Map positioning method, system and device and storage medium

Abstract

The invention discloses a map positioning method, system and device and a storage medium, and the method comprises the steps: collecting a plurality of scene images, extracting SIFT feature points, pose information and three-dimensional point position information, carrying out the iteration of the information through a preset algorithm, obtaining an optimization value, replacing the SIFT feature points with Superpoint descriptors, and integrating the related information to generate an offline map. The method comprises the following steps: receiving a new scene image output by external equipment in real time, extracting new SIFT feature points and Superpoint descriptors, matching the features in an offline map, finding out a plurality of matched images, finally calculating the matched images based on a preset estimation algorithm to obtain positioning pose information, and adopting the offline map constructed in advance to obtain positioning pose information. The method effectively achieves accurate positioning, high frequency and low calculation amount, facilitates the use of low-configuration equipment, greatly reduces the use cost, and gives consideration to a good positioning effect.

Description

[0001] Technology Neighborhood

[0002] The present invention relates to the field of positioning technology, and in particular to a map positioning method, system, device and storage medium. Background Art

[0003] Current positioning technology is highly dependent on the device's built-in high-precision and fast-response IMU (inertial measurement unit) sensor and high-precision camera, which together form the core of the vision-assisted positioning system. The IMU sensor can provide key information such as the device's attitude, speed and position in real time, while the high-precision camera is responsible for capturing image data of the surrounding environment and providing the system with rich visual features.

[0004] However, this combination of technologies places extremely high demands on hardware. The R&D and production costs of high-precision IMU sensors and high-resolution cameras are relatively high, resulting in high costs for the entire device. At the same time, in order to process the massive amounts of data generated by these high-precision sensors, the device also needs to be equipped with powerful computing power, which further increases the cost of the device. In actual applications, due to the interference of various environmental factors and the limitations of hardware performance, the positioning effect is often unstable, which further reduces the user experience. Summary of the invention

[0005] The purpose of the present invention is to address the technical problems existing in the background technology and to propose a map positioning method, system, device and storage medium.

[0006] In order to achieve the above technical objectives, the technical solution adopted by the present invention is as follows:

[0007] A first implementation of the first aspect of the present invention provides a map positioning method, which includes:

[0008] S101, collecting multiple scene images, and extracting SIFT feature points, posture information and three-dimensional point position information from each of the scene images;

[0009] S102, based on a preset first algorithm, iterating the posture information and the three-dimensional point position information in each of the scene images to obtain iterative posture information and iterative three-dimensional point position information;

[0010] S103, replacing the SIFT feature points in each of the scene images with Superpoint descriptors, and integrating and storing a plurality of the Superpoint descriptors, a corresponding plurality of the iterative pose information, and a plurality of the iterative three-dimensional point position information to obtain an offline map;

[0011] S104, receiving multiple new scene images transmitted by a preset external device in real time, and extracting new SIFT feature points and new Superpoint descriptors in each of the new scene images;

[0012] S105, matching the new SIFT feature points and the new Superpoint descriptors in the offline map to obtain a plurality of matching images;

[0013] S106: Based on a preset estimation algorithm, calculate the multiple matching images to obtain positioning posture information.

[0014] Optionally, in a second implementation of the first aspect of the present invention, S101 includes:

[0015] S1011, collecting a plurality of scene images with an overlap rate of at least 50% between adjacent images;

[0016] S1012, extracting SIFT feature points from each of the scene images based on a preset SIFT algorithm;

[0017] S1013, matching a plurality of the SIFT feature points, and calculating the relative rotation data and translation data between adjacent scenes images to obtain the posture information and three-dimensional point position information corresponding to the scene images.

[0018] Optionally, in a third implementation of the first aspect of the present invention, S103 includes:

[0019] S1031, according to the corresponding vector relationship between the SIFT feature points and the Superpoint descriptor, replacing the SIFT feature points in each of the scene images with the Superpoint descriptor;

[0020] S1032, based on a preset greedy algorithm, calculating the effective range covered by each of the scene images, and screening out a plurality of screening scene images that cover the widest effective range;

[0021] S1033, integrating and storing the plurality of the filtered scene images, the plurality of the Superpoint descriptors, the corresponding plurality of the iterative pose information and the plurality of the iterative three-dimensional point position information to obtain an offline map.

[0022] Optionally, in a fourth implementation of the first aspect of the present invention, after S104, the step further includes:

[0023] S1041, cutting the offline map into N sub-maps, where N is greater than 1;

[0024] S1042: With the external device as the main body, a sub-map area is loaded in real time, wherein the sub-map area includes a plurality of sub-maps.

[0025] Optionally, in a fifth implementation of the first aspect of the present invention, after S1042, the step further includes:

[0026] S1043: Based on the moving path of the external device and the range of the sub-map area, the sub-map is added or deleted in real time and dynamically.

[0027] Optionally, in a sixth implementation of the first aspect of the present invention, S105 includes:

[0028] S1051, receiving IMU integral data of the external device, and obtaining new posture information corresponding to a plurality of new scene images;

[0029] S1052: According to the new SIFT feature points and the new Superpoint descriptors, the scene images close to the multiple new pose information are respectively selected in the sub-map area to obtain multiple matching images.

[0030] Optionally, in a seventh implementation manner of the first aspect of the present invention, S106 includes:

[0031] S1061, calculating the plurality of matching images based on a preset RANSAC-PNP algorithm to obtain a plurality of pose estimation information;

[0032] S1062. Perform weighted summation on the plurality of posture estimation information and calculate an average value to obtain positioning posture information.

[0033] A first implementation of the second aspect of the present invention provides a map positioning system, comprising:

[0034] The acquisition module is used to acquire multiple scene images and respectively extract SIFT feature points, posture information and three-dimensional point position information from each of the scene images;

[0035] An iteration module, configured to iterate the posture information and the three-dimensional point position information in each of the scene images based on a preset first algorithm to obtain iterative posture information and iterative three-dimensional point position information;

[0036] An offline module, used for replacing the SIFT feature points in each of the scene images with Superpoint descriptors, and integrating and storing a plurality of the Superpoint descriptors, a corresponding plurality of the iterative pose information, and a plurality of the iterative three-dimensional point position information to obtain an offline map;

[0037] A receiving module, used for receiving in real time a plurality of new scene images transmitted by a preset external device, and extracting new SIFT feature points and new Superpoint descriptors in each of the new scene images;

[0038] A matching module, used for matching the new SIFT feature points and the new Superpoint descriptors in the offline map to obtain a plurality of matching images;

[0039] The positioning module is used to calculate the multiple matching images based on a preset estimation algorithm to obtain positioning posture information.

[0040] A first implementation of the third aspect of the present invention provides a map positioning device, the map positioning device comprising: a memory and at least one processor, the memory storing instructions, the memory and the at least one processor being interconnected via a line;

[0041] The at least one processor calls the instructions in the memory to enable the map positioning device to execute the map positioning method as described in any one of the first aspects of the present invention.

[0042] A first implementation of the fourth aspect of the present invention provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the map positioning method as described in any one of the first aspects of the present invention is implemented.

[0043] Compared with the prior art, the present invention has the following beneficial technical effects: by collecting multiple scene images, extracting SIFT feature points, posture information and three-dimensional point position information, using a preset algorithm to iterate this information to obtain an optimized value, then replacing the SIFT feature points with Superpoint descriptors, and integrating relevant information to generate an offline map, receiving new scene images output by an external device in real time, extracting new SIFT feature points and Superpoint descriptors, matching these features in the offline map, finding multiple matching images, and finally calculating the matching images based on a preset estimation algorithm to obtain positioning posture information, and using a pre-constructed offline map, it can effectively achieve accurate positioning, high frequency and low calculation amount, facilitate the use of low-configuration devices, greatly reduce the cost of use while taking into account a good positioning effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 This is a schematic diagram of a first embodiment of a map positioning method in an embodiment of the present invention;

[0045] Figure 2 Schematic diagram of a second embodiment of the map positioning method in an embodiment of the present invention;

[0046] Figure 3 Schematic diagram of a third embodiment of the map positioning method in an embodiment of the present invention;

[0047] Figure 4 Schematic diagram of a fourth embodiment of a map positioning method according to an embodiment of the present invention;

[0048] Figure 5 Schematic diagram of a fifth embodiment of a map positioning method according to an embodiment of the present invention;

[0049] Figure 6 Schematic diagram of a sixth embodiment of a map positioning method according to an embodiment of the present invention;

[0050] Figure 7 Schematic diagram of a seventh embodiment of a map positioning method according to an embodiment of the present invention;

[0051] Figure 8 A schematic diagram of an embodiment of a map positioning system in an embodiment of the present invention;

[0052] Fig. 9 A schematic diagram of an embodiment of a map positioning device in an embodiment of the present invention;

[0053] Fig.10 The figure is a flowchart of the execution of the map positioning method in the embodiment of the present invention. DETAILED DESCRIPTION

[0054] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0055] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or component referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", and the like are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, features defined as "first", "second", and the like may explicitly or implicitly include one or more features. In the description of the present invention, unless otherwise specified, "multiple" means two or more.

[0056] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or a specific connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood by specific circumstances.

[0057] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1-Figure 7 as well as Fig.10 In an embodiment of the present invention, a map positioning method is used to detect a road. The map positioning method includes:

[0058] S101, collecting multiple scene images, and extracting SIFT feature points, posture information and three-dimensional point position information from each scene image;

[0059] In this embodiment, the step of collecting multiple scene images is to serve as a basis for the subsequent construction of an offline map. The devices used in this method are AR devices and VR devices, so the collection work is also performed by AR devices and VR devices. Before the collection work, it is necessary to complete the internal and external parameter calibration of the AR device and VR device. The internal and external parameter calibration refers to determining the focal length of the camera, the coordinates of the principal point (optical center), and the spatial position and rotation angle of the camera, which is crucial for subsequent image processing and three-dimensional reconstruction.

[0060] Furthermore, step S101 specifically includes:

[0061] S1011, collecting a plurality of scene images with an overlap rate of at least 50% between adjacent images;

[0062] S1012, extracting SIFT feature points in each scene image based on a preset SIFT algorithm;

[0063] S1013, matching multiple SIFT feature points, calculating the relative rotation data and translation data between adjacent scene images, to obtain the posture information and three-dimensional point position information of the corresponding scene image.

[0064] In this embodiment, during the shooting process using an AR device and a VR device, it is ensured that there is a certain overlapping area between two adjacent images. In this embodiment, it is preferred that there is an overlap rate of at least 50% to ensure that there are enough feature points that can be matched between adjacent images, thereby improving the accuracy and stability of feature matching. By using the matched SIFT feature points, the relative rotation data and translation data between the two scene images are calculated, wherein these rotation data and translation data can reflect the spatial transformation relationship between the scene images, which is beneficial to subsequent pose estimation and three-dimensional reconstruction.

[0065] S102, iterating the posture information and the three-dimensional point position information in each scene image based on a preset first algorithm to obtain iterative posture information and iterative three-dimensional point position information;

[0066] In this embodiment, the bundle adjustment technology is used to iteratively optimize the pose information and three-dimensional point position information in each scene image. By iteratively adjusting the pose information and three-dimensional point positions of all scene images, the reprojection error is minimized to obtain an accurate three-dimensional point cloud, i.e., iterative pose information and iterative three-dimensional point position information.

[0067] S103, replacing the SIFT feature points in each scene image with Superpoint descriptors, and integrating and storing multiple Superpoint descriptors, corresponding multiple iterative pose information, and multiple iterative three-dimensional point position information to obtain an offline map;

[0068] Furthermore, step S103 specifically includes:

[0069] S1031, according to the corresponding vector relationship between the SIFT feature points and the Superpoint descriptor, the SIFT feature points in each scene image are replaced with the Superpoint descriptor;

[0070] S1032, based on a preset greedy algorithm, calculating the effective range covered by each scene image, and screening out multiple screening scene images with the widest effective range;

[0071] S1033, integrating and storing the multiple filtered scene images, the multiple Superpoint descriptors, the corresponding multiple iterative pose information and the multiple iterative three-dimensional point position information to obtain an offline map.

[0072] In this embodiment, the Superpoint descriptor is a more modern feature point detection and description method, which aims to provide a more efficient and accurate feature description than traditional methods (such as SIFT). Therefore, the SIFT feature points are replaced with Superpoint descriptors through vector relationships to improve the jealousy and robustness of subsequent feature matching. According to the greedy algorithm, the coverage of each scene image is calculated, which refers to the size of the scene range seen in this image. The specific calculation method is to obtain the three-dimensional point cloud corresponding to the image based on the depth map of each scene image, and calculate the bounding box of the point cloud, that is, to obtain the coverage range corresponding to the image. In the implementation of this embodiment, a scene image is used as the starting point, and the next one is continuously added. The basis for adding the next scene image is that the overlapping range with the bounding box of the current scene image is minimized, and finally the purpose of covering the entire scene with the least scene images is achieved, thereby improving the computing efficiency. The greedy algorithm specifically adopts K-cover.

[0073] S104, receiving multiple new scene images transmitted by a preset external device in real time, and extracting new SIFT feature points and new Superpoint descriptors in each new scene image;

[0074] Furthermore, after step S104, the method further includes:

[0075] S1041, cutting the offline map into N sub-maps, where N is greater than 1;

[0076] S1042, taking the external device as the main body center, loading a sub-map area in real time, wherein the sub-map area includes a plurality of sub-maps;

[0077] S1043, based on the movement path of the external device and the range of the sub-map area, sub-maps are added or deleted in real time and dynamically.

[0078] In this embodiment, the external device here also uses AR devices and VR devices to collect scene images. It should be noted that the scene image collection here is not the same as the aforementioned image collection, but different collection steps under the same scene. When the offline map is loaded, it is intelligent and dynamic loading, and only the sub-map area composed of multiple sub-maps at the initial position will be loaded, where the initial position is the main center position of the external device. As the main body moves, the sub-map area will automatically unload the sub-maps that are out of range and reload the new sub-map, so that real-time positioning can be achieved using very small memory. The sub-map area includes 3*3 blocks of sub-maps.

[0079] S105, matching the new SIFT feature points and the new Superpoint descriptors in the offline map to obtain multiple matching images;

[0080] Furthermore, step S105 specifically includes:

[0081] S1051, receiving IMU integral data from an external device, and obtaining new posture information corresponding to a plurality of new scene images;

[0082] S1052: According to the new SIFT feature points and the new Superpoint descriptor, scene images close to the multiple new pose information are respectively selected in the sub-map area to obtain multiple matching images.

[0083] In this embodiment, the IMU can provide real-time data about the acceleration and rotation rate of the external device in space, and assist the positioning system to quickly estimate the position of the external device, that is, to obtain new posture information. Although these new posture information are relatively rough, they can still be used in subsequent image matching to reduce the search range and thus improve the matching efficiency. Multiple scene images similar to the new posture information are selected, specifically 20 images. In order to ensure the reliability of the matching, when performing feature point matching on these scene images, it is necessary to select matching points greater than 500.

[0084] S106: Based on a preset estimation algorithm, multiple matching images are calculated to obtain positioning posture information.

[0085] Furthermore, step S106 specifically includes:

[0086] S1061, based on a preset RANSAC-PNP algorithm, calculating multiple matching images to obtain multiple pose estimation information;

[0087] S1062. Perform weighted summation on the multiple pose estimation information and calculate the average value to obtain positioning pose information.

[0088] In this embodiment, the RANSAC-PNP algorithm is used to solve and obtain a series of pose estimation information from multiple matching images. The RANSAC algorithm can robustly estimate model parameters from a data set containing outliers, perform weighted summation on a series of pose estimates obtained by RANSAC-PNP, and then calculate the average value to obtain the final positioning pose. This step takes into account the uncertainty of each pose estimate and improves the accuracy and robustness of positioning by weighted averaging.

[0089] Specifically, it collects multiple scene images, extracts SIFT feature points, posture information and 3D point position information, uses a preset algorithm to iterate this information to obtain the optimized value, then replaces the SIFT feature points with Superpoint descriptors, and integrates relevant information to generate an offline map. It receives new scene images output by external devices in real time, extracts new SIFT feature points and Superpoint descriptors, matches these features in the offline map, finds multiple matching images, and finally calculates the matching images based on the preset estimation algorithm to obtain the positioning posture information. The pre-built offline map is used to effectively achieve accurate positioning, high frequency and low calculation amount, which is convenient for use with low-configuration devices, greatly reduces the cost of use while taking into account good positioning effects.

[0090] The map positioning method in the embodiment of the present invention is described above. The map positioning system in the embodiment of the present invention is described below. Figure 8 , the map positioning system comprises:

[0091] The acquisition module 201 is used to acquire multiple scene images and extract SIFT feature points, posture information and three-dimensional point position information from each scene image;

[0092] An iteration module 202, configured to iterate the pose information and the three-dimensional point position information in each scene image based on a preset first algorithm to obtain iterative pose information and iterative three-dimensional point position information;

[0093] The offline module 203 is used to replace the SIFT feature points in each scene image with Superpoint descriptors, and integrate and store multiple Superpoint descriptors, corresponding multiple iterative pose information and multiple iterative three-dimensional point position information to obtain an offline map;

[0094] The receiving module 204 is used to receive multiple new scene images transmitted by a preset external device in real time, and extract new SIFT feature points and new Superpoint descriptors in each new scene image;

[0095] A matching module 205 is used to match the new SIFT feature points and the new Superpoint descriptors in the offline map to obtain multiple matching images;

[0096] The positioning module 206 is used to calculate the multiple matching images based on a preset estimation algorithm to obtain positioning posture information.

[0097] Specifically, it collects multiple scene images, extracts SIFT feature points, posture information and 3D point position information, uses a preset algorithm to iterate this information to obtain the optimized value, then replaces the SIFT feature points with Superpoint descriptors, and integrates relevant information to generate an offline map. It receives new scene images output by external devices in real time, extracts new SIFT feature points and Superpoint descriptors, matches these features in the offline map, finds multiple matching images, and finally calculates the matching images based on the preset estimation algorithm to obtain the positioning posture information. The pre-built offline map is used to effectively achieve accurate positioning, high frequency and low calculation amount, which is convenient for use with low-configuration devices, greatly reduces the cost of use while taking into account good positioning effects.

[0098] The acquisition module 201 may also be specifically used for:

[0099] Collect multiple scene images with an overlap rate of at least 50% between adjacent images;

[0100] Based on the preset SIFT algorithm, SIFT feature points are extracted from each scene image;

[0101] Match multiple SIFT feature points and calculate the relative rotation data and translation data between adjacent scene images to obtain the pose information and 3D point position information of the corresponding scene image.

[0102] The offline module 203 may also be specifically used for:

[0103] According to the corresponding vector relationship between SIFT feature points and Superpoint descriptors, SIFT feature points in each scene image are replaced with Superpoint descriptors;

[0104] Based on the preset greedy algorithm, the effective range covered by each scene image is calculated, and multiple scene images with the widest effective range are screened out;

[0105] Multiple filtered scene images, multiple Superpoint descriptors, corresponding multiple iterative pose information and multiple iterative three-dimensional point position information are integrated and stored to obtain an offline map.

[0106] After receiving module 204, it can also be specifically used for:

[0107] Cut the offline map into N sub-maps, where N is greater than 1;

[0108] With the external device as the main center, the sub-map area is loaded in real time, wherein the sub-map area includes multiple sub-maps;

[0109] Based on the movement path of the external device and the scope of the sub-map area, sub-maps are added and deleted in real time and dynamically.

[0110] The matching module 205 may also be specifically used for:

[0111] Receive IMU integral data from an external device to obtain new pose information corresponding to multiple new scene images;

[0112] According to the new SIFT feature points and the new Superpoint descriptor, scene images similar to the multiple new pose information are screened out in the sub-map area to obtain multiple matching images.

[0113] The positioning module 206 may also be specifically used for:

[0114] Based on the preset RANSAC-PNP algorithm, multiple matching images are calculated to obtain multiple pose estimation information;

[0115] The weighted sum of multiple pose estimation information is performed and the average value is calculated to obtain the positioning pose information.

[0116] above Figure 8 The map positioning system in the embodiment of the present invention is described in detail from the perspective of modular functional entities, and the map positioning device in the embodiment of the present invention is described in detail from the perspective of hardware processing.

[0117] Fig. 9 : is a structural schematic diagram of a map positioning device provided by an embodiment of the present invention. The map positioning device 300 may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units, CPU) 310 (for example, one or more processors) and a memory 320, and one or more storage media 330 (for example, one or more mass storage devices) storing application programs 333 or data 332. Among them, the memory 320 and the storage medium 330 can be short-term storage or permanent storage. The program stored in the storage medium 330 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations in the map positioning device 300. Furthermore, the processor 310 can be configured to communicate with the storage medium 330 to execute a series of instruction operations in the storage medium 330 on the map positioning device 300.

[0118] The map-based positioning device 300 may also include one or more power supplies 340, one or more wired or wireless network interfaces 350, one or more input and output interfaces 360, and / or one or more operating systems 331, such as Windows Serve, Mac OS X, Unix, Linux, FreeBSD, etc. It will be appreciated by those skilled in the art that Figure 3 The map positioning device structure shown does not constitute a limitation on the communication protocol device based on LAN screen projection, and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0119] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of the map positioning method.

[0120] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program codes.

[0121] The above is a map positioning method or multiple implementation methods provided in combination with specific content, and it is not intended that the specific implementation of the present invention is limited to these descriptions. Any method, structure, etc. similar to or identical to the present invention, or any technical deduction or replacement based on the concept of the present invention, shall be deemed to be within the protection scope of the present invention.

Claims

1. A map positioning method, characterized in that: include: S101, collecting multiple scene images, and extracting SIFT feature points, posture information and three-dimensional point position information from each of the scene images; S102, based on a preset first algorithm, iterating the posture information and the three-dimensional point position information in each of the scene images to obtain iterative posture information and iterative three-dimensional point position information; S103, replacing the SIFT feature points in each of the scene images with Superpoint descriptors, and integrating and storing a plurality of the Superpoint descriptors, a corresponding plurality of the iterative pose information, and a plurality of the iterative three-dimensional point position information to obtain an offline map; S104, receiving multiple new scene images transmitted by a preset external device in real time, and extracting new SIFT feature points and new Superpoint descriptors in each of the new scene images; S105, matching the new SIFT feature points and the new Superpoint descriptors in the offline map to obtain a plurality of matching images; S106: Based on a preset estimation algorithm, calculate the multiple matching images to obtain positioning posture information.

2. A map positioning method according to claim 1, characterized in that: The S101 includes: S1011, collecting a plurality of scene images with an overlap rate of at least 50% between adjacent images; S1012, extracting SIFT feature points from each of the scene images based on a preset SIFT algorithm; S1013, matching a plurality of the SIFT feature points, and calculating the relative rotation data and translation data between adjacent scenes images to obtain the posture information and three-dimensional point position information corresponding to the scene images.

3. A map positioning method according to claim 2, characterized in that: The S103 includes: S1031, according to the corresponding vector relationship between the SIFT feature points and the Superpoint descriptor, replacing the SIFT feature points in each of the scene images with the Superpoint descriptor; S1032, based on a preset greedy algorithm, calculating the effective range covered by each of the scene images, and screening out a plurality of screening scene images that cover the widest effective range; S1033, integrating and storing the plurality of the filtered scene images, the plurality of the Superpoint descriptors, the corresponding plurality of the iterative pose information and the plurality of the iterative three-dimensional point position information to obtain an offline map.

4. A map positioning method according to claim 3, characterized in that: After S104, the method further includes: S1041, cutting the offline map into N sub-maps, where N is greater than 1; S1042: With the external device as the main body, a sub-map area is loaded in real time, wherein the sub-map area includes a plurality of sub-maps.

5. A map positioning method according to claim 4, characterized in that: After S1042, the method further includes: S1043: Based on the moving path of the external device and the range of the sub-map area, the sub-map is added or deleted in real time and dynamically.

6. A map positioning method according to claim 5, characterized in that: The S105 further includes: S1051, receiving IMU integral data of the external device, and obtaining new posture information corresponding to the plurality of new scene images; S1052: According to the new SIFT feature points and the new Superpoint descriptors, the scene images close to the multiple new pose information are respectively selected in the sub-map area to obtain multiple matching images.

7. A map positioning method according to claim 6, characterized in that: The S106 includes: S1061, calculating the plurality of matching images based on a preset RANSAC-PNP algorithm to obtain a plurality of pose estimation information; S1062. Perform weighted summation on the plurality of posture estimation information and calculate an average value to obtain positioning posture information.

8. A map positioning system, characterized in that: include: The acquisition module is used to acquire multiple scene images and respectively extract SIFT feature points, posture information and three-dimensional point position information from each of the scene images; An iteration module, configured to iterate the posture information and the three-dimensional point position information in each of the scene images based on a preset first algorithm to obtain iterative posture information and iterative three-dimensional point position information; An offline module, used for replacing the SIFT feature points in each of the scene images with Superpoint descriptors, and integrating and storing a plurality of the Superpoint descriptors, a corresponding plurality of the iterative pose information, and a plurality of the iterative three-dimensional point position information to obtain an offline map; A receiving module, used for receiving in real time a plurality of new scene images transmitted by a preset external device, and extracting new SIFT feature points and new Superpoint descriptors in each of the new scene images; A matching module, used for matching the new SIFT feature points and the new Superpoint descriptors in the offline map to obtain a plurality of matching images; The positioning module is used to calculate the multiple matching images based on a preset estimation algorithm to obtain positioning posture information.

9. A map positioning device, characterized in that: The map positioning device comprises: a memory and at least one processor, the memory stores instructions, and the memory and the at least one processor are interconnected via a line; The at least one processor calls the instructions in the memory to enable the map positioning device to execute the map positioning method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the map positioning method according to any one of claims 1 to 7 is implemented.

Images