Map data completion method and device, equipment, storage medium and program product

Abstract

The invention provides a map data complementing method and device, equipment, a storage medium and a program product. The method comprises the steps that first map data are segmented into a plurality of first data fragments, the data missing type to which each first data fragment belongs is determined, each first data fragment comprises a missing data fragment, and each missing data fragment comprises at least one continuous missing data; for each first data slot, calling a data completion model corresponding to the data missing type of the first data slot, performing data completion on the first data slot to obtain a second data slot, the second data slot comprising first completion data of each missing data; splicing the plurality of second data fragments to obtain second map data; and calling the data correction model, and correcting each piece of first complementation data in the second map data into second complementation data to obtain third map data. According to the invention, the complementation precision of the map data can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence, and in particular relates to a map data completion method and device, equipment, a storage medium and a program product. BACKGROUND

[0002] As the core foundation of the digital operation and management of expressways, high-precision maps are necessary tools for realizing the digital transformation of expressways and providing high-quality travel services. In actual application, map data often has missing or abnormal data due to problems such as sensor obstruction, signal interference, and high data collection costs. In related technologies, data completion methods are mostly implemented by interpolation algorithms based on statistical inference or rules, and the completion accuracy of map data is poor. SUMMARY

[0003] The embodiments of the present application provide a map data completion method, device, electronic equipment, computer readable storage medium and computer program product, which can improve the completion accuracy of map data.

[0004] The technical solutions of the embodiments of the present application are implemented as follows: The embodiments of the present application provide a map data completion method, which comprises the following steps: For first map data comprising a plurality of missing data segments, the first map data is divided into a plurality of first data segments, and a data missing type to which each first data segment belongs is determined, each first data segment comprises one missing data segment, and each missing data segment comprises at least one continuous missing data. For each first data segment, a data completion model corresponding to the data missing type of the first data segment is called to complete the data of the first data segment, and a second data segment is obtained, the second data segment comprises first completion data of each missing data. The plurality of second data segments are spliced to obtain second map data. A data correction model is called to correct each first completion data in the second map data to second completion data, and third map data is obtained.

[0005] The embodiments of the present application also provide a map data completion device, which comprises: The division module is configured to divide, for first map data comprising a plurality of missing data segments, the first map data into a plurality of first data segments, and determine a data missing type to which each first data segment belongs, each first data segment comprises one missing data segment, and each missing data segment comprises at least one continuous missing data. a data completion module, configured to, for each first data segment, call a data completion model corresponding to the data missing type of the first data segment, and perform data completion on the first data segment to obtain a second data segment, the second data segment including first completion data of each missing data; a splicing module, configured to splice a plurality of second data segments to obtain second map data; a data correction module, configured to call a data correction model, and correct each first completion data in the second map data into second completion data to obtain third map data.

[0006] The embodiment of the present application also provides an electronic device, comprising: a memory, configured to store computer executable instructions; a processor, configured to execute the computer executable instructions stored in the memory, and implement the map data completion method provided by the embodiment of the present application.

[0007] The embodiment of the present application also provides a computer readable storage medium, which stores computer executable instructions or computer programs, and the computer executable instructions or computer programs are executed by a processor to implement the map data completion method provided by the embodiment of the present application.

[0008] The embodiment of the present application also provides a computer program product, which comprises computer executable instructions or computer programs, and the computer executable instructions or computer programs are executed by a processor to implement the map data completion method provided by the embodiment of the present application.

[0009] The embodiment of the present application has the following beneficial effects: According to the above embodiment of the present application, the first map data is first divided into a plurality of first data segments (each first data segment includes a missing data segment), and the data missing type to which each first data segment belongs is determined, then a data completion model corresponding to the data missing type is called to perform data completion on the first data segment to obtain a second data segment, and a plurality of second data segments are spliced to obtain second map data, so that each first completion data in the second map data is corrected into second completion data by calling a data correction model to obtain third map data. In this way, the corresponding data completion model can be called for different data missing types for completion processing, which is suitable for diversified data missing types and has pertinence to a certain data missing type, and the completion accuracy of the map data is initially improved; the splicing operation restores the data continuity, and the data completion result of the data completion model is optimized in combination with the data correction model, so that the completion accuracy of the map data is further improved, and the reliability of the completed map data is improved. BRIEF DESCRIPTION OF DRAWINGS

[0010] Figure 1 is an architecture schematic diagram of a map data completion system provided by an embodiment of the present application; Figure 2 is a structure schematic diagram of an electronic device provided by an embodiment of the present application; Figure 3 is a first flow schematic diagram of a map data completion method provided by an embodiment of the present application; Figure 4 is a second flow schematic diagram of a map data completion method provided by an embodiment of the present application; Figure 5 is a third flow schematic diagram of a map data completion method provided by an embodiment of the present application; Figure 6 is a segmentation schematic diagram of first map data provided by an embodiment of the present application; Figure 7 is a schematic diagram of an initial sample data set provided by an embodiment of the present application; Figure 8 is a flow schematic diagram of selecting observation features based on a genetic algorithm provided by an embodiment of the present application; Figure 9 is a schematic diagram of extracting a completion data segment from second map data provided by an embodiment of the present application.

[0011] It should be noted that the above-mentioned "first", "second" are only used to distinguish different schemes, and do not represent the degree of superiority or priority in the implementation process. DETAILED DESCRIPTION

[0012] In order to make the purpose, technical scheme and advantages of the present application clearer, the present application will be described in further detail below with reference to the accompanying drawings, and the described embodiments should not be regarded as limiting the present application, and all other embodiments obtained by a person of ordinary skill in the art without making creative labor are within the scope of protection of the present application.

[0013] In the following description, "some embodiments" are described, which describe a subset of all possible embodiments, but it can be understood that "some embodiments" can be the same subset or different subset of all possible embodiments, and can be combined with each other without conflict.

[0014] In the following description, the terms "first\second\third" are only used to distinguish similar objects, and do not represent a specific order of the objects, and it can be understood that "first\second\third" can be interchanged in a specific order or sequence as allowed, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0015] In the embodiments of this application, the term "module" or "unit" refers to a computer program or a part of a computer program with a predetermined function, and works together with other related parts to achieve a predetermined target, and can be implemented entirely or partially by using software, hardware (such as a processing circuit or a memory) or a combination thereof. Similarly, one processor (or multiple processors or memories) can be used to implement one or more modules or units. In addition, each module or unit can be a part of an integral module or unit that contains the functions of the module or unit.

[0016] Unless otherwise defined, all technical and scientific terms used in the embodiments of the present application have the same meanings as commonly understood by one of ordinary skill in the art. The terms used in the embodiments of the present application are only for the purpose of describing the embodiments of the present application and are not intended to limit the present application.

[0017] The relevant data collection process in the embodiments of the present application should strictly comply with the requirements of relevant laws and regulations, obtain the informed consent or separate consent of the personal information subject, and within the scope of authorization of laws and regulations and the personal information subject, carry out subsequent data use and processing behavior.

[0018] Before the embodiments of the present application are further described in detail, the terms and phrases involved in the embodiments of the present application are explained, which are applicable to the following explanations.

[0019] 1) In response to, for indicating the condition or state on which the operation is dependent, when the dependent condition or state is met, one or more operations performed can be real-time or have a set delay; in the absence of special instructions, there is no restriction on the execution order of multiple operations performed.

[0020] 2) Map data refers to a structured information set directly associated with a geographic spatial location, used to describe the attributes, forms, and spatial relationships of geographic elements on the surface of the Earth (or other celestial bodies). It is the core foundation of geographic information systems (GIS), navigation applications, spatial analysis, and other fields. It not only contains spatial location information of "where", but also covers attribute descriptions of "what", and can directly or indirectly reflect various characteristics of the geographic environment. From the perspective of composition dimensions, map data can generally be divided into the following key types and elements: (1) Data types: mainly divided into vector data and raster data. Vector data stores geometric objects such as "points", "lines", and "surfaces" (e.g. "points" representing landmark buildings, "lines" representing roads, and "surfaces" describing lake ranges), with high precision and ease of editing; raster data is presented in the form of a pixel matrix (e.g. satellite remote sensing images, aerial photographs), which is good at restoring continuous geographic scenes (e.g. terrain undulations, vegetation coverage). (2) Core features: first, it has a spatial reference system, which needs to rely on a unified coordinate system to achieve accurate positioning of geographic elements and avoid "position deviation"; second, it is associated with attribute information, for example, a vector data of a road will contain descriptive information such as road name, number of lanes, speed limit, and road surface material in addition to spatial coordinates. (3) Main sources: including professional surveying and mapping (traditional geodetic surveying, unmanned aerial vehicle surveying), satellite remote sensing (such as Gaofen satellites, Landsat satellites), user collaboration (such as OpenStreetMap), and official or enterprise-built geographic databases. (4) Typical applications: widely serving navigation and positioning (such as path planning in map APPs), urban planning (such as land layout and traffic flow simulation), emergency rescue (such as disaster area terrain analysis and material delivery route optimization), and logistics distribution (such as order distribution based on regional geography), etc. It is a key support for "spatial perception" and "decision optimization".

[0021] For example, the map data here can refer to high-precision map data, which is the core basic data supporting the digital operation management of expressways, realizing the digital transformation of expressways, and providing high-quality travel services. It can include road segment coordinates, lane line information, traffic sign locations, and monitoring point states, and needs to have spatial continuity and temporal dynamics.

[0022] The missing data is a case where spatial position information, attribute description information, or spatial relationship information of part of geographic elements in the map data is missing, blank, or invalid (continuous or single), which is not simply data "not existing", but also includes cases where data exists but cannot be used (such as format error, logical contradiction), and directly affects the positioning accuracy, analysis effectiveness, and application reliability of the map, and is a problem that needs to be solved in geographic information processing. In the whole process of map data from collection to application, data missing may occur in multiple links, such as data collection link caused by sensor obstruction, signal interference, surveying and mapping equipment failure, data collection cost, etc., such as data processing link caused by different coordinate system conversion, data format batch conversion, etc. Missing data can be of the front missing type, middle missing type, and rear missing type. Missing data is the target object that is completed by a data completion model (such as an Elman model) and optimized by a data correction model (such as a Mamba model). Non-missing data is valid data in the map data other than missing data.

[0023] 3) Map data completion refers to supplementing, repairing, or inferring missing data to restore the spatial integrity, attribute integrity, and logical consistency of the map data, and ultimately improve the data usability and analysis reliability of the map data, through scientific technical methods and geographic rules, for the problems of spatial position missing, attribute information missing, or topological relationship missing in the map data. The core goal is not simply to "fill in the blanks", but to ensure that the completed data conforms to the geographic reality rules (such as road direction, regional boundary logic), and can adapt to subsequent visualization, spatial analysis, and decision-making application requirements. The core goal of map data completion is: first, to restore data integrity and make the originally "invalid" or "biased" data regain its value; second, to ensure data accuracy and avoid introducing new errors in the completion process (such as assigning incorrect road coordinates, unreasonable attribute values), and ensure that the data is consistent with the real geographic scene.

[0024] Embodiments of the present application provide a map data completion method, device, electronic equipment, computer readable storage medium, and computer program product, which can improve the completion accuracy of map data. Next, based on the above description of the terms and terms involved in the embodiments of the present application, the embodiments of the present application will be described in detail.

[0025] The following describes a map data completion system provided by the embodiments of the present application. Referring to Figure 1 , Figure 1FIG. 1 is a schematic diagram of an architecture of a map data completion system provided by an embodiment of the present application. To implement a supporting example application, the map data completion system 100 includes a server 200, a network 300, and a terminal 400. The terminal 400 connects to the server 200 through the network 300, and the network 300 can be a wide area network or a local area network, or a combination of the two, and data transmission is implemented using wireless or wired links.

[0026] Here, the terminal 400 sends a data completion request for first map data (including a plurality of missing data segments) to the server 200 in response to a data completion instruction for the first map data; the server 200 receives the data completion request sent by the terminal 400; in response to the data completion request, the first map data is divided into a plurality of first data segments, and the data missing type to which each first data segment belongs is determined, each first data segment includes a missing data segment, and each missing data segment includes at least one missing data in sequence; for each first data segment, a data completion model corresponding to the data missing type of the first data segment is called to complete the data of the first data segment, to obtain a second data segment, the second data segment includes first completion data of each missing data; the plurality of second data segments are spliced to obtain second map data; a data correction model is called to correct each first completion data in the second map data to second completion data, to obtain third map data; the third map data is returned to the terminal 400; the terminal 400 receives the third map data sent by the server 200, and the third map data is the map data obtained after the data of the first map data is completed.

[0027] The map data completion method provided in the embodiments of the present application is implemented by an electronic device, for example, can be implemented by a terminal alone, can be implemented by a server alone, or can be implemented by a terminal and a server in cooperation. The electronic device implementing the map data completion method provided in the embodiments of the present application can be various types of terminals or servers. The server (for example, the server 200) can be a stand-alone physical server, or can be a server cluster or a distributed system composed of multiple physical servers, or can be a cloud server providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and basic cloud computing services such as big data and artificial intelligence platforms. The terminal (for example, the terminal 400) can be a notebook computer, a tablet computer, a desktop computer, a smart phone, a smart voice interaction device (for example, a smart speaker), a smart home appliance (for example, a smart television), a smart watch, a vehicle-mounted terminal, a wearable device, a virtual reality (VR) device, a flying device, and the like, but is not limited thereto. The terminal and the server can be connected directly or indirectly through wired or wireless communication, and the embodiments of the present application do not limit this.

[0028] In some embodiments, the terminal or the server can implement the map data completion method provided in the embodiments of the present application by running various computer executable instructions or computer programs. For example, the computer executable instructions can be microprogram level commands, machine instructions or software instructions. The computer program can be a native program or a software module in an operating system; can be a native application program (APP), that is, a program that needs to be installed in an operating system to run; or can be a small program that can be embedded into any APP, that is, a program that only needs to be downloaded into a browser environment to run. In summary, the above computer executable instructions can be any form of instructions, and the above computer programs can be any form of application programs, modules or plug-ins.

[0029] The electronic device implementing the map data completion method provided in the embodiments of the present application is described below. Referring to Figure 2 , Figure 2 is a structural schematic diagram of the electronic device provided in the embodiments of the present application. The electronic device 500 provided in the embodiments of the present application can be a terminal or a server. As shown in Figure 2As shown, the electronic device 500 includes at least one processor 510, memory 550, at least one network interface 520, and a user interface 530. The various components of the electronic device 500 are coupled together by a bus system 540, which is used for the communication of information among the components and between the components and a storage system (not shown) and a peripheral device system (not shown). It will be appreciated by those skilled in the art that the bus 540 is illustrative only and other types of communications links can be used. Figure 2 The various buses can be combined into a single bus, or they can be implemented as separate buses.

[0030] The processor 510 can be an integrated circuit chip with processing capability that can be a general purpose processor, a Digital Signal Processor (DSP), or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof. The processor 510 can be a microprocessor, or any conventional processor, etc.

[0031] The user interface 530 includes one or more output devices 531 that enable presentation of media content, including one or more speakers and / or one or more visual display screens. The user interface 530 also includes one or more input devices 532 that facilitate user input, such as a keyboard, mouse, microphone, touch screen display, camera, other input buttons and controls.

[0032] The memory 550 can be removable, non-removable, or a combination thereof. The memory 550 can include one or more memory devices physically located in proximity to the processor 510. The memory 550 includes volatile memory or non-volatile memory, and can include both volatile and non-volatile memory. Non-volatile memory can be read only memory (ROM), and volatile memory can be random access memory (RAM). The memory 550 described in the embodiments of the present application is intended to encompass any suitable type of memory.

[0033] In some embodiments, the memory 550 is capable of storing data to support various operations, examples of which include programs, modules, and data structures or a subset or superset thereof, examples of which are illustrated below. Among them, the operating system 551 includes a system program for processing various basic system services and performing hardware-related tasks, such as a framework layer, a core library layer, a driver layer, etc., for implementing various basic services and processing hardware-based tasks; the network communication module 552 is used to reach other electronic devices via one or more (wired or wireless) network interfaces 520, example network interfaces 520 including Bluetooth, Wireless Fidelity (Wi-Fi), and Universal Serial Bus (USB), etc.; the presentation module 553 is used to enable the presentation of information via one or more output devices 531 (e.g., a display screen, a speaker, etc.) associated with the user interface 530 (e.g., a user interface for operating peripheral devices and displaying content and information); the input processing module 554 is used to detect and translate one or more user inputs or interactions from one or more input devices 532.

[0034] In some embodiments, the map data completion device provided by the embodiments of the present application can be implemented in a software manner, Figure 2 The map data completion device 555 stored in the memory 550 is shown, which can be software in the form of programs and plug-ins, including the following software modules: segmentation module 5551, data completion module 5552, splicing module 5553, and data correction module 5554. These modules are logical, so they can be combined or further split according to the functions implemented, and the functions of each module will be described below.

[0035] The map data completion method provided by the embodiments of the present application is described below. The map data completion method provided by the embodiments of the present application is implemented by an electronic device, for example, can be implemented by a server or a terminal alone, or by a server and a terminal in cooperation. Therefore, the execution subject of each step will not be repeated below. Referring to Figure 3 , Figure 3 is the first flowchart of the map data completion method provided by the embodiments of the present application. The map data completion method provided by the embodiments of the present application includes: Step 101: For the first map data including a plurality of missing data segments, the first map data is segmented into a plurality of first data segments, and the data missing type to which each first data segment belongs is determined.

[0036] Among them, each first data segment includes a missing data segment, and each missing data segment includes at least one continuous missing data.

[0037] For step 101, the first map data is the map data to be completed, and the first map data includes a plurality of missing data segments, each missing data segment including at least one missing data, which is one missing data or a plurality of continuous missing data. For example, referring to Figure 6 , the map data includes three missing data segments, i.e., data segment 1 including one missing data (i.e., missing data segment 1), data segment 2 including three missing data (i.e., missing data segment 2), and data segment 3 including two missing data (i.e., missing data segment 3).

[0038] The first map data is divided into a plurality of first data segments, each first data segment including one missing data segment. Then, the data missing type to which each first data segment belongs is determined. The data missing type to which the first data segment belongs is the data missing type to which the missing data segment belongs. The data missing type includes a front-segment missing type, a middle-segment missing type, and a rear-segment missing type. If there is no adjacent non-missing data segment before the missing data segment, it is determined that the data missing type to which the missing data segment belongs is the front-segment missing type, and thus the data missing type to which the first data segment belongs is also the front-segment missing type. If there is an adjacent non-missing data segment before the missing data segment and there is an adjacent non-missing data segment after the missing data segment, it is determined that the data missing type to which the missing data segment belongs is the middle-segment missing type, and thus the data missing type to which the first data segment belongs is also the middle-segment missing type. If there is no adjacent non-missing data segment after the missing data segment, it is determined that the data missing type to which the missing data segment belongs is the rear-segment missing type, and thus the data missing type to which the first data segment belongs is also the rear-segment missing type. For example, referring to Figure 6 , the data missing type to which data segment 1 belongs is the front-segment missing type, the data missing type to which data segment 2 belongs is the middle-segment missing type, and the data missing type to which data segment 3 belongs is the rear-segment missing type.

[0039] In actual applications, in addition to including the missing data segment, the first data segment also includes a non-missing data segment (including one non-missing data or a plurality of continuous non-missing data) associated with the missing data segment. Specifically, if the first data segment belongs to the front-segment missing type, the first data segment includes the missing data segment and a non-missing data segment located after and adjacent to the missing data segment, such as data segment 1 in Figure 6 . If the first data segment belongs to the middle-segment missing type, the first data segment includes the missing data segment, a non-missing data segment located before and adjacent to the missing data segment, and a non-missing data segment located after and adjacent to the missing data segment, such as data segment 2 in Figure 6the second data segment in the first map data. If the first data segment belongs to the post-segment missing type, the first data segment includes a missing data segment and a non-missing data segment located before and adjacent to the missing data segment, as shown in FIG. 2B. Figure 6 the third data segment in the first map data.

[0040] In some embodiments, referring to FIG. 10, Figure 4 the step 101 of "segmenting the first map data into a plurality of first data segments" can be implemented by performing the following steps: a step 1011 of obtaining a first quantity of missing data in each missing data segment; a step 1012 of, for each missing data segment, determining a first non-missing data segment located before and adjacent to the missing data segment in the first map data, and obtaining a second quantity of non-missing data in the first non-missing data, and determining a second non-missing data segment located after and adjacent to the missing data segment in the first map data, and obtaining a third quantity of non-missing data in the second non-missing data segment; a step 1013 of determining position information of each missing data segment; and a step 1014 of, for each missing data segment, segmenting the first data segment in which the missing data segment is located from the first map data based on the first quantity, the second quantity, the third quantity and the position information.

[0041] For the step 1011, for each missing data segment, a quantity of missing data included in the missing data segment is obtained, denoted as a first quantity . In determining the first quantity, the missing data segment can be directly located in the first map data, and then the first quantity of missing data included in the missing data segment is counted. Alternatively, a missing data can be located in the first map data, and then a quantity of consecutive missing data adjacent to the missing data is counted, and the first quantity is obtained by adding the quantity of the missing data.

[0042] For the step 1012, the following processing is performed for each missing data segment: first, a first non-missing data segment located before and adjacent to the missing data segment in the first map data is determined, as shown in FIG. 2A Figure 6 , the first non-missing data segment located before and adjacent to the missing data segment 2 in the map data: (3, 1, 0), (3, 3, 0), (4, 1, 0), (5, 2, 0), (6, 6, 0), (7, 2, 0), (8, 7, 0); then a second quantity of non-missing data in the first non-missing data is obtained, denoted as a second quantity ; and then a second non-missing data segment located after and adjacent to the missing data segment in the first map data is determined, as shown in FIG. 2B Figure 6As shown, the second non-missing data segment after and adjacent to the missing data segment 2 in the map data: (5, 2, 0), (6, 1, 0), (5, 3, 0), (7, 2, 0), (8, 1, 0); then a third number of non-missing data in the second non-missing data is obtained, denoted as the third number .

[0043] For step 1013, the position information of each missing data segment is determined . Specifically, if there is no adjacent non-missing data segment before the missing data segment, the position information of the missing data segment is determined to be front missing (indicating that there is no adjacent non-missing data segment before the missing data segment); if there is an adjacent non-missing data segment before the missing data segment and there is an adjacent non-missing data segment after the missing data segment, the position information of the missing data segment is determined to be middle missing (indicating that there is an adjacent non-missing data segment before the missing data segment and there is an adjacent non-missing data segment after the missing data segment); if there is no adjacent non-missing data segment after the missing data segment, the position information of the missing data segment is determined to be rear missing (indicating that there is no adjacent non-missing data segment after the missing data segment). In actual application, the position information of the missing data segment can be represented by a corresponding symbol, such as a numerical symbol, i.e.: front missing = 1; middle missing = 2; rear missing = 3.

[0044] For step 1014, the following processing is performed for each missing data segment: the first number, the second number, the third number and the position information can all be denoted as the missing characteristics of the missing data segment, so based on the first number, the second number, the third number and the position information, the first data segment in which the missing data segment is located is extracted from the first map data to obtain the first data segment. Specifically, around each missing data segment, the principle of "completely containing the missing data segment + covering the front and rear key non-missing data" is used for segmentation, i.e.: according to the first number to ensure that the segmentation range completely includes all continuous missing data (i.e. missing data segments); according to the second number , the third number respectively to ensure that the segmentation range (i.e. the obtained first data segment) includes all non-missing data adjacent to the missing data segment before and after (to ensure that there is enough reference feature for subsequent completion processing); in combination with the position information to verify whether the segmentation range matches the corresponding data missing type, such as front missing, which requires ensuring that there is no redundant non-missing data before the missing data segment in the first data segment obtained after segmentation. Each first data segment obtained by final segmentation contains only one missing data segment, and carries , , , the complete features.

[0045] By applying the steps 1011-1014, 1) by quantifying the first number of missing data, the second / third number of adjacent non-missing data before and after, and combining the position information of the missing data segment, the first data segment is segmented, which can accurately capture the core features of each missing data segment (corresponding to the quantization parameters , , , ), providing a unified quantization standard for subsequent classification of data missing types. 2) Each first data segment after segmentation contains only one missing data segment, avoiding the mixing of different missing scene data, so that the corresponding data completion model can be called specifically later, reducing redundant information interference.

[0046] In some embodiments, "determining the data missing type to which each first data segment belongs" can be achieved by performing the following processing for each first data segment: if the position information indicates that there is no adjacent non-missing data segment before the missing data segment, it is determined that the data missing type to which the first data segment belongs is the front segment missing type; if the position information indicates that there is an adjacent non-missing data segment before the missing data segment, and there is an adjacent non-missing data segment after the missing data segment, it is determined that the data missing type to which the first data segment belongs is the middle segment missing type; if the position information indicates that there is no adjacent non-missing data segment after the missing data segment, it is determined that the data missing type to which the first data segment belongs is the rear segment missing type.

[0047] Here, each first data segment includes a missing data segment, so when determining the data missing type to which the first data segment belongs, the position information of the missing data segment needs to be determined. Specifically, if the position information indicates that there is no adjacent non-missing data segment (including one non-missing data or multiple consecutive non-missing data) before the missing data segment, it is determined that the data missing type to which the first data segment belongs is the front segment missing type, such as the data segment 1 in Figure 6 the data missing type to which the first data segment belongs is the front segment missing type. If the position information indicates that there is an adjacent non-missing data segment before the missing data segment, and there is an adjacent non-missing data segment after the missing data segment, it is determined that the data missing type to which the first data segment belongs is the middle segment missing type, such as the data segment 2 in Figure 6 the data missing type to which the first data segment belongs is the middle segment missing type. If the position information indicates that there is no adjacent non-missing data segment after the missing data segment, it is determined that the data missing type to which the first data segment belongs is the rear segment missing type, such as the data segment 3 in Figure 6 the data missing type to which the first data segment belongs is the rear segment missing type.

[0048] By applying the above embodiments, the three types of data missing, i.e., the front segment, the middle segment and the rear segment, can be accurately divided based on the presence or absence of non-missing data before and after the missing data segment, so as to avoid confusion of different data missing scenarios. This provides a basis for subsequent targeted calling of data completion models, so that the data completion models of different data missing types can be used to process the first data segment to be completed, such as the first data segment of the front segment missing type can be completed based on the subsequent non-missing data, the first data segment of the middle segment missing type can be completed based on the front and rear non-missing data, and the first data segment of the rear segment missing type can be completed based on the previous non-missing data. In this way, the data completion model focuses on the corresponding effective reference features, reduces the interference of redundant information, and improves the data completion accuracy.

[0049] Step 102: For each first data segment, a data completion model corresponding to the data missing type of the first data segment is called to complete the data of the first data segment, and a second data segment is obtained.

[0050] The second data segment includes the first completion data of each missing data.

[0051] For step 102, the following processing is performed for each first data segment: a data completion model corresponding to the data missing type to which the first data segment belongs is called to complete the data of the first data segment, and a second data segment is obtained. The data completion model is used to predict the first completion data of each missing data based on the first data segment including the missing data. The second data segment is a data segment obtained by completing each missing data in the first data segment. The second data segment includes the first completion data of each missing data, i.e., the second data segment is obtained by replacing each missing data in the first data segment with the corresponding first completion data. The missing data and the first completion data are one-to-one corresponding.

[0052] In some embodiments, different data missing types correspond to different data completion models. In actual applications, the data completion model can be a model constructed based on an Elman neural network. The Elman neural network is a typical model of shallow deep learning, and its advantages of simple structure and few parameters can reduce the dependence on the overall amount of data. To ensure that different data missing types correspond to different data completion models, different sample data of different data missing types can be used to train the corresponding data completion models, and the model structures of different data completion models can be the same, such as both being constructed based on an Elman neural network, but the model parameters of different data completion models are different; of course, the model structures of different data completion models can also be different, which is not limited here. In this way, the data completion model can accurately adapt to the differentiated data missing scenarios of front missing, middle missing, and rear missing - front missing only has non-missing data in the sequence, middle missing has non-missing data before and after, and rear missing only has non-missing data in the front sequence, so that the data completion model focuses on the corresponding effective reference features, avoids irrelevant information interference, reduces overfitting, and improves the preliminary completion accuracy of map data.

[0053] In some embodiments, the data completion model corresponding to the data missing type is obtained by performing the following processing: obtaining an initial data completion model, and obtaining a sample data set, the sample data set including a plurality of sample data belonging to the data missing type, each sample data including observation data and target data, the observation data representing non-missing data, and the target data representing a label of missing data; for each sample data, calling the initial data completion model to complete the data of the sample data with missing target data, to obtain a completed sample data corresponding to the sample data; based on the difference between each sample data and the corresponding completed sample data, updating the model parameters of the initial data completion model to obtain the data completion model.

[0054] Here, first, an initial data completion model constructed in advance is obtained, which can be constructed based on an Elman neural network and includes initial model parameters. The data completion model is obtained by training the initial data completion model. The data completion model corresponding to different data missing types can be trained based on sample data of different data missing types, so for each data missing type, a sample data set corresponding to the data missing type is obtained, and the sample data set includes a plurality of sample data belonging to the data missing type. In actual applications, the data sample set is generated based on historical map data (excluding missing data), and specifically, missing data can be randomly marked (i.e., non-missing data is randomly marked as missing data, but the non-missing data is not actually deleted) in the historical map data to form historical map data including a plurality of missing data segments, and then the first number of missing data in each missing data segment is obtained For each missing data segment, identify the first non-missing data segment in the first map data that is preceding and adjacent to the missing data segment, and obtain the second number of non-missing data segments in the first non-missing data segment. And, determine the second non-missing data segment in the first map data that is located after and adjacent to the missing data segment, and obtain the third number of non-missing data in the second non-missing data segment. Determine the location information of each missing data segment. According to the first quantity Second quantity Third quantity and location information According to the principle of merging like items (i.e., ensuring that each sample data belongs to the same missing data type), , , , It is a match, for example: each sample data of the missing segment type is: >0、 =0、 >0、 =1; Each sample data of the middle segment missing type is: >0、 >0、 >0、 =2; Each sample data of the latter missing type is: >0、 >0、 =0、 =3), extract multiple sample data corresponding to each data missing type from the map data marked with missing data. , , , ,in, This represents a sample dataset representing a type of missing data. The first sample in the dataset representing this data missing type Sample data.

[0055] Wherein each sample data includes observation data and target data, the observation data represents non-missing data in the historical map data, and the target data represents non-missing data marked as missing data in the historical map data, which can be understood as a label of missing data (i.e. marked as missing data). Based on this, the initial data completion model is trained using the sample data set of this data missing type to obtain the data completion model of this data missing type. Specifically, the initial data completion model is called to complete data for each sample data of the missing target data, to obtain the corresponding completed sample data of each sample data; based on the difference between each sample data and the corresponding completed sample data, the value of the loss function is determined, so that based on the value of the loss function, the model parameters of the initial data completion model are updated to obtain the data completion model.

[0056] By applying the above embodiments, 1) the sample data set is divided according to the data missing type and contains "observation data (non-missing data) + target data (missing data label)", which is adapted to the quantization parameter 、 、 、 , which can make the initial data completion model focus on the "observation-target" mapping relationship of the same missing scenario, and avoid the learning deviation caused by the mixed data of different missing types. 2) Through the closed loop of "calling the model to complete → calculating the difference between the sample and the completed sample → updating the parameters", the learning of the data completion model on the association rule between the non-missing data and the missing data can be continuously optimized, the completion precision is improved, especially the characteristics of "anti-noise and low data dependence" of the Elman model, and the overfitting is reduced. 3) The "data completion model" finally obtained can accurately match the corresponding data missing type, lay a foundation for subsequent calling of the exclusive data completion model for the first data segment of different data missing types, guarantee the preliminary completion quality, support the subsequent data splicing and data correction process, and help to improve the completeness and reliability of the map data.

[0057] In some embodiments, referring to Figure 5The "obtaining a sample data set" can be achieved by performing the following steps: step 201, obtaining an initial sample data set, the initial sample data set being a matrix of M rows*N columns, the initial sample data set including M rows of initial sample data of a data missing type, each row of initial sample data including observation data and target data, one column of observation data forming an observation feature, one column of target data forming a target feature, the initial sample data set including N1 columns of observation features and N2 columns of target features, N=N1+N2; step 202, determining a correlation coefficient between each observation feature and each target feature; step 203, from the N1 columns of observation features, determining a first observation feature with a correlation coefficient less than a first threshold value and a second observation feature with a correlation coefficient greater than a second threshold value, the second threshold value being greater than the first threshold value; step 204, for a third observation feature other than the first observation feature and the second observation feature in the N1 columns of observation features, selecting a fourth observation feature from the third observation feature based on a genetic algorithm; and step 205, based on the fourth observation feature, the second observation feature, and the N2 columns of target features, constructing a sample data set.

[0058] For step 201, obtaining an initial sample data set, the data completion model corresponding to different data missing types can be trained based on sample data of different data missing types, so for each data missing type, an initial sample data set corresponding to the data missing type is obtained, and the initial sample data set includes a plurality of sample data belonging to the data missing type. In actual application, the initial sample data set is generated based on historical map data (not including missing data), specifically, missing data can be randomly marked (i.e. non-missing data is randomly marked as missing data, but the non-missing data is not actually deleted) in the historical map data to form historical map data including a plurality of missing data segments, then a first number of missing data in each missing data segment is obtained , a second number of non-missing data in a first non-missing data segment adjacent to the missing data segment before the missing data segment in the first map data is obtained , and a third number of non-missing data in a second non-missing data segment adjacent to the missing data segment after the missing data segment in the first map data is obtained ; the position information of each missing data segment is determined ; according to the first number , the second number , the third number , and the position information , , , is matched, such as: each initial sample data of the front segment missing type is: > 0, = 0, > 0, = 1; each initial sample data of the middle segment missing type is: > 0, > 0, > 0, = 2; each initial sample data of the rear segment missing type is: > 0, > 0, = 0, = 3), from the map data marked with missing data, a plurality of initial sample data corresponding to each data missing type is extracted , , , wherein, represents a set of initial sample data of a data missing type, represents the i-th initial sample data in the set of initial sample data of the data missing type.

[0059] wherein, as shown in formula (1), the set of initial sample data is a matrix of M rows and N columns, that is: M rows belong to initial sample data of a data missing type, and N columns are features (including observation features and target features). Each row of initial sample data includes observation data and target data, one column of observation data forms an observation feature, and one column of target data forms a target feature. The set of initial sample data includes N1 (depending on b+c) columns of observation features and N2 columns of target features (depending on a), N=N1+N2, N, N1 and N2 are all integers greater than 0. In the set of initial sample data, N1 columns of observation features are located on the left, and N2 columns of target features are located on the right. Figure 7

[0060] For step 202, the correlation coefficient between each observation feature and each target feature is determined, such as the Spearman correlation coefficient.

[0061] For step 203, a first observation feature with a correlation coefficient less than a first threshold value is determined from the N1 columns of observation features; and a second observation feature with a correlation coefficient greater than a second threshold value is determined from the N1 columns of observation features. Wherein the second threshold value is greater than the first threshold value, such as the first threshold value can be 0.3, and the second threshold value can be 0.8.

[0062] ​​For step 204, a fourth observation feature is selected from the third observation features based on a genetic algorithm. Thus, by pre-excluding the first observation feature and the second observation feature through the first threshold and the second threshold, the calculation amount of the genetic algorithm for selecting the fourth observation feature can be reduced, and a lightweight observation feature selection effect can be achieved.

[0063] Specifically, referring to Figure 8 , the selecting the fourth observation feature from the third observation features based on the genetic algorithm includes: (1) randomly selecting a plurality of observation features from the third observation features to generate an initial population of the i-th round .

[0064] (2) calculating a fitness function value of a population (an observation feature combination composed of a plurality of observation features) of the i-th (i [0, 100]) round .

[0065] The fitness function value can be calculated by the following formula (1) : Formula (1) Wherein, is the fitness function value; is the number of observation features in the population; is the real data marked as missing data in the historical map data; is the completed data obtained by completing data based on the observation feature combination of the current population through the data completion model.

[0066] (3) based on the fitness function value of the i-th round population , calculating the selection rate of the i-th round population .

[0067] The selection rate is calculated by the following formula (2): Formula (2) Wherein, is the selection rate.

[0068] (4) randomly selecting individuals from the initial population as the optimal observation feature subset of the current round according to the selection rate.

[0069] (5) randomly replacing the observation features in the current optimal observation feature subset according to the mutation rate to generate the population of the i+1-th round. Wherein, the mutation position is only the missing data of the non-main position.

[0070] (6) If the termination function requirement is met (i.e., i = 100), the optimal observation feature subset is directly output, and if not, steps (1)-(5) are repeated until the maximum number of iterations i is reached, and the optimal observation feature subset is output.

[0071] Thus, the final output optimal observation feature subset is: the fourth observation feature selected from the third observation feature.

[0072] For step 205, based on the fourth observation feature, the second observation feature and the N2 column target feature, a sample data set is constructed.

[0073] By applying the above steps 201-205, first, the correlation coefficient is used for screening to eliminate redundant first observation features less than a first threshold (such as 0.3) and retain high-value second observation features greater than a second threshold (such as 0.8), thereby reducing invalid feature interference and reducing subsequent calculation burden; second, the genetic algorithm is used to select a fourth observation feature from the third observation feature, thereby accurately mining features useful for target features (missing data) and avoiding overfitting. The finally constructed sample data set only contains high-quality observation features and target features, which meets the training needs of the data completion model and improves the model completion accuracy, thereby laying a high-quality data foundation for subsequent map data completion, splicing and data correction, and ensuring efficient and reliable completion process.

[0074] In some embodiments, each completion sample data includes third completion data corresponding to each target data; based on this, the following processing can be performed for each completion sample data to obtain sample data segments for training the initial data correction model: for the first third completion data in the completion sample data, the sixth number of non-missing data before the first third completion data is determined, and for the last third completion data in the completion sample data, the seventh number of non-missing data after the last third completion data is determined; for each third completion data, a sample data segment including the third completion data is extracted from the completion sample data, the sixth number of data before the third completion data in the sample data segment, and the seventh number of data after the third completion data in the sample data segment. Based on this, the initial data correction model can be trained by performing the following steps: obtaining the initial data correction model; for each sample data segment, calling the initial data correction model to correct the third completion data in the sample data segment to obtain fourth completion data; based on the difference between each fourth completion data and the corresponding target data, updating the model parameters of the initial data correction model to obtain the data correction model.

[0075] Here, after obtaining each sample data, each sample data includes third complete data corresponding to each target data. Each data correction processing of the data correction model is performed on one sample data, that is, the data correction model adopts a single-point missing data correction strategy. Therefore, the sixth number of non-missing data before the first third complete data in the sample data is determined, and the seventh number of non-missing data after the last third complete data in the sample data is determined. Based on this, the training sample of the initial data correction model is constructed based on each third complete data. Specifically, the following processing is performed on each third complete data: from the sample data, a sample data segment including the third complete data is extracted, and the sample data segment needs to meet the following conditions: the sixth number of data before the third complete data in the sample data segment, and the seventh number of data after the third complete data in the sample data segment.

[0076] Based on this, the initial data correction model can be trained by multiple sample data segments to obtain the data correction model. Specifically, the initial data correction model is called to correct the third complete data in each sample data segment to obtain fourth complete data, and then the value of the loss function is determined based on the difference between each fourth complete data and the corresponding target data, so that the model parameters of the initial data correction model are updated based on the value of the loss function to obtain the data correction model.

[0077] Applying the above embodiment, 1) the sixth data and the seventh number are used to extract the sample data segment, which can unify the dimensions of each sample data segment and avoid dimension confusion interference in training. 2) The sample data segment contains sufficient non-missing data before and after the third complete data, which can provide sufficient spatio-temporal context features for the initial data correction model (such as the Mamba model) to help it learn the association rules between the complete data and the real data. 3) Through the closed loop of "correcting to obtain fourth complete data-difference updating parameters", the model is continuously optimized to accurately adapt to the correction requirements, and the final data correction model can efficiently improve the accuracy of the complete data, lay a foundation for subsequent map data correction, and ensure the reliability of the map data.

[0078] Step 103: Splicing multiple second data segments to obtain second map data.

[0079] For step 103, after obtaining the multiple second data segments preliminarily completed by the data completion model, the multiple second data segments are spliced to restore data continuity to obtain the second map data.

[0080] In some embodiments, step 102, "stitching together multiple second data segments to obtain second map data", can be achieved by performing the following steps: stitching together multiple second data segments according to the arrangement order of the first data segments corresponding to each second data segment in the first map data to obtain stitched map data; deleting overlapping data between two adjacent second data segments from the stitched map data to obtain second map data.

[0081] Here, we first determine the order of the first data segments corresponding to each second data segment in the first map data. See also... Figure 6 Data segment 1 is the first element in the first data segment arrangement; data segment 2 is the second element; and data segment 3 is the third element. Therefore, multiple second data segments are stitched together according to the order of the first data segments corresponding to each second data segment in the first map data to obtain the stitched map data. However, because there is overlap between adjacent first data segments when acquiring them, for example… Figure 6 Data segments 1 and 2 in the image both contain overlapping data: (3,1,0), (3,3,0), (4,1,0), (5,2,0), (6,6,0), (7,2,0), and (8,7,0). Therefore, there is also overlap between the two second data segments corresponding to the two adjacent first data segments. Thus, for the stitched map data, the overlapping data between adjacent second data segments is deleted to obtain the second map data. It should be noted that deleting the overlapping data between adjacent second data segments means deleting the overlapping data only from one of the two adjacent second data segments, not deleting the overlapping data from both second data segments.

[0082] For example, suppose the second data segments include data segment 4, data segment 5, and data segment 6. Among them, data segment 4 and data segment 5 both include overlapping data: (3, 1, 0), (3, 3, 0), (4, 1, 0), (5, 2, 0), (6, 6, 0), (7, 2, 0), (8, 7, 0); data segment 5 and data segment 6 both include overlapping data: (5, 2, 0), (6, 1, 0), (5, 3, 0), (7, 2, 0), (8, 1, 0). Therefore, the data segment 4-data segment 6 are spliced to obtain the spliced map data, and then the overlapping data between the adjacent two second data segments is deleted from the spliced map data to obtain the second map data, that is, from the data segment 5, the overlapping data (3, 1, 0), (3, 3, 0), (4, 1, 0), (5, 2, 0), (6, 6, 0), (7, 2, 0), (8, 7, 0) of data segment 4 and data segment 5 is deleted; from data segment 6, the overlapping data (5, 2, 0), (6, 1, 0), (5, 3, 0), (7, 2, 0), (8, 1, 0) of data segment 5 and data segment 6 is deleted.

[0083] By applying the above embodiment, splicing the second data segments in the original arrangement order in the first map data can strictly preserve the spatial (such as the milepost number) or temporal continuity of the high-precision map data, avoid data logic disorder, and ensure that the data meets the actual use requirements of scenarios such as highway digital operation. Deleting the overlapping data between adjacent segments can eliminate redundant information, reduce the computational burden of subsequent data correction models, and avoid numerical conflict errors caused by overlapping data. The final second map data is continuous and has no redundancy, provides high-quality input for subsequent accurate correction of missing data, ensures the efficiency and data reliability of the entire completion process, and helps to improve the completeness of high-precision map data.

[0084] Step 104: calling a data correction model to correct each first completed data in the second map data into second completed data to obtain third map data.

[0085] For step 104, a data correction model is called to correct each first completed data in the second map data to correct each first completed data in the second map data into second completed data, thereby obtaining third map data. The third map data is the final map data obtained after data completion of the first map data.

[0086] Here, the data correction model is obtained by pre-training, and in actual application, the data correction model can be constructed by using a Mamba neural network. The Mamba neural network has strong high-order feature extraction and information learning capabilities. Therefore, in the embodiments of the present application, first, the strong anti-interference capability of the data completion model constructed based on an Elman neural network is used to complete the data completion task, and then the end-to-end adaptive feature extraction and high-order spatial feature learning capability of the data correction model constructed based on a Mamba neural network is used to complete the correction task of the missing data completion, thereby providing more accurate and rich feature expression for the reconstruction of the missing data and further improving the completion quality of the missing data.

[0087] It should be noted that each data correction processing of the data correction model is performed on one first completed data, that is, the data correction model adopts a single-point missing data correction strategy.

[0088] In some embodiments, before the data correction model is called to correct the first completed data, the following steps can also be performed: determining a fourth number of non-missing data before the first first completed data in the second map data, and determining a fifth number of non-missing data after the last first completed data in the second map data; for each first completed data, extracting a completed data segment including the first completed data from the second map data, wherein the completed data segment includes the fourth number of data before the first completed data, and the completed data segment includes the fifth number of data after the first completed data. Based on this, the third map data can be obtained by performing the following processing on each first completed data: calling the data correction model to correct the first completed data based on the completed data segment of the first completed data to obtain second completed data.

[0089] Here, the fourth number of non-missing data before the first first completed data in the second map data needs to be determined, and the fifth number of non-missing data after the last first completed data in the second map data needs to be determined. As an example, referring to Figure 9 , the first first completed data in the second map data is (6, 4, 0), and the last first completed data is (9, 1, 0). The number of continuous non-missing data before the first completed data (6, 4, 0) is 4, so the fourth number is 4; the number of continuous non-missing data after the first completed data (9, 1, 0) is 5, so the fifth number is 5.

[0090] Based on this, the input data of the input data correction model is constructed based on each first completed data. Specifically, the following processing is performed for each first completed data: from the second map data, a completed data segment including the first completed data is extracted, the completed data segment is a data segment including one first completed data in the second map data, and it is required to ensure that the completed data segment meets the following conditions: the first completed data is followed by a fourth number of data in the completed data segment, and the first completed data is followed by a fifth number of data in the completed data segment. Continuing to refer to Figure 9 , the completed data segment includes data segment 7, data segment 8, and data segment 9, wherein data segment 7 is a data segment including first completed data (6, 4, 0), which is followed by 4 data, and which is followed by 5 data; data segment 8 is a data segment including first completed data (7, 3, 0), which is followed by 4 data, and which is followed by 5 data; and data segment 9 is a data segment including first completed data (8, 1, 0), which is followed by 4 data, and which is followed by 5 data.

[0091] Based on this, the third map data can be obtained by performing the following processing for each first completed data: based on the completed data segment of the first completed data, the data correction model is called to correct the first completed data to obtain the second completed data.

[0092] By applying the above embodiments, 1) the fourth number and the fifth number of non-missing data before and after the first and last first completed data are determined first, and then the completed data segment containing the first completed data is extracted according to the number, which can unify the feature dimension of the data correction model and avoid affecting the data correction effect due to chaotic feature dimensions. 2) The completed data segment contains sufficient data before and after the first completed data, which can provide sufficient context features for the data correction model, help the data correction model accurately capture the spatio-temporal correlation rule, and improve the correction accuracy of the first completed data. 3) Adapt to the collaborative logic of "preliminary completion of the data completion model + data correction of the data correction model", provide high-quality input for the data correction model, further optimize the completion result, and ensure that the second map data is more accurate after correction, and improve the completeness and reliability of the map data.

[0093] According to the above embodiments of the present application, first map data is first divided into a plurality of first data segments (each first data segment includes a missing data segment), and a data missing type to which each first data segment belongs is determined, then a data completion model corresponding to the data missing type is called to complete data of the first data segment to obtain second data segments, the plurality of second data segments are spliced to obtain second map data, and each first completion data in the second map data is corrected to second completion data by calling a data correction model to obtain third map data. In this way, a corresponding data completion model can be called for different data missing types for completion processing, which is suitable for diversified data missing types and has pertinence to a certain data missing type, and the completion accuracy of map data is initially improved. The splicing operation restores data continuity, and the data completion result of the data completion model is optimized in combination with the data correction model, and the completion accuracy of map data is further improved, thereby improving the reliability of the completed map data.

[0094] The following continues to describe an exemplary structure of the implementation of the map data completion device 555 provided by the embodiments of the present application as a software module. In some embodiments, as shown in Figure 2 The software module stored in the map data completion device 555 of the memory 550 can include: a division module 5551 configured to divide first map data including a plurality of missing data segments into a plurality of first data segments, and determine a data missing type to which each first data segment belongs, each first data segment including a missing data segment, and each missing data segment including at least one missing data in series; a data completion module 5552 configured to, for each first data segment, call a data completion model corresponding to the data missing type of the first data segment to complete data of the first data segment to obtain second data segments, each second data segment including first completion data of each missing data; a splicing module 5553 configured to splice a plurality of second data segments to obtain second map data; and a data correction module 5554 configured to call a data correction model to correct each first completion data in the second map data to second completion data to obtain third map data.

[0095] In some embodiments, the cutting module 5551 is further configured to obtain a first number of missing data in each of the missing data segments; for each of the missing data segments, determine a first non-missing data segment adjacent to the missing data segment and located before the missing data segment in the first map data, and obtain a second number of non-missing data in the first non-missing data segment, and determine a second non-missing data segment adjacent to the missing data segment and located after the missing data segment in the first map data, and obtain a third number of non-missing data in the second non-missing data segment; determine position information of each of the missing data segments; and for each of the missing data segments, based on the first number, the second number, the third number and the position information, cut the first data segment in which the missing data segment is located from the first map data.

[0096] In some embodiments, the cutting module 5551 is further configured to, if the position information indicates that there is no adjacent non-missing data segment before the missing data segment, determine that the data missing type to which the first data segment belongs is a front-segment missing type; if the position information indicates that there is an adjacent non-missing data segment before the missing data segment and there is an adjacent non-missing data segment after the missing data segment, determine that the data missing type to which the first data segment belongs is a middle-segment missing type; and if the position information indicates that there is no adjacent non-missing data segment after the missing data segment, determine that the data missing type to which the first data segment belongs is a rear-segment missing type.

[0097] In some embodiments, the splicing module 5553 is further configured to splice a plurality of the second data segments in accordance with the arrangement order of each of the second data segments in the corresponding first data segment in the first map data, to obtain spliced map data; and delete overlapping data between two adjacent second data segments from the spliced map data, to obtain the second map data.

[0098] In some embodiments, the data correction module 5554 is further configured to, for a first first-completion data in the second map data, determine a fourth quantity of non-missing data before the first first-completion data, and for a last first-completion data in the second map data, determine a fifth quantity of non-missing data after the last first-completion data; for each of the first-completion data, extract a completion data segment including the first-completion data from the second map data, the completion data segment including the fourth quantity of data before the first-completion data and the fifth quantity of data after the first-completion data; and the data correction module 5554 is further configured to, for each of the first-completion data, perform the following processing respectively to obtain the third map data: based on the completion data segment of the first-completion data, invoke the data correction model to correct the first-completion data to obtain second-completion data.

[0099] In some embodiments, the data completion module 5552 is further configured to obtain the data completion model corresponding to the data missing type by performing the following processing: obtaining an initial data completion model, and obtaining a sample data set, the sample data set including a plurality of sample data belonging to the data missing type, each of the sample data including observation data and target data, the observation data representing non-missing data, and the target data representing a label of missing data; for each of the sample data, invoking the initial data completion model to perform data completion on the sample data missing the target data to obtain a completion sample data corresponding to the sample data; and based on a difference between each of the sample data and the corresponding completion sample data, updating model parameters of the initial data completion model to obtain the data completion model.

[0100] In some embodiments, the data completion module 5552 is further configured to obtain an initial sample data set, the initial sample data set being a matrix of M rows and N columns, the initial sample data set including M rows of initial sample data of the data missing type, each row of the initial sample data including observation data and target data, one column of the observation data constituting an observation feature, one column of the target data constituting a target feature, the initial sample data set including N1 columns of the observation features and N2 columns of the target features, N=N1+N2; determining a correlation coefficient between each of the observation features and each of the target features; determining, from the N1 columns of the observation features, first observation features with a correlation coefficient less than a first threshold value and second observation features with a correlation coefficient greater than a second threshold value, the second threshold value being greater than the first threshold value; selecting, based on a genetic algorithm, fourth observation features from the third observation features for third observation features other than the first observation features and the second observation features in the N1 columns of the observation features; and constructing the sample data set based on the fourth observation features, the second observation features, and the N2 columns of the target features.

[0101] In some embodiments, each of the completed sample data includes third completed data corresponding to each of the target data; and the data correction module 5554 is further configured to, for each of the completed sample data, perform the following processing: determining a sixth number of non-missing data before a first third completed data in the completed sample data and a seventh number of non-missing data after a last third completed data in the completed sample data; extracting, from the completed sample data, a sample data segment including the third completed data for each of the third completed data, the sample data segment including the sixth number of data before the third completed data and the seventh number of data after the third completed data; obtaining an initial data correction model; correcting the third completed data in the sample data segment to obtain fourth completed data by calling the initial data correction model for each of the sample data segments; and updating model parameters of the initial data correction model to obtain the data correction model based on a difference between each of the fourth completed data and corresponding target data.

[0102] It should be noted that the description of the device embodiments in the present application is similar to the description of the above-mentioned method embodiments, and has similar beneficial effects as the method embodiments, which will not be repeated here. For technical details not described in the map data completion device provided in the embodiments of the present application, the technical details can be understood based on the description of the above-mentioned method embodiments.

[0103] The embodiment of the present application further provides a computer program product, which comprises computer executable instructions or a computer program, and the computer executable instructions or the computer program are stored in a computer readable storage medium. The processor of the electronic device reads the computer executable instructions or the computer program from the computer readable storage medium, and the processor executes the computer executable instructions or the computer program, so that the electronic device executes the map data completion method provided by the embodiment of the present application.

[0104] The embodiment of the present application further provides a computer readable storage medium, which stores computer executable instructions or a computer program, and when the computer executable instructions or the computer program are executed by the processor, the processor will execute the map data completion method provided by the embodiment of the present application.

[0105] In some embodiments, the computer readable storage medium can be RAM, ROM, flash memory, magnetic surface memory, optical disc, or CD-ROM memory, etc.; or can be various devices comprising one or any combination of the above storage medium.

[0106] In some embodiments, the computer executable instructions can be in the form of programs, software, software modules, scripts or codes, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and can be deployed in any form, including being deployed as independent programs or being deployed as modules, components, subroutines or other units suitable for use in a computing environment.

[0107] As an example, the computer executable instructions can but not necessarily correspond to files in a file system, can be stored in a part of a file storing other programs or data, for example, stored in one or more scripts in a Hyper Text Markup Language (HTML) document, stored in a single file dedicated to the program in question, or stored in multiple cooperative files (for example, files storing one or more modules, subroutines or code parts).

[0108] As an example, the computer executable instructions can be deployed to be executed on one electronic device, or executed on multiple electronic devices located in one place, or executed on multiple electronic devices distributed in multiple places and interconnected through a communication network.

[0109] The above is only an embodiment of the present application, and is not used to limit the protection scope of the present application. Any modification, equivalent replacement and improvement made within the spirit and scope of the present application shall be included in the protection scope of the present application.

Claims

1. A method for completing map data, characterized in that, The method includes: For a first map data that includes multiple missing data segments, the first map data is divided into multiple first data segments, and the data missing type of each first data segment is determined. Each first data segment includes one missing data segment, and each missing data segment includes at least one consecutive missing data segment. For each of the first data segments, the data completion model corresponding to the data missing type of the first data segment is called to complete the data of the first data segment, and a second data segment is obtained. The second data segment includes the first completed data for each of the missing data. Multiple second data segments are stitched together to obtain second map data; The data correction model is invoked to correct each of the first completed data in the second map data to the second completed data, thus obtaining the third map data.

2. The method as described in claim 1, characterized in that, The step of dividing the first map data into multiple first data segments includes: Obtain the first number of missing data in each of the missing data segments; For each missing data segment, a first non-missing data segment in the first map data that is located before and adjacent to the missing data segment is determined, and a second number of non-missing data in the first non-missing data is obtained; and a second non-missing data segment in the first map data that is located after and adjacent to the missing data segment is determined, and a third number of non-missing data in the second non-missing data segment is obtained. Determine the location information of each of the missing data segments; For each missing data segment, based on the first quantity, the second quantity, the third quantity, and the location information, the first data segment containing the missing data segment is obtained from the first map data.

3. The method as described in claim 2, characterized in that, Determining the data missing type of each of the first data segments includes: For each of the first data segments, perform the following processing: If the location information indicates that there is no adjacent non-missing data segment before the missing data segment, the data missing type of the first data segment is determined to be the preceding segment missing type; If the location information indicates that there is an adjacent non-missing data segment before the missing data segment and an adjacent non-missing data segment after the missing data segment, the data missing type of the first data segment is determined to be the middle segment missing type; If the location information indicates that there is no adjacent non-missing data segment after the missing data segment, the data missing type of the first data segment is determined to be the subsequent segment missing type.

4. The method as described in claim 1, characterized in that, The step of stitching together multiple second data segments to obtain second map data includes: According to the arrangement order of the first data segments corresponding to each second data segment in the first map data, multiple second data segments are spliced ​​together to obtain spliced ​​map data; From the stitched map data, the overlapping data between two adjacent second data segments is deleted to obtain the second map data.

5. The method as described in claim 1, characterized in that, The method further includes: For the first first complete data in the second map data, determine the fourth number of non-missing data before the first first complete data, and for the last first complete data in the second map data, determine the fifth number of non-missing data after the last first complete data; For each of the first completed data, a completed data segment including the first completed data is extracted from the second map data. The completed data segment includes the fourth number of data before the first completed data and the completed data segment includes the fifth number of data after the first completed data. The data correction model is invoked to correct each of the first completed data points in the second map data to the second completed data points, thereby obtaining the third map data, including: For each of the first completed data points, the following processing is performed to obtain the third map data: Based on the completed data fragment of the first completed data, the data correction model is invoked to correct the first completed data to obtain the second completed data.

6. The method as described in claim 1, characterized in that, The method further includes: The data completion model corresponding to the missing data type is obtained by performing the following processing: Obtain an initial data completion model and a sample dataset, which includes multiple sample data belonging to the data missing type. Each sample data includes observation data and target data, where the observation data represents non-missing data and the target data represents the label of the missing data. For each of the sample data, the initial data completion model is invoked to complete the sample data that is missing the target data, thereby obtaining the completed sample data corresponding to the sample data; Based on the difference between each sample data and the corresponding completed sample data, the model parameters of the initial data completion model are updated to obtain the data completion model.

7. The method as described in claim 6, characterized in that, The acquisition of the sample dataset includes: Obtain an initial sample dataset, which is an M-row * N-column matrix. The initial sample dataset includes M rows of initial sample data belonging to the data missing type. Each row of the initial sample data includes observation data and target data. One column of the observation data constitutes an observation feature, and one column of the target data constitutes a target feature. The initial sample dataset includes N1 columns of the observation features and N2 columns of the target features, where N = N1 + N2. Determine the correlation coefficient between each observed feature and each target feature; From the observed features in column N1, determine a first observed feature whose correlation coefficient is less than a first threshold and a second observed feature whose correlation coefficient is greater than a second threshold, wherein the second threshold is greater than the first threshold; For the third observation feature in column N1, which is other than the first and second observation features, a fourth observation feature is selected from the third observation feature based on a genetic algorithm. The sample dataset is constructed based on the fourth observation feature, the second observation feature, and the target features in column N2.

8. The method as described in claim 6, characterized in that, Each of the completed sample data includes third completed data corresponding to each of the target data; the method further includes: For each of the completed sample data, perform the following processing: For the first third complete data in the completed sample data, determine the sixth number of non-missing data before the first third complete data, and for the last third complete data in the completed sample data, determine the seventh number of non-missing data after the last third complete data. For each of the third complete data, a sample data segment including the third complete data is extracted from the complete sample data. The sample data segment includes the sixth number of data before the third complete data and the seventh number of data after the third complete data. The method further includes: Obtain initial data to revise the model; For each of the sample data segments, the initial data correction model is invoked to correct the third complete data in the sample data segment, resulting in the fourth complete data. Based on the difference between each of the fourth complete data and the corresponding target data, the model parameters of the initial data correction model are updated to obtain the data correction model.

9. A map data completion device, characterized in that, The device includes: The segmentation module is used to segment the first map data, which includes multiple missing data segments, into multiple first data segments, and determine the data missing type of each first data segment. Each first data segment includes one missing data segment, and each missing data segment includes at least one consecutive missing data segment. The data completion module is used to call the data completion model corresponding to the data missing type of the first data segment for each first data segment, and to complete the data of the first data segment to obtain a second data segment, wherein the second data segment includes the first completed data for each missing data; The stitching module is used to stitch together multiple second data segments to obtain second map data; The data correction module is used to call the data correction model to correct each of the first complete data in the second map data to the second complete data, thereby obtaining the third map data.

10. An electronic device, characterized in that, The electronic device includes: Memory is used to store executable instructions for a computer; A processor, when executing computer-executable instructions stored in the memory, implements the map data completion method according to any one of claims 1 to 8.

11. A computer-readable storage medium storing computer-executable instructions or a computer program, characterized in that, When the computer-executable instructions or computer program are executed by a processor, the map data completion method according to any one of claims 1 to 8 is implemented.

12. A computer program product comprising computer-executable instructions or a computer program, characterized in that, When the computer-executable instructions or computer program are executed by a processor, the map data completion method according to any one of claims 1 to 8 is implemented.

Images