0114] In an embodiment of the present invention, the reduplication expansion of the registered address information data after preliminary segmentation is performed based on the segmentation and annotation results, which can avoid the ambiguity caused by the ambiguity of the segmentation; the address segmentation processing of the registered address information data after the reduplication expansion is performed based on the Bayesian probability model, which can improve the accuracy of the address segmentation, and reduce the human cost invested without affecting the reliability of the obtained address segmentation information. The address type identification is performed based on the address segmentation information combined with the address element index analysis, which can ensure the reliability of the address type identification and provide sufficient feature information for address matching. The address matching is performed based on the address segmentation information and the target address type information, respectively, and there are address precise matching and address fuzzy matching. Even if the user's address data input has an inaccurate description, efficient and accurate matching of the address can be achieved. The contour extraction is performed based on the obtained building mask information, and the building contour information is obtained to generate the target building view in combination with the building label, which can avoid the deviation of the obtained building view from the actual situation. The target matching address is marked in the corresponding position of the target building view, so that the matched user position can be located and viewed more quickly.

[0115] Embodiment 4

[0116] See also Figure 4 , Figure 4 1 is a schematic diagram of the structure of a GIS-based address data intelligent analysis and processing device in an embodiment of the present invention, the device is applied to a client device and a geographic information system GIS server, the client device is connected to the GIS server in communication; the device includes:

[0117] Data preprocessing module 41: used for the GIS server to receive the registration address information data input by the user in the client device, and preprocess the registration address information data to obtain the preprocessed registration address information data;

[0118] The word segmentation and marking module 42 is used to perform preliminary segmentation on the preprocessed registered address information data to obtain the registered address information data after preliminary segmentation, and perform word segmentation and marking on the registered address information data after preliminary segmentation to obtain the word segmentation and marking result;

[0119] The address segmentation module 43 is used to perform reduplication expansion on the reported address information data after the preliminary segmentation based on the segmentation and marking results to obtain the reported address information data after the reduplication expansion, and perform address segmentation processing on the reported address information data after the reduplication expansion based on the Bayesian probability model to obtain address segmentation information;

[0120] Address type identification module 44: used to identify the address type based on the address segmentation information combined with address element index analysis to obtain target address type information;

[0121] Address matching module 45: used to perform address matching based on the address segmentation information and the target address type information to obtain a target matching address;

[0122] The building mask detection module 46 is used to call the GIS map to extract the building real scene image corresponding to the target matching address, and perform building mask detection on the building real scene image to obtain building mask information;

[0123] Building view generation module 47: used to extract contours based on the building mask information to obtain building contour information, generate a target building view based on the building contour information combined with the building label, and mark the target matching address in the corresponding position of the target building view.

[0124] In the specific implementation process of the present invention, the specific implementation methods of the device items can be referred to the above embodiments, which will not be repeated here.

[0125] In an embodiment of the present invention, the reduplication expansion of the registered address information data after preliminary segmentation is performed based on the segmentation and annotation results, which can avoid the ambiguity caused by the ambiguity of the segmentation; the address segmentation processing of the registered address information data after the reduplication expansion is performed based on the Bayesian probability model, which can improve the accuracy of the address segmentation, and reduce the human cost invested without affecting the reliability of the obtained address segmentation information. The address type identification is performed based on the address segmentation information combined with the address element index analysis, which can ensure the reliability of the address type identification and provide sufficient feature information for address matching. The address matching is performed based on the address segmentation information and the target address type information, respectively, and there are address precise matching and address fuzzy matching. Even if the user's address data input has an inaccurate description, efficient and accurate matching of the address can be achieved. The contour extraction is performed based on the obtained building mask information, and the building contour information is obtained to generate the target building view in combination with the building label, which can avoid the deviation of the obtained building view from the actual situation. The target matching address is marked in the corresponding position of the target building view, so that the matched user position can be located and viewed more quickly.

[0126] A person skilled in the art may understand that all or part of the steps in the various methods of the above embodiments may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, and the storage medium may include: a read-only memory (ROM), a random access memory (RAM), a disk or an optical disk, etc.

[0127] In addition, the above is a detailed introduction to a GIS-based address data intelligent analysis and processing method and related devices provided by an embodiment of the present invention. Specific examples are used herein to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scopes. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A method for intelligent analysis and processing of address data based on GIS, characterized in that: Applied to a client device and a geographic information system (GIS) server, the client device is in communication connection with the GIS server; the method comprises: The GIS server receives the registration address information data input by the user in the client device, and pre-processes the registration address information data to obtain the pre-processed registration address information data; Preliminarily segmenting the preprocessed registered address information data to obtain the preliminarily segmented registered address information data, and performing word segmentation and tagging on the preliminarily segmented registered address information data to obtain the word segmentation and tagging results; Based on the word segmentation and marking results, the preliminarily segmented registered address information data is expanded with reduplication to obtain the registered address information data after reduplication expansion, and the registered address information data after reduplication expansion is processed with address word segmentation based on the Bayesian probability model to obtain address word segmentation information; Based on the address segmentation information combined with address element index analysis, address type identification is performed to obtain target address type information; Performing address matching based on the address segmentation information and the target address type information to obtain a target matching address; Calling the GIS map to extract the real-scene image of the building corresponding to the target matching address, and performing building mask detection on the real-scene image of the building to obtain building mask information; Performing contour extraction based on the building mask information to obtain building contour information, generating a target building view based on the building contour information combined with a building label, and marking the target matching address in a corresponding position of the target building view; The preliminarily segmenting the preprocessed registration address information data to obtain the preliminarily segmented registration address information data, and performing word segmentation and marking on the preliminarily segmented registration address information data to obtain the word segmentation and marking results, includes: preliminarily segmenting the preprocessed registration address information data based on a preset segmentation unit to obtain the preliminarily segmented registration address information data; obtaining a preset reduplication marking relationship table and a segmentation training corpus, performing model statistical learning based on the preset reduplication marking relationship table and the segmentation training corpus to obtain a target reduplication marking model; performing word segmentation and marking on the preliminarily segmented registration address information data based on the target reduplication marking model to obtain the word segmentation and marking results; The method of performing reduplication expansion on the registration address information data after preliminary segmentation based on the word segmentation and annotation results to obtain the registration address information data after reduplication expansion, and performing address segmentation processing on the registration address information data after reduplication expansion based on a Bayesian probability model to obtain address segmentation information, includes: performing reduplication expansion on the characters with word segmentation annotations in the registration address information data after preliminary segmentation to obtain the registration address information data after reduplication expansion; generating a bag-of-words frequency vector based on a preset stop word list, and constructing a Bayesian probability model based on the bag-of-words frequency vector; performing a topic-term distribution analysis based on the Bayesian probability model to obtain a feature word library; performing mapping identifier traversal on the segmentation training corpus based on the word segmentation character string to obtain a target mapping word segmentation identifier, and constructing a word segmentation array based on the target mapping word segmentation identifier; performing address segmentation processing on the registration address information data after reduplication expansion based on the word segmentation array and the feature word library to obtain address segmentation information.

2. The GIS-based address data intelligent analysis and processing method according to claim 1 is characterized in that: The preprocessing of the installation address information data to obtain the preprocessed installation address information data includes: Cleaning the reported address information data to obtain the cleaned reported address information data; The cleaned address information data is standardized to obtain pre-processed address information data.

3. The GIS-based address data intelligent analysis and processing method according to claim 1 is characterized in that: The address type identification is performed based on the address segmentation information in combination with the address element index analysis to obtain the target address type information, including: Acquire a corresponding regional code based on the address segmentation information, compare index information based on the regional code with a pre-stored dictionary tree, and obtain a target dictionary tree; Determine the point of interest information corresponding to the address segmentation information based on the target dictionary tree; Extracting address elements from the address segmentation information based on an address element dictionary to obtain address element information, and performing spatial index analysis based on the address element information to obtain spatial index information; Address type identification is performed based on the spatial index information in combination with the point of interest information to obtain target address type information.

4. The GIS-based address data intelligent analysis and processing method according to claim 1 is characterized in that: The performing address matching based on the address segmentation information and the target address type information to obtain the target matching address includes: Based on the address segmentation information and the target address type information, a syntax tree recursive operation is used to perform address similarity calculation in a standard address database to obtain a corresponding similarity calculation result, and based on the similarity calculation result, an address precise matching is performed to obtain an address precise matching result, and based on the address precise matching result, it is determined whether a target matching address is matched; If it is determined that the target matching address is not matched, the address segmentation information and the target address type information are vectorized based on the word vector model to obtain an address segmentation information vector and a target address type information vector; Inputting the address segmentation information vector and the target address type information into the address matching model to perform fuzzy matching processing based on edit distance to obtain a fuzzy matching result; A target matching address is determined based on the fuzzy matching result and user feedback information obtained from the client device.

5. The GIS-based address data intelligent analysis and processing method according to claim 1 is characterized in that: The performing building mask detection on the building real scene image to obtain building mask information includes: Preprocessing the building real scene image to obtain a preprocessed building real scene image; Binarizing the preprocessed building real scene image to obtain a binary building real scene image; Performing corrosion processing on the binary building real scene image to obtain a binary building real scene image after corrosion processing, and performing inversion processing on the binary building real scene image after corrosion processing to obtain a binary building real scene image after inversion processing; The binary building real scene image after inversion processing is subjected to connected domain filtering to obtain the building mask information.

6. The GIS-based address data intelligent analysis and processing method according to claim 1 is characterized in that: The extracting the contour based on the building mask information to obtain the building contour information includes: Performing triangular mesh processing based on the building real scene image to obtain a target triangular mesh, and performing semantic segmentation processing on the target triangular mesh to obtain a plurality of building clusters; Extracting building contours using a plurality of building clusters based on the building mask information, performing merged line matching based on the building contours using a collinear equation to obtain a target merged line; A two-dimensional outline of the building is generated based on the target merged line and the building outline to obtain building outline information.

7. A GIS-based address data intelligent analysis and processing device, characterized in that: Applicable to a client device and a geographic information system (GIS) server, wherein the client device is connected to the GIS server for communication; the device comprises: Data preprocessing module: used for the GIS server to receive the registration address information data input by the user in the client device, and preprocess the registration address information data to obtain the preprocessed registration address information data; Word segmentation and annotation module: used to perform preliminary segmentation on the preprocessed registered address information data, obtain the registered address information data after preliminary segmentation, and perform word segmentation and annotation on the registered address information data after preliminary segmentation to obtain the word segmentation and annotation results; An address segmentation module is used to perform reduplication expansion on the reported address information data after preliminary segmentation based on the segmentation and marking results to obtain the reported address information data after reduplication expansion, and perform address segmentation processing on the reported address information data after reduplication expansion based on a Bayesian probability model to obtain address segmentation information; Address type identification module: used to identify the address type based on the address segmentation information combined with address element index analysis to obtain target address type information; Address matching module: used to perform address matching based on the address segmentation information and the target address type information to obtain a target matching address; Building mask detection module: used to call GIS map to extract the building real scene image corresponding to the target matching address, and perform building mask detection on the building real scene image to obtain building mask information; Building view generation module: used for performing contour extraction based on the building mask information to obtain building contour information, generating a target building view based on the building contour information combined with the building label, and marking the target matching address in the corresponding position of the target building view; The preliminarily segmenting the preprocessed registration address information data to obtain the preliminarily segmented registration address information data, and performing word segmentation and marking on the preliminarily segmented registration address information data to obtain the word segmentation and marking results, includes: preliminarily segmenting the preprocessed registration address information data based on a preset segmentation unit to obtain the preliminarily segmented registration address information data; obtaining a preset reduplication marking relationship table and a segmentation training corpus, performing model statistical learning based on the preset reduplication marking relationship table and the segmentation training corpus to obtain a target reduplication marking model; performing word segmentation and marking on the preliminarily segmented registration address information data based on the target reduplication marking model to obtain the word segmentation and marking results; The method of performing reduplication expansion on the registration address information data after preliminary segmentation based on the word segmentation and annotation results to obtain the registration address information data after reduplication expansion, and performing address segmentation processing on the registration address information data after reduplication expansion based on a Bayesian probability model to obtain address segmentation information, includes: performing reduplication expansion on the characters with word segmentation annotations in the registration address information data after preliminary segmentation to obtain the registration address information data after reduplication expansion; generating a bag-of-words frequency vector based on a preset stop word list, and constructing a Bayesian probability model based on the bag-of-words frequency vector; performing a topic-term distribution analysis based on the Bayesian probability model to obtain a feature word library; performing mapping identifier traversal on the segmentation training corpus based on the word segmentation character string to obtain a target mapping word segmentation identifier, and constructing a word segmentation array based on the target mapping word segmentation identifier; performing address segmentation processing on the registration address information data after reduplication expansion based on the word segmentation array and the feature word library to obtain address segmentation information.

8. A GIS-based address data intelligent analysis and processing system, characterized in that: The system includes a client device and a geographic information system (GIS) server, and the client device is communicatively connected to the GIS server; the system is configured to execute the GIS-based intelligent analysis and processing method for address data as described in any one of claims 1 to 6.

Images