A navigation enhancement network system based on "body reference" and its construction method

By using a navigation enhancement network system based on 'volume reference' and combining 5G transmission and multi-source fusion technology, it provides enhanced services based on 'volume' information, solving the problem of converting point coordinates to volume coordinates and achieving high-precision seamless indoor and outdoor positioning.

CN114646991BActive Publication Date: 2024-11-26AEROSPACE INFORMATION RES INST CAS
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210248804.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-03-14
Publication Date
2024-11-26
Estimated Expiration
2042-03-14

AI Technical Summary

Technical Problem

Existing technologies cannot effectively solve the conversion from point coordinates to volume coordinates, and cannot meet the needs of scenario-based, entity-based, and refined management, resulting in inaccurate positioning and high recognition error rate.

Method used

A navigation enhancement network system based on 'volume reference' is adopted. Through reference station determination, position calculation, data processing and information transmission system, combined with 5G transmission and multi-source fusion technology, it provides enhanced information services based on 'volume' information, and realizes the function of calculating unknown points from known points.

Benefits of technology

It improves positioning accuracy and recognition accuracy, achieves seamless indoor and outdoor positioning, and solves the problems of inaccurate positioning and high recognition error rate in traditional navigation systems.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114646991B_ABST
    Figure CN114646991B_ABST
Patent Text Reader

Abstract

The present invention discloses a navigation enhancement network system based on "body reference" and a construction method, including a base station determination system, a base station model position solution system, a base station network data processing system, an information transmission system and a terminal enhancement information processing system. The network system constructs a "base body" network based on 5G, vision, and GNSS multi-source fusion information, and forms an enhanced service of the "base body" for multi-source fusion terminals, thereby improving the positioning and recognition accuracy of three-dimensional measurement terminals and panoramic terminals. Compared with traditional navigation enhancement systems that can only provide differential correction information based on point positions, the network system can provide enhanced information based on "body" information, which can enhance the recognition accuracy in addition to position enhancement. At the same time, based on the characteristics of rapid 5G information transmission, services are provided to application terminals that support Beidou, INS, and vision functions, and the function of inferring the position of unknown points from known points is realized in any scenario, solving the problem of seamless positioning indoors and outdoors.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of navigation and positioning technology, and in particular to a navigation enhancement network system based on "body reference" and a construction method thereof. Background Art

[0002] Satellite navigation positioning is a basic space-time network that gives people precise location and time. The coordinate system obtained by GNSS is point coordinates (longitude, latitude and altitude). Surveying and mapping technology has changed from the original basic measurement to the current holographic surveying and mapping. In the future, real-life three-dimensional maps will replace the existing two-dimensional and image maps and become the mainstream of basic geographic information. The concept of positioning must also adapt to the development of new technologies. The measurement of objects abstracted as a point cannot meet the needs of scene-based, entity-based and refined management. The original geographic entity is changed from a particle position to a group of space-time point clouds to identify. The trend from point coordinates to body coordinates is inevitable, and no reasonable solution has been proposed yet. Summary of the invention

[0003] The purpose of the present invention is to provide a navigation enhancement network system and construction method based on "body reference" to solve the problems raised in the above background technology.

[0004] To achieve the above-mentioned purpose, the present invention provides the following technical solutions: A navigation enhancement network system based on "body reference", comprising: a reference station determination system, a reference station model position solution system, a reference station network data processing system, an information transmission system and a terminal enhancement information processing system;

[0005] The base station determination system selects static ground object points as reference points on the remote sensing image through visual interpretation, extracts and segments the target objects, and obtains the coordinate information of the ground object contour;

[0006] The reference station model position solving system utilizes the target recognition in the scene, and obtains the three-dimensional coordinate sequence of the reference station and the attribute information of the ground object through the twin scene and three-dimensional reconstruction technology;

[0007] The reference station network data processing system gathers the reference station network data provided by the reference station model position solution system in a certain area, and extracts the 5G, vision, and GNSS multi-source fusion information of each reference station in the reference station network data, obtains the feature matrix and precise coordinate position information, and builds an enhanced station database to prepare for the enhanced information service of the terminal enhanced information processing system;

[0008] The information transmission system uses 5G transmission to send information of surrounding reference points to the terminal enhanced information processing system;

[0009] The terminal enhanced information processing system is scattered in each terminal. It obtains its own precise position based on the terminal's own GNSS, INS and other sensors, and requests information about surrounding reference stations from the reference station network data processing system. After obtaining the surrounding visual reference stations and GNSS high-precision differential information, it enhances its own positioning function.

[0010] According to another aspect of the present invention, a method for constructing a navigation enhancement network system based on a "volume reference" is proposed, comprising the following steps:

[0011] Step 1: On the remote sensing image, through the method of visual interpretation, select static ground object points as reference points, extract and segment the target objects, and obtain the coordinate information of the ground object contour;

[0012] Step 2: The reference station model position solving system identifies the target in the scene, and obtains the three-dimensional coordinate sequence of the reference station and the ground feature attribute information through the twin scene and three-dimensional reconstruction technology. The ground feature attribute information is expressed in the form of a feature matrix;

[0013] Step 3: When the reference station network data of the predetermined area is uniformly gathered to the reference station network data processing system, the reference station network data processing system constructs an enhanced station database for the extracted feature matrix and precise coordinate position of each reference station, so as to prepare for the enhanced information service of the terminal;

[0014] Step 4: Use 5G transmission to send information about surrounding benchmark points to surrounding terminals that need services;

[0015] Step 5: The terminal obtains its own precise position based on its own GNSS sensor, and requests the information of surrounding reference stations from the reference station network data processing module. After obtaining the surrounding visual reference stations and GNSS high-precision differential information, it enhances its own positioning function.

[0016] Furthermore, the "volume reference" refers to the "point" on the ground in a two-dimensional map, which is displayed in the form of a "volume" in the real three-dimensional scene based on the "object-oriented" theory.

[0017] Furthermore, the mode position refers to the phase center position of the antenna obtained by Beidou positioning, specifically the particle position P (X, Y, Z), and then a set of points with spatial topological relationship is obtained by "Beidou + intelligent recognition" fusion positioning, which is called the mode position, specifically the particle position set P{p1, p2…pn}, where pn is the nth particle.

[0018] Furthermore, remote sensing pattern recognition technology is used to obtain static targets as "reference stations", GNSS is used to provide accurate spatiotemporal references to obtain the positions of reference stations, and remote sensing is used to obtain feature points of reference features, which are expressed in the form of feature matrices.

[0019] Furthermore, the reference station network is constructed in a uniformly distributed form within a certain area, and the reference stations are characterized by "body references", and enhanced information based on the "body references" is used as a basic component unit of information.

[0020] Furthermore, the information transmission module utilizes the 5G transmission network, and the data center provides the model position P{p1, p2…} of the surrounding "body reference" to the multi-source fusion terminal. The terminal calculates its own position or the position of surrounding objects through GNSS positioning and reference objects in the scene obtained by its own sensors. Different from the differential correction information based on point position provided by traditional navigation enhancement systems, the "body reference" provides enhanced information based on "body" information. In addition to enhancing the position information, it also enhances the recognition accuracy.

[0021] Furthermore, the terminal enhanced information processing system relies on the terminal to work. The terminal enhanced information processing system obtains the base station network data processing system information through the 5G transmission network and works in coordination with the terminal's own sensor information to realize the function of inferring the position of unknown points from known points, thereby solving the problem of seamless positioning indoors and outdoors.

[0022] Furthermore, in the reference station module position solving system in step S2, module position calculation includes the following steps:

[0023] Step S2.1>: Instance segmentation

[0024] ResNet-101+FPN is used as the backbone, and instance segmentation is achieved through two branches, namely the prototype generation branch and the mask coefficient branch. The prototype generation branch Protonet is mainly based on FCN to generate mask prototypes. The mask coefficient branch mainly predicts the mask coefficient through the anchor-based target detector. The two branches are calculated independently, and then the mask is synthesized through matrix multiplication and Sigmoid function, and the segmentation result is finally output;

[0025] Step S2.2>: Target depth acquisition

[0026] A binocular camera is used to obtain depth information. Assuming the target at coordinate (i, j), its depth value is Z(i, j); then a neighborhood window is used to estimate the depth value at each pixel position; where S is the size of the neighborhood window; is a symbolic function;

[0027] Step S2.3>: Calculation of external parameter matrix

[0028] The camera pose estimation is performed through a set of n 3D points in the world coordinate system and their corresponding 2D coordinates in the image. The EPnP (Efficient PnP) algorithm is used to achieve pose estimation.

[0029] Step S2.4>: Position calculation

[0030] By transforming the pixel coordinate system, the three-dimensional position of the target in the world coordinate system can be obtained.

[0031] The conversion relationship between pixel coordinates (u, v) and image coordinate system is:

[0032] If fx and fy are the focal lengths of the camera on the x-axis and y-axis, and Xc, Yc, and Zc are the coordinates of the target in the camera coordinate system, then:

[0033]

[0034] The (x, y) refers to the position of the target in the image coordinate system, the (u, v) refers to the position of the target in the pixel coordinate system, the (u0, v0) refers to the parameter for transferring the image coordinate origin from the center of the image to the upper left corner, and the (dx, dy) refers to the size of each pixel on the x-axis and y-axis;

[0035] If fx and fy are the focal lengths of the camera on the x-axis and y-axis, and Xc, Yc, and Zc are the coordinates of the target in the camera coordinate system, then:

[0036]

[0037] Given the pixel coordinates (u, v) of the target and the intrinsic matrix of the camera, the position information is solved based on the previously solved camera extrinsics and target depth; R represents the rotation matrix, that is, the rotation of the camera relative to the world coordinate system, and T represents the translation matrix, that is, the translation of the camera relative to the world coordinate system;

[0038] The (Xw, Yw, Zw) refers to the three-dimensional position of the target in the world coordinate system.

[0039] Beneficial effects:

[0040] Compared with the prior art, the traditional navigation enhancement system and construction method of the present invention provide differential correction information based on point position, while the present invention provides enhancement information based on "body" information. In addition to position enhancement, the recognition accuracy is also enhanced. Based on the characteristics of 5G information rapid transmission, services are provided for comprehensive application terminals based on Beidou / inertial navigation / vision, and the function of inferring the position of unknown points from known points is realized in any scenario, which completely solves the problem of seamless positioning indoors and outdoors. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0042] Figure 1 It is a schematic diagram of a navigation enhancement network construction principle based on a “body reference” according to an embodiment of the present invention.

[0043] Figure 2 4 is a schematic diagram of a navigation enhancement network service based on “body reference” according to an embodiment of the present invention. DETAILED DESCRIPTION

[0044] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the protection scope of the present invention.

[0045] See also Figure 1-2 To achieve the above purpose according to an embodiment of the present invention, the present invention provides the following technical solution: A navigation enhancement network system based on "body reference", comprising:

[0046] Reference station determination system, reference station model position solution system, reference station network data processing system, information transmission system and terminal enhanced information processing system;

[0047] The base station determination system selects static ground points as reference points on high-precision remote sensing images through visual interpretation, extracts and segments the target objects, and obtains coordinate information of the ground object contour; the high precision is at the centimeter level.

[0048] The reference station model position solving system utilizes the target recognition in the scene, and obtains the three-dimensional coordinate sequence of the reference station and the attribute information of the ground object through the twin scene and three-dimensional reconstruction technology;

[0049] The reference station network data processing system gathers the reference station network data provided by the reference station model position solution system in a certain area, and extracts data from the 5G, vision, GNSS and other multi-source fusion information of each reference station in the reference station network data, obtains information such as feature matrix and precise coordinate position, and builds an enhanced station database to prepare for the enhanced information service of the terminal enhanced information processing system;

[0050] The information transmission system utilizes the high-speed transmission characteristics of 5G to send information of surrounding reference points to the terminal enhanced information processing system;

[0051] The terminal enhanced information processing system is scattered in each terminal. It obtains its own precise position based on the terminal's own GNSS, INS and other sensors, and requests information about surrounding reference stations from the reference station network data processing system. After obtaining information such as surrounding visual reference stations and GNSS high-precision differentials, it enhances its own positioning function.

[0052] According to another embodiment of the present invention, a method for constructing a navigation enhancement network system based on a "volume reference" is proposed, characterized in that it includes the following steps:

[0053] Step S1, on a high-precision remote sensing image, by visual interpretation, a static object point is selected as a reference point, the target object is extracted and segmented, and the coordinate information of the object contour is obtained;

[0054] Step S2: The model position of the reference station is obtained by using the target recognition in the scene, through the twin scene and three-dimensional reconstruction technology, to obtain the three-dimensional coordinate sequence of the reference station and the ground feature attribute information, wherein the ground feature attribute information is expressed in the form of a feature matrix;

[0055] Step S3: When the reference station network data of a certain area are uniformly gathered to the reference station network data processing system, the reference station network data processing system constructs an enhanced station database for the extracted feature matrix, precise coordinate position, etc. of each reference station to prepare for the enhanced information service of the terminal;

[0056] Step S4: Using the high-speed transmission characteristics of 5G, information of surrounding reference points is sent to surrounding terminals that need services;

[0057] Step S5: The terminal obtains its own precise position based on its own GNSS sensor, and requests the information of surrounding reference stations from the reference station network data processing system. After obtaining information such as surrounding visual reference stations and GNSS high-precision differentials, the terminal enhances its own positioning function.

[0058] Furthermore, the "volume benchmark" refers to the "point" on the ground at a certain scale in a two-dimensional map, which is displayed in the form of a "volume" in the real three-dimensional scene based on the "object-oriented" theory.

[0059] Furthermore, the mode position refers to the phase center position of the antenna obtained by Beidou positioning, specifically the particle position P (X, Y, Z), and then a set of points with spatial topological relationship is obtained by "Beidou + intelligent recognition" fusion positioning, which is called the mode position, specifically the particle position set P{p1, p2...}.

[0060] Furthermore, remote sensing pattern recognition technology is used to obtain static targets as "reference stations", GNSS is used to provide accurate spatiotemporal references to obtain the positions of reference stations, and remote sensing is used to obtain feature points of reference features, which are expressed in the form of feature matrices.

[0061] Furthermore, the reference station network is constructed in a uniformly distributed form within a certain area. Furthermore, the reference station is characterized by a "body reference", and enhanced information based on the "body reference" is used as a basic constituent unit of the present technology.

[0062] Furthermore, the information transmission system utilizes the 5G high-transmission network, and the data center provides the model position P{p1, p2…} of the surrounding "body reference" to the multi-source fusion terminal. The terminal calculates its own position or the position of surrounding objects through GNSS positioning and reference objects in the scene obtained by its own sensors.

[0063] Furthermore, the terminal enhanced information processing system relies on the terminal to work. Furthermore, the terminal enhanced information processing system obtains the base station network data processing system information through the 5G transmission network and works in coordination with the terminal's own sensor information, thereby realizing the function of inferring the position of unknown points from known points in any scenario, completely solving the problem of seamless positioning indoors and outdoors.

[0064] According to one embodiment of the present invention, the mold position calculation in step S2 includes the following steps:

[0065] Step S2.1: Instance Segmentation

[0066] ResNet-101+FPN is used as the backbone, and instance segmentation is achieved through two branches, namely the prototype generation branch and the mask coefficient branch. The prototype generation branch Protonet is mainly based on FCN to generate mask prototypes. The mask coefficient branch mainly predicts the mask coefficient through the anchor-based target detector. The two branches are calculated independently, and then the mask is synthesized through matrix multiplication and Sigmoid function, and the segmentation result is finally output;

[0067] Step S2.2: Target depth acquisition

[0068] A binocular camera is used to obtain depth information. Assuming the target at coordinate (i, j), its depth value is Z(i, j); then a neighborhood window is used to estimate the depth value at each pixel position; where S is the size of the neighborhood window. is a symbolic function;

[0069] Step S2.3: Calculation of the external parameter matrix

[0070] The camera pose estimation is performed through a set of n 3D points in the world coordinate system and their corresponding 2D coordinates in the image. The EPnP (Efficient PnP) algorithm is used to achieve pose estimation.

[0071] Step S2.4: Position calculation

[0072] By transforming the pixel coordinate system, the three-dimensional position of the target in the world coordinate system can be obtained.

[0073] The conversion relationship between pixel coordinates (u, v) and image coordinate system is:

[0074]

[0075] The (x, y) refers to the position of the target in the image coordinate system, the (u, v) refers to the position of the target in the pixel coordinate system, the (u0, v0) refers to the parameter for transferring the image coordinate origin from the center of the image to the upper left corner, and the (dx, dy) refers to the size of each pixel on the x-axis and y-axis;

[0076]

[0077] If fx and fy are the focal lengths of the camera on the x-axis and y-axis, and Xc, Yc, and Zc are the coordinates of the target in the camera coordinate system, then, given the pixel coordinates (u, v) of the target and the intrinsic parameter matrix of the camera, the position information can be solved based on the previously solved camera extrinsic parameters and target depth;

[0078] The (Xw, Yw, Zw) refers to the three-dimensional position of the target in the world coordinate system.

[0079] Compared with the prior art, the traditional navigation enhancement system of the present invention provides differential correction information based on point position, while the present technology provides enhancement information based on "body" information. In addition to position enhancement, it also enhances the recognition accuracy. Based on the characteristics of 5G information rapid transmission, it provides services for comprehensive application terminals based on Beidou / inertial navigation / vision, and realizes the function of inferring the position of unknown points from known points in any scenario, thus solving the problem of seamless positioning indoors and outdoors.

[0080] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for constructing a navigation enhancement network system based on a "body reference", the navigation enhancement network system comprising: Reference station determination system, reference station model position solution system, reference station network data processing system, information transmission system and terminal enhanced information processing system; The base station determination system selects static ground object points as reference points on the remote sensing image through visual interpretation, extracts and segments the target objects, and obtains the coordinate information of the ground object contour; The reference station model position solving system utilizes the target recognition in the scene, and obtains the three-dimensional coordinate sequence of the reference station and the attribute information of the ground object through the twin scene and three-dimensional reconstruction technology; The reference station network data processing system gathers the reference station network data provided by the reference station model position solution system in a certain area, and extracts the 5G, vision, and GNSS multi-source fusion information of each reference station in the reference station network data, obtains the feature matrix and precise coordinate position information, and builds an enhanced station database to prepare for the enhanced information service of the terminal enhanced information processing system; The information transmission system uses 5G transmission to send information of surrounding reference points to the terminal enhanced information processing system; The terminal enhanced information processing system is dispersed in each terminal, and obtains its own precise position based on the terminal's own GNSS or INS sensor, and requests the information of surrounding reference stations from the reference station network data processing system, and enhances its own positioning function after obtaining the surrounding visual reference stations and GNSS high-precision differential information; It is characterized by comprising the following steps: Step 1: On the remote sensing image, through the method of visual interpretation, select static ground object points as reference points, extract and segment the target objects, and obtain the coordinate information of the ground object contour; Step 2: The reference station model position solution system identifies the target in the scene, and obtains the three-dimensional coordinate sequence of the reference station and the attribute information of the ground object through the twin scene and three-dimensional reconstruction technology. The attribute information of the ground object is expressed in the form of a feature matrix; the static target is obtained as the "reference station" by using remote sensing pattern recognition technology, the position of the reference station is obtained by using GNSS to provide a precise spatiotemporal reference, and the feature points of the surface image of the reference ground object are obtained by remote sensing. The feature points are expressed in the form of a feature matrix; The base station model position solution system in step 2, the model position calculation includes the following steps: Step 2.1: Instance Segmentation Based on the neural network algorithm, the target object is identified and the image is segmented to propose the image features of the target object; Step 2.2: Target depth acquisition A binocular camera is used to obtain depth information. Assuming the target at coordinate (i, j), its depth value is Z(i, j); then a neighborhood window is used to estimate the depth value at each pixel position; Step 2.3: Calculation of the external parameter matrix The camera pose estimation is performed through a set of n 3D points in the world coordinate system and their corresponding 2D coordinates in the image. The EPnP (Efficient PnP) algorithm is used to achieve pose estimation. Step 2.4: Position calculation By transforming the coordinate system of pixels, the three-dimensional position of the target in the world coordinate system is obtained; The conversion relationship between pixel coordinates (u, v) and image coordinate system is: , The (x, y) refers to the position of the target in the image coordinate system, the (u, v) refers to the position of the target in the pixel coordinate system, the (u0, v0) refers to the parameter for transferring the image coordinate origin from the center of the image to the upper left corner, and the (dx, dy) refers to the size of each pixel on the x-axis and y-axis; If fx and fy are the focal lengths of the camera on the x-axis and y-axis, and Xc, Yc, and Zc are the coordinates of the target in the camera coordinate system, then: , Given the pixel coordinates (u, v) of the target and the intrinsic matrix of the camera, the position information is solved based on the previously solved camera extrinsics and target depth; R represents the rotation matrix, that is, the rotation of the camera relative to the world coordinate system, and T represents the translation matrix, that is, the translation of the camera relative to the world coordinate system; The (Xw, Yw, Zw) refers to the three-dimensional position of the target in the world coordinate system; Step 3: When the reference station network data of the predetermined area is uniformly gathered to the reference station network data processing system, the reference station network data processing system constructs an enhanced station database for the extracted feature matrix and precise coordinate position of each reference station, so as to prepare for the enhanced information service of the terminal; Step 4: Use 5G transmission to send information about surrounding benchmark points to surrounding terminals that need services; Step 5: The terminal obtains its own precise position based on its own GNSS sensor, and requests the information of surrounding reference stations from the reference station network data processing module. After obtaining the surrounding visual reference stations and GNSS high-precision differential information, it enhances its own positioning function.

2. According to claim 1, a method for constructing a navigation enhancement network system based on "volume reference" is characterized in that: The "volume benchmark" refers to the theory based on "object-oriented". In a two-dimensional map, the feature is abstracted into a "point" on the ground, and in the real three-dimensional scene, it is displayed in the form of a "volume" of "a series of points + semantic information".

3. According to the method for constructing a navigation enhancement network system based on "volume reference" in claim 1, it is characterized in that: The mode position refers to the phase center position of the antenna obtained by Beidou positioning, specifically the particle position P (X, Y, Z), and then a set of points with spatial topological relationship is obtained by "Beidou + intelligent recognition" fusion positioning, which is called the mode position, specifically the particle position set P{p1, p2...pn}, where pn is the nth particle.

4. According to the method for constructing a navigation enhancement network system based on "volume reference" in claim 1, it is characterized in that: The reference station network is constructed in a uniformly distributed form within a certain area. The reference station is characterized by a "body reference" and enhanced information based on the "body reference" is used as a basic component unit of information.

5. According to the method for constructing a navigation enhancement network system based on "volume reference" in claim 1, it is characterized in that: The information transmission module uses the 5G transmission network, and the data center provides the model position P{p1, p2...} of the surrounding "body reference" to the multi-source fusion terminal. The terminal uses GNSS positioning and the reference objects in the scene obtained by its own sensors to infer its own or surrounding ground objects. Different from the differential correction information based on point position provided by the traditional navigation enhancement system, the "body reference" provides enhanced information based on "body" information. In addition to enhancing the position information, it also enhances the recognition accuracy.

6. According to the method for constructing a navigation enhancement network system based on "volume reference" in claim 1, it is characterized in that: The terminal enhanced information processing system relies on the terminal to work. The terminal enhanced information processing system obtains the base station network data processing system information through the 5G transmission network and works in coordination with the terminal's own sensor information to realize the function of inferring the position of unknown points from known points and solve the problem of seamless positioning indoors and outdoors.