A visible light communication-assisted indoor positioning method for perspective circles and arcs

By using the perspective circular and arc method assisted by visible light communication in indoor positioning, using the characteristics of cameras and circular lamps to establish auxiliary coordinate systems and space-time encoding models, the low positioning accuracy and equipment dependence in the prior art are solved. Problem, high-precision indoor positioning is achieved without the need for IMU and tilt angle limitations.

CN114299133BActive Publication Date: 2024-12-06BEIJING UNIV OF POSTS & TELECOMM
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111563989.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-12-20
Publication Date
2024-12-06
Estimated Expiration
2041-12-20

AI Technical Summary

Technical Problem

The existing indoor positioning technology based on circular lamps requires additional equipment or ideal projection models, and is limited by the receiver inclination angle and environmental changes, resulting in low positioning accuracy and difficult to widely use in practical applications.

Method used

A indoor positioning method for perspective circles and arcs assisted by visible light communication is proposed. By building four independent coordinate systems in the room, capturing the circular and arc features with the camera, establishing auxiliary coordinate systems and space-time encoding models, and computing Normal vector and Euler angle realize high-precision positioning without IMU and no tilt angle limitation.

Benefits of technology

High-precision positioning based on complete circular and arc features in circular lamp layout is realized, the system design is simplified, and positioning accuracy and feasibility is improved. The simulation results show that the proportion of samples with positioning error within 10cm reaches 95%.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114299133B_ABST
    Figure CN114299133B_ABST
Patent Text Reader

Abstract

The present invention proposes a visible light communication-assisted perspective circle and arc indoor positioning method, which belongs to the field of visible light communication and indoor positioning technology. Specifically, first, a transceiver communication scene is built indoors, and the receiving end is used as the object to be located to capture the light information of the transmitting end and form an image to find the target lamp; then, the normal vector is calculated through the complete circular image of the target lamp and the adjacent local lamp images that exist at the same time; then, the constructed VLC space-time coding model is used, combined with the principle of geometric projection, to calculate the center G of the target lamp and the virtual marker point M. According to the correspondence between the normal vector of the target lamp and the horizontal vector #imgabs0# in the CCS coordinate system and the WCS coordinate system, the rotation matrix and translation vector of the target lamp from CCS to WCS are calculated, which are the posture matrix and position coordinates of the camera in the WCS coordinate system respectively. The indoor system of the present invention is simpler to implement, and can achieve higher positioning accuracy and more accurate camera posture solution.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of visible light communication and indoor positioning, and in particular is a visible light communication-assisted indoor positioning method for perspective circles and arcs. Background Art

[0002] In recent years, the demand for location-based services, including positioning and navigation, has surged, making indoor precise positioning increasingly important. However, existing technologies such as WiFi, radio frequency identification (RFID), ultra-wideband (UWB) and Bluetooth all face the disadvantages of low accuracy or high installation costs. In contrast, with the popularization of light-emitting diodes (LEDs), visible light positioning (VLP) has gradually become a positioning technology with great development potential, with the characteristics of low cost, easy implementation and high positioning accuracy.

[0003] VLP positioning methods can be classified as follows: nearest neighbor method, fingerprint method, received signal strength method (RSS), time of arrival (TOA) and angle of arrival (AOA) method and image sensor method. Among them, the nearest neighbor method and fingerprint method can use a single LED to estimate the position of the receiver; however, the nearest neighbor method is not accurate enough, and the fingerprint method is also easily affected by environmental changes; the RSS algorithm determines the position of the receiver based on the power of the received signal and needs to rely on an accurate channel model; in addition, the TOA and AOA algorithms require complex hardware implementation to ensure positioning accuracy. Unlike the algorithm that uses a photodiode (PD) as a receiver, the image sensing algorithm uses a camera to receive the signal and determines the position of the receiver by analyzing the geometric relationship between the LED and its projection on the image plane.

[0004] Most existing VLP algorithms achieve positioning based on point light sources; however, in addition to point-shaped lamps, circular lamps are also a common type of light source. Most of the visible light positioning focused on circular lamps is based on image sensors. For example, reference: 1: R. Zhang, W.-D. Zhong, Q. Kemao, and S. Zhang, "A single LED positioning system based on circle projection," IEEE Photon. J., vol. 9, no. 4, pp. 1–9, Aug. 2017. By establishing a weak projection model, a point is marked on the edge of the circular lamp to assist in estimating the camera's pose. However, the large tilt angle results in an inaccurate approximation of the center of the ellipse projected onto the image plane, which limits the application of the positioning method.

[0005] Reference 2: H.Cheng, C.Xiao, Y.Ji, J.Ni, and T.Wang, "A single LED visible light positioning system based on geometric features and CMOS camera," IEEE Photon.Technol.Lett., vol.32, no.17, pp.1097–1100, Sep.2020. A simple positioning model based on the IMU component of a smartphone is proposed. The radius of the illuminator projection is approximated through the plane-intersection-line mechanism to estimate the position of the receiver. However, this positioning method relies on an additional device IMU, which may cause errors in the measured angle due to the influence of the magnetic field.

[0006] Reference 3: J.Hao, J.Chen, and R.Wang, “Visible light positioning using a single LED luminaire,” IEEE Photon.J., vol.11, no.5, pp.1–13, Oct.2019. The camera’s posture is estimated using IMU and weak projection model, and the azimuth error measured by IMU is reduced through calibration, thereby improving positioning accuracy.

[0007] In general, existing studies on indoor localization based on circular luminaires require additional equipment or ideal projection models, and these requirements may affect the practical application of localization technology.

[0008] In addition, circular features can also be located using algorithms based on computer vision. In reference 4: B. Huang, Y. Sun, Y. Zhu, Z. Xiong, and J. Liu, "Vision pose estimation from planar dual circles in a single image," Optik, vol. 127, no. 10, pp. 4275–4280, May 2016., the authors proposed a pose estimation method that uses two coplanar circles to eliminate the ambiguity of the plane. However, this method must capture two complete circles at the same time, which is sometimes not possible in practice.

[0009] In addition, Reference 5: C. Meng, Z. Li, H. Sun, D. Yuan, X. Bai, and F. Zhou, "Satellite pose estimation via single perspective circle and line," IEEE Trans. Aerosp. Electron. Syst., vol. 54, no. 6, pp. 3084–3095, June 2018. proposed a perspective circle and line method to estimate the satellite's pose. In indoor positioning scenarios, there may not be enough geometric features that can be captured by the camera.

[0010] In contrast, in an indoor environment with a circular lamp layout, it is easier to capture the geometric features of circles or arcs. Therefore, there may be a practical indoor positioning algorithm based on circular visual features, which can achieve high positioning accuracy without the need for an IMU and without limiting the receiver tilt angle. Summary of the invention

[0011] In view of the above problems, the present invention proposes a visible light communication-assisted perspective circle and arc indoor positioning method, which can achieve high-precision positioning based on complete circle and arc features in a circular lamp layout without IMU requirements or receiver tilt angle restrictions.

[0012] The specific steps are as follows:

[0013] Step 1: Build a communication scene including the receiving end and the transmitting end indoors, and establish four independent coordinate systems;

[0014] In the indoor positioning scenario, the transmitting end is a number of circular panel lamps randomly fixed on the ceiling of the room, and a number of LEDs are evenly distributed in each lamp; the receiving end is a general camera.

[0015] The transmitting end and the receiving end form an elliptical cone Γ, the bottom of Γ is a circular lamp, and the vertex is the receiving end.

[0016] Four independent coordinate systems, including pixel coordinate system PCS, image coordinate system ICS, camera coordinate system CCS and world coordinate system WCS.

[0017] Step 2: The transmitting end emits light, and the receiving end, as the object to be located, captures information and forms an image until a target lamp that meets the conditions is found;

[0018] The conditions are met if: the receiving end images have at least one complete circle and one partial light image at the same time;

[0019] Step 3: Establish an auxiliary coordinate system ACS; set the projection of the target lamp on the imaging plane of the receiving end to be an ellipse Πe , around the ellipse Π e The plane of the ellipse is rotated by the major axis and is aligned with the side surface Γ of the elliptical conic surface Γ c Intersect, there are two circular sections Π1 and Π2; find the two normal vectors corresponding to them in the ACS coordinate system and

[0020] In the ACS coordinate system, the side surface Γ of the elliptical cone Γ is c can be represented as a standard quadratic surface.

[0021] Step 4: Similarly, for another partial image or complete circular light image in the imaging, rotate it around the major axis of the ellipse and intersect with the side of the elliptical cone Γ to form two circular sections Π3 and Π4, and the corresponding two normal vectors are and

[0022] Step 5: Calculate the deviations between the four normal vectors, select the normal vector with the smallest deviation as the normal vector of the target lamp, and normalize it;

[0023] The calculation formula is:

[0024]

[0025] The normalized normal vector of the luminaire plane in the CCS coordinate system is expressed as:

[0026]

[0027] is the rotation matrix from the ACS coordinate system to the CCS coordinate system; k i is the coefficient that determines the direction of the normal vector of the circular plane;

[0028] Step 6: Build a VLC space-time coding model, drive different LEDs to form a light-emitting array, and combine the principle of geometric projection to calculate the intersection point as the center G of the target lamp and the virtual marker point M.

[0029] The virtual marker point M is located at the edge of the lamp and directly above the center of the circle G, satisfying the vector After normalization in the world coordinate system, we have

[0030] Step 7: According to the normal vector and horizontal vector of the target lamp The correspondence between WCS and CCS is calculated by calculating the target lamps around the CCS coordinate system axis x c ,y c and z c Euler angles of the axis θ and ψ;

[0031] First, the normal vector of the lamp in the world coordinate system is Set the normalized normal vector of the luminaire plane to Then we have:

[0032]

[0033] By solving and θ.

[0034] Then, Normalized in the WCS coordinate system Normalized in the CCS coordinate system Vector by Find out, then we have:

[0035]

[0036] The rotation angle ψ is obtained by solving.

[0037] Among them, G c is the coordinate of the center G in the CCS coordinate system; M c is the coordinate of the marker point M in the CCS coordinate system; M Tr is the coordinate of point M in the ACS coordinate system, G Tr is the coordinate of point G in the ACS coordinate system;

[0038] Step 8: In the CCS coordinate system, for the Euler angle θ and ψ, calculate the rotation matrix from the CCS coordinate system to the WCS coordinate system

[0039] The rotation matrix is ​​expressed as follows:

[0040]

[0041] Among them, R X Represents around x c The rotation matrix of the axis, R Y Represents around y c The rotation matrix of the axis, R Z Represents around z c The rotation matrices of the axes; respectively:

[0042]

[0043]

[0044]

[0045] Step 9: Use the coordinates of the center G of the target lamp and the marker point M in the CCS coordinate system and the WCS coordinate system respectively, combined with the rotation matrix Calculate the translation vector from the CCS coordinate system to the WCS coordinate system

[0046] The calculation formula is as follows:

[0047]

[0048] G w is the coordinate of the center G in the WCS coordinate system, M w is the coordinate of the marker point M in the WCS coordinate system,

[0049] Step 10: Rotation matrix from CCS coordinate system to WCS coordinate system and translation vectors That is, they are the posture matrix and position coordinates of the camera in the WCS coordinate system.

[0050] The advantages of the present invention are:

[0051] 1) A visible light communication-assisted indoor positioning method for perspective circles and arcs. In the visible light communication system, the receiver only uses a camera to obtain the image information of the LED, and no longer uses sensor devices such as photodiodes, and there is no need to use complex channel models.

[0052] 2) A visible light communication-assisted indoor positioning method for perspective circles and arcs. The indoor system is simpler to implement. Only two circular panel lamps and a camera are needed to form a complete transceiver system. The solution has stronger feasibility.

[0053] 3) A visible light communication-assisted indoor positioning method for perspective circles and arcs. Compared with other visible light indoor positioning solutions for circular lamps, it does not need to rely on IMU or limit the receiver tilt angle, and can achieve higher positioning accuracy and more accurate camera attitude solution; BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 A flowchart of a visible light communication-assisted indoor positioning method for perspective circles and arcs according to the present invention;

[0055] Figure 2 The communication scene constructed for the present invention using VLP and selected LED lamps for display;

[0056] Figure 3 It is a schematic diagram of four mutually independent coordinate systems adopted by the present invention;

[0057] Figure 4 It is the VLC system space-time coding model adopted by the present invention;

[0058] Figure 5 The elliptical conical surface formed by the center of the lamp and the LED plane in the auxiliary coordinate system x Tr Oz Tr The projection diagram below;

[0059] Figure 6 is a graph of cumulative distribution of errors of the present invention and other types of positioning algorithms;

[0060] Figure 7 This is a comparison chart of the pose solution errors of the present invention and other positioning technologies within a certain noise range;

[0061] Figure 8 This is a comparison chart of the posture solution error of the present invention and other positioning technologies as the LED radius changes; DETAILED DESCRIPTION

[0062] The present invention is further described in detail below in conjunction with the embodiments and drawings.

[0063] The present invention proposes a visible light communication assisted perspective circle and arc indoor positioning method (Visible Light Communication Assisted Perspective Circle and Arc, V-PCA); the present invention establishes an imaging model and the required plane and space rectangular coordinate systems to give the conversion relationship between the systems; and designs a space-time coding model based on VLC, thereby determining the positions where the center of a circular lamp and a virtual marker point form a projection in an image, and transmitting their world coordinate information to a receiver through a VLC system; then, the coordinates of the center of the circular lamp and the virtual marker point under CCS are calculated using spatial geometric information; then, the circular feature of a circular lamp and the arc feature of an adjacent lamp are used to obtain the normal vector of the target lamp in the camera coordinate system; finally, based on the single view geometry theory, the conversion relationship between corresponding points in different coordinate systems is analyzed, and combined with the existing coordinate information, the posture and position of the camera in space are estimated to complete the posture calculation.

[0064] Specifically: First, a space-time coding model is designed using VLC to determine the position of the projection of the center of the circular lamp and the virtual marker point in the image, and their world coordinates are transmitted to the receiver through the VLC system; then, by establishing an auxiliary coordinate system, the normal vector of the plane where the lamp is located in the camera coordinate system is estimated by using the captured complete circle of a lamp and the arc geometric features of another adjacent lamp. The coordinates of the center of the circle and the marker point and the normal vector under the camera system are combined with the single view geometry theory to solve the posture and position of the camera in the world coordinate system. The present invention can achieve high-precision positioning based on complete circle and arc features in the circular lamp layout, without IMU requirements or receiver tilt angle restrictions. Simulation results show that the V-PCA algorithm can make the proportion of samples with positioning errors within 10cm reach 95%.

[0065] like Figure 1 As shown, the specific steps are as follows:

[0066] Step 1: Build a communication scene including the receiving end and the transmitting end indoors, and establish four independent coordinate systems;

[0067] In indoor positioning scenarios, such as Figure 2 As shown, the transmitting end is a number of circular panel lamps, which are randomly fixed on the ceiling of the room, responsible for lighting and sending visible light information. The original lighting equipment and layout in the room can be reused through modification; a number of LED arrays are evenly distributed in each lamp as a transmitter; the receiving end is a universal camera; common mobile phones and tablets have built-in universal cameras.

[0068] The transmitting end and the receiving end form an elliptical cone Γ, the bottom of Γ is a circular lamp, and the vertex is the receiving end.

[0069] The projection diagram of the transmitting end is as follows: Figure 3 As shown in FIG. 1 , four independent coordinate systems are established in the projection model, including a pixel coordinate system (PCS). p -u p v p 、Image coordinate system (ICS) i -x i y i 、Camera coordinate system (CCS) c -x c y c z c and the world coordinate system (WCS) w-x w y w z w .

[0070] In PCS, ICS and CCS, the coordinate axis u p 、x i and x c Parallel to each other, v p ,y i and c The focal length f of the camera is equal to the two centers O i and O c The distance between two points, so the z coordinate of the image plane in CCS is expressed as z c =f.

[0071] The coordinates of the corresponding point i in CCS are obtained from the coordinates of any point i in ICS, expressed as:

[0072]

[0073] in represents the coordinates of point i in CCS; Represents the coordinates of the projection point i' of point i in ICS.

[0074] use Represents the pixel coordinates of any point i on the image plane. According to the single view geometry theory, the expression of the image coordinates of this point is obtained:

[0075]

[0076] in, and They are respectively the image plane along x i Axis and y i The physical size of the pixel block of the axis. u and f v are the focal ratios along the u-axis and v-axis in the pixel respectively. (u0,v0) T Represents the principal point O i The pixel coordinates of the camera are (u0, v0). T and focal ratio f u and f v , can be calibrated in advance.

[0077] Step 2: The transmitting end emits light, and the receiving end, as the object to be located, captures information and forms an image until a target lamp that meets the conditions is found;

[0078] The conditions are met if: the receiving end images at least one complete circle and one local light image simultaneously; or the receiving end images two complete circles; or three or more complete circles; or a combination of two complete circles and a local light image; or a combination of three or more complete circles and a local light image; and a lamp corresponding to a complete circle is randomly selected from the multiple complete circles as the target lamp.

[0079] Step 3: Establish an auxiliary coordinate system ACS; set the projection of the target lamp on the imaging plane of the receiving end to be an ellipse Π e , around the ellipse Π e The plane of the ellipse is rotated by the major axis and is aligned with the side surface Γ of the elliptical conic surface Γ c Intersect, there are two circular sections Π1 and Π2; find the two normal vectors corresponding to them in the ACS coordinate system and

[0080] The receiving end is a standard pinhole camera that can be tilted arbitrarily except being coplanar with the lamp; therefore, the transmitter and camera form an elliptical conical surface Γ, with the base being the circular lamp and the vertex being O c In the communication scenario, the circular lamp is used as a transmitter, and the projection of the circular lamp captured by the camera on the image plane is elliptical, which is defined as Π e ;

[0081] In ICS, let the ellipse Π e The equation is:

[0082] Ax i2 +Bx i y i +Cy i2 +Dx i +Ey i +1=0 (3)

[0083] Among them, A, B, C, D and E are parameters determined by curve fitting; the ellipse Π is obtained by image processing. e The image coordinates of the upper part; then the least squares fit is used to estimate the parameters A, B, C, D and E.

[0084] By introducing an auxiliary coordinate system (ACS) and representing the conical surface Γ of the elliptical cone Γ in this system c , the surface is symmetrical; the side surface Γ of the elliptical cone Γ in CCS c The general equation of is:

[0085] A 2 x c2 +Bf 2 xc y c +Cf 2 y c2 +Dfx c z c +Efy c z c +z c2 =0 (4)

[0086] It can also be expressed as:

[0087] [x c y c z c ]Q[x c y c z c ] T =0 (5)

[0088] In the formula is a real symmetric matrix, not a diagonal matrix. This is because the axis of the elliptical cone Γ does not coincide with any coordinate axis of the CCS, which makes the conical surface Γ c The equation is relatively complex. Angularize the matrix Q, that is, express equation (5) in the standard quadratic coordinate system; by setting the rotation matrix from ACS to CCS get:

[0089]

[0090] In the formula represents the coordinates of point i in the ACS. By replacing (6) with (5), we obtain:

[0091]

[0092] From equation (7), we can get is a matrix consisting of the eigenvectors of Q, which can diagonalize Q.

[0093] Then there are: Therefore, we can deduce from (7) that Γ in ACS c The curve equation is simplified to:

[0094] λ1(x Tr ) 2 +λ2(y Tr ) 2 +λ3(z Tr ) 2 =0 (8)

[0095] where λ1, λ2, and λ3 are the eigenvalues ​​of Q. Solve After that, the corresponding ACS coordinate system can be determined.

[0096] One of the coordinate axes in the ACS coordinate system is aligned with the central axis of the elliptical cone Γ and has the same origin as the CCS. i The values ​​of (i = 1, 2, 3) determine the relative positions of the coordinate axes of the elliptical cone Γ and the ACS. There are six possible relationships between λ1, λ2, and λ3, but only if one of them is less than 0 and the other two are greater than 0 can (8) hold. For example, when λ1>λ2>0 and λ3<0, the central axis of the elliptical cone Γ is Tr Axis-aligned, the major axis of the ellipse is parallel to y Tr axis, while the minor axis is parallel to the x Tr When λ1, λ2, and λ3 have other relationships, similar relationships exist between the elliptical cone and the coordinate axes.

[0097] Around the ellipse Π e The principal axis rotates the plane of the ellipse and is related to Γ c There must be two circular cross-sections, denoted as Π1 and Π2, of which only one must be parallel to the lamp. Π1 and Π2 have different directions, but both circles are parallel to the major axis. We define planes Π1 and Π2 as:

[0098] z Tr =k1x Tr +b (9)

[0099] z Tr =k2x Tr +b (10)

[0100] in, and They respectively determine the direction of the normal vector of the circular plane. b is a constant corresponding to k1 and k2. There are two possible normal vectors of the two planes in the coordinate system, which are expressed as and In fact, there should be only one circular plane corresponding to the lamp, so there is ambiguity between the plane and the corresponding normal vector.

[0101] Step 4: Similarly, for another partial image or complete circular light image in the imaging, rotate it around the major axis of the ellipse and intersect with the side of the elliptical conical surface Γ to form two circular sections Π3 and Π4, and the corresponding two normal vectors are and

[0102] In order to eliminate the plane ambiguity, the arc of another LED lamp is used to find the correct plane. This method does not require the shape of the adjacent lamp to be complete. This arc can be fitted to another ellipse Π f Similarly, using Π f Construct an elliptical cone with the camera vertex and rotate π fThe normal vector is of two planes.

[0103] Step 5: Calculate the deviations between the four normal vectors, select the normal vector with the smallest deviation as the normal vector of the target lamp, and normalize it;

[0104] Since the arc and the complete circular plane are both on the ceiling of the room, There must be two identical vectors between them. Then, the deviation between the vectors is calculated using the following expression:

[0105]

[0106] The optimal plane is obtained by calculation and the normal vector where i * is the optimal solution i obtained in (11).

[0107] The normalized normal vector of the luminaire plane in the CCS coordinate system is expressed as:

[0108]

[0109] is the rotation matrix from the auxiliary coordinate system ACS coordinate system to the CCS coordinate system; k i is the coefficient that determines the direction of the normal vector of the circular plane;

[0110] Step 6: Build a VLC space-time coding model, drive different LEDs to form a light-emitting array, and combine the principle of geometric projection to calculate the intersection point as the center G of the target lamp and the marking point M.

[0111] The geometric features of the luminaire are used for positioning, including circular outline, ellipse, e The center G of the plane and the virtual marker point M, points G and M are determined by the space-time coding model; the marker point M is located at the edge of the lamp and directly above the center G, satisfying the horizontal vector In the world coordinate system,

[0112] In WCS, the coordinates of point G and point M are and Broadcast via LEDs.

[0113] Point G' and point M' are the projections of point G and point M on the image plane respectively; in order to find the projections G' and M' while avoiding affecting the lighting, a spatiotemporal coding model is designed for the lamp;

[0114] like Figure 4As shown, the local brightness and darkness of the transmitter is controlled to find the coordinates of the target point. First, the projection point G' is found, which is the intersection of the line segments P'Q' and M'N', and P'Q' and M'N' are the projections of PQ and MN respectively. Similarly, the projection point M' can be used as the intersection of M'Q' and P'M', which are the projections of the lines MQ and PM respectively. The coordinates of points M', N', P', Q' can be obtained by image processing; therefore, the straight line in ICS where the line segment formed by points M', N', P', Q' (i.e., P'Q', M'N', M'Q', P'M') is located is represented by a two-point equation. Therefore, the coordinates of points M' and G' can be obtained by calculating the intersection of the straight lines.

[0115] For example, using the straight line equations of P'Q' and P'M', the coordinates of the center G' are obtained.

[0116] In addition, all LEDs on the same luminaire broadcast the luminaire's ID information and the world coordinate information of points G and M; generally, the luminaire is oriented vertically downward. Therefore, the unit normal vector of the luminaire in WCS is in(·) T Represents the transpose of a matrix.

[0117] The conversion from CCS to WCS is expressed as:

[0118]

[0119] G c represents the coordinates of point G in CCS, and represents the rotation matrix and translation vector from CCS to WCS; in addition, It also includes the rotation of the three orthogonal coordinate axes in CCS corresponding to the three orthogonal coordinate axes in WCS, so it can also be regarded as the posture of the camera. c =(0,0,0) T Sometimes, there are therefore It can be regarded as the position of the camera. From the above content, we can know that the positioning goal is to solve and

[0120] like Figure 5 As shown, the elliptical cone Γ is at x Tr Oz Tr Projection on the plane. Line segment L1L2 is the complete projection of the lamp captured by the camera, and the length of L1L2 is equal to the diameter 2R of the circular LED. Line segment L1L2 is the ellipse Π e In x Tr Oz Tr Projection on the plane. Line segment L1′L2′ is In x Tr OzTr In addition, the line L1O Tr and L2O Tr The equation is expressed as:

[0121]

[0122] By solving (9) and (14) simultaneously, the coordinates of points L1′ and L2′ can be obtained.

[0123] Here, it means and The plane equation of the lamp can be expressed as follows:

[0124] z Tr =k LED x Tr +b LED (15)

[0125] in, and M' in ACS Tr coordinate, The coordinates of Export, where M' c It can be obtained according to Section III-A and formula (1). Therefore, M'O Tr The two-point equation of a line can be expressed as:

[0126]

[0127] from Figure 5 It can be seen that the circular lamp plane and the straight line M'O Tr They intersect at point M, so by solving (15) and (16) together, we can get the coordinates M of point M in ACS Tr Similarly, by solving G′O Tr The coordinates of the intersection of the line (15) and the point G in ACS are obtained Tr Finally, through and G can be calculated c and M c .

[0128] Step 7: According to the normal vector and horizontal vector of the target lamp The correspondence between WCS and CCS is calculated by calculating the target lamps around the CCS coordinate system axis x c ,y c and z c Euler angles of the axis θ and ψ;

[0129] According to the single view geometry theory, the normalized normal vector of the luminaire plane is and The relationship between The coordinates can be further expressed as:

[0130]

[0131] By solving and θ.

[0132] Then, Normalized in the WCS coordinate system Normalized in the CCS coordinate system Vector by Find out;

[0133]

[0134] G c is the coordinate of the center G in the CCS coordinate system; M c is the coordinate of the marker point M in the CCS coordinate system;

[0135] The rotation angle ψ is obtained by solving.

[0136] Step 8: In the CCS coordinate system, for the Euler angle θ and ψ, calculate the rotation matrix from the CCS coordinate system to the WCS coordinate system

[0137] The rotation matrix is ​​expressed as follows:

[0138]

[0139] Assume R X Represents around x c The rotation matrix of the axis, R Y Represents around y c The rotation matrix of the axis, R Z Represents around z c The rotation matrices of the axes; respectively:

[0140]

[0141]

[0142]

[0143] Step 9: Use the coordinates of the center G of the target lamp and the marker point M in the CCS coordinate system and the WCS coordinate system respectively, combined with the rotation matrix Calculate the translation vector from the CCS coordinate system to the WCS coordinate system

[0144] The calculation formula is as follows:

[0145]

[0146] G w is the coordinate of the center G in the WCS coordinate system, M w is the coordinate of the marker point M in the WCS coordinate system, In equation (23), and Averaging is performed to reduce estimation errors.

[0147] Step 10: Rotation matrix from CCS coordinate system to WCS coordinate system and translation vectors That is, the posture matrix and position coordinates of the camera in the CCS coordinate system.

[0148] The indoor positioning scheme designed by the present invention is compared with the visible light positioning algorithm based on IMU (V-IMU) and the traditional PnP algorithm; first, the performance of the three algorithms in terms of the cumulative distribution function (CDF) of the position error is shown in Figure 2. Figure 6 As shown in the figure, it can be observed that V-PCA performs best among the three algorithms, and most samples under the V-PCA algorithm can achieve an accuracy of less than 10 cm. In contrast, 68% of the samples under the PnP algorithm can achieve an accuracy of about 10 cm, while for the V-IMU algorithm, only 20% of the samples can achieve the corresponding accuracy. This is because the V-IMU algorithm can only achieve a reliable approximation of weak projection when the tilt angle is small.

[0149] Comparison of pose solution errors of the three algorithms within a certain noise range, such as Figure 7 As shown, the image noise is modeled as having a zero expectation and a standard deviation σ n is Gaussian white noise with a range of 0 to 4 pixels. Figure 7 As can be seen from (a), the position error of the present invention is basically caused by image noise. As the image noise gradually increases from 0 to 4 pixels, the average positioning error of the present invention only increases from 0 cm to 9 cm, which is more stable than other algorithms. In addition, Figure 7 (b) further demonstrates the relationship between the posture error and the image noise mean. As the radius increases, the posture error mean value of the present invention only increases from 0% to 1.6%, which is more robust to image noise.

[0150] The present invention shows the change of positioning accuracy with the increase of the radius of the circular lamp, that is, the change curve of the mean value of the position error with the increase of the radius from 6 cm to 20 cm. Figure 8As shown in (a), the accuracy of position estimation increases with the increase of the radius of the lamp. The present invention shows the best performance in the algorithm comparison, and the mean of all position errors is kept below 15 cm, which is much better than the comparison algorithm. Figure 8 (b) The average values ​​of the pose errors of V-PCA and PnP are compared. It can be seen that the present invention is more accurate in pose estimation than PnP. As the radius increases, the pose error of the present invention decreases from 2% to 0.8%, while the average pose error of PnP decreases from 12% to 1.8%. This shows that the present invention is relatively less susceptible to the influence of the radius of the lamp than the PnP algorithm and has a higher adaptability to scene changes.

[0151] In summary, the present invention realizes the posture and position estimation of circular LEDs without IMU and tilt angle restrictions at the receiving end. A complete communication system can be constructed using only circular lamps and cameras, and effective image information is obtained through a special coding model. Compared with other positioning methods under the same layout, the present invention has higher positioning accuracy and better environmental adaptability. The simulation demonstration of the invention shows that the present invention can make the proportion of samples with an error within 10 cm reach 95%, and it has the feasibility of both platform laying and theoretical implementation.

[0152] The above is a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the principle of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A visible light communication-assisted indoor positioning method for perspective circles and arcs, characterized in that: The specific steps are as follows: Step 1: Build a communication scene including the receiving end and the transmitting end indoors, and establish four independent coordinate systems; The transmitting end and the receiving end form an elliptical cone Γ, with the circular lamp at the bottom and the receiving end at the apex; Four independent coordinate systems, including pixel coordinate system PCS, image coordinate system ICS, camera coordinate system CCS and world coordinate system WCS; Step 2: The transmitting end emits light, and the receiving end, as the object to be located, captures information and forms an image until a target lamp that meets the conditions is found; The conditions are met if: the receiving end images have at least one complete circle and one partial light image at the same time; Step 3: Using the complete circle of the target lamp in the plane imaging of the receiving end and the adjacent local lamp graphics that exist at the same time, calculate the two normal vectors corresponding to each in the auxiliary coordinate system ACS, and select the minimum value as the normal vector of the target lamp by calculating the deviation between the normal vectors; Step 4: Build a VLC space-time coding model, drive different LEDs to form a light-emitting array, and combine the principle of geometric projection to calculate the intersection point as the center G of the target lamp and the virtual marker point M; The virtual marker point M is located at the edge of the lamp and directly above the center of the circle G, satisfying the vector In the world coordinate system, Step 5: According to the normal vector and horizontal vector of the target lamp Correspondence between WCS and CCS, calculate the rotation matrix of the target lamp from the CCS coordinate system to the WCS coordinate system The rotation matrix is ​​expressed as follows: Among them, R X Represents the coordinate system x around the CCS c The rotation matrix of the axis, R Y Represents the CCS coordinate system y c The rotation matrix of the axis, R Z Represents the coordinate system z around the CCS c The rotation matrices of the axes; respectively: θ and ψ are the target lamps around the CCS coordinate system axis x c ,y c and z c Euler angles of the axes; Step 6: Use the coordinates of the center G of the target lamp and the marker point M in the CCS coordinate system and the WCS coordinate system respectively, combined with the rotation matrix Calculate the translation vector from the CCS coordinate system to the WCS coordinate system The calculation formula is as follows: G w is the coordinate of the center G in the WCS coordinate system, M w is the coordinate of the marker point M in the WCS coordinate system, G c is the coordinate of the center G in the CCS coordinate system; M c is the coordinate of the marker point M in the CCS coordinate system; is the rotation matrix from the ACS coordinate system to the CCS coordinate system; G Tr is the coordinate of point G in the ACS coordinate system; M Tr is the coordinate of point M in the ACS coordinate system; Rotation matrix from CCS coordinate system to WCS coordinate system and translation vectors That is, they are the posture matrix and position coordinates of the camera in the WCS coordinate system.

2. A visible light communication-assisted indoor positioning method for perspective circles and arcs as claimed in claim 1, characterized in that: In the indoor positioning scenario, the transmitting end is a number of circular panel lamps randomly fixed on the ceiling of the room, and a number of LEDs are evenly distributed in each lamp; the receiving end is a general camera.

3. The indoor positioning method of visible light communication-assisted perspective circles and arcs as claimed in claim 1, characterized in that: The calculation process of the normal vector of the target lamp is as follows: First, establish the auxiliary coordinate system ACS and set the projection of the target lamp on the imaging plane of the receiving end as the ellipse Π e , around the ellipse Π e The plane of the ellipse is rotated by the major axis and is aligned with the side surface Γ of the elliptical conic surface Γ c Intersecting, there are two circular cross sections Π1 and Π2; corresponding to the two normal vectors and Similarly, for another local image or complete circular light image in the imaging, it is rotated around the major axis of the ellipse and intersects with the side of the elliptical conical surface Γ to form two circular sections Π3 and Π4, and the corresponding two normal vectors are and Then, by calculating the deviations between the four normal vectors, the normal vector with the smallest deviation is selected as the normal vector of the target lamp and normalized; The calculation formula is: The normalized normal vector of the luminaire plane in the CCS coordinate system is expressed as: k i is the coefficient that determines the direction of the normal vector of the circular plane.

4. The indoor positioning method of visible light communication-assisted perspective circles and arcs as claimed in claim 1, characterized in that:The rotation matrix Central Euler Angle The calculation process of θ and ψ is: First, the normal vector of the lamp in the world coordinate system is Set the normalized normal vector of the luminaire plane to Then we have: By solving and θ; Then, Normalized in the WCS coordinate system Normalized in the CCS coordinate system Vector by Find out, then we have: The rotation angle ψ is obtained by solving.