Radar echo-based three-dimensional coordinate system registration method without prior information

Abstract

The invention discloses a radar echo-based no-prior-information three-dimensional coordinate system registration method, which comprises the following steps of: performing moving target detection on target echo signals acquired by two radars to obtain distance-Doppler domain signals of the two radars, screening a unit with the highest amplitude, and further obtaining distance-Doppler domain indexes of the radars; the distance-Doppler domain indexes are registered through a conversion formula, an objective function is established according to the registered distance-Doppler domain indexes and solved through a genetic algorithm, and finally parameters of a rotation matrix and a translation vector between the radar A and the radar B are efficiently and accurately obtained through gradient descent optimization.

Description

Technical Field

[0001] The invention belongs to the technical field of radar echoes, and in particular relates to a three-dimensional coordinate system registration method without prior information based on radar echoes. Background Art

[0002] Denied environment refers to environmental conditions that have an obstructive, restrictive or blocking effect on specific actions or activities. In the military field, the denied environment includes powerful enemy air defense systems, anti-ship missile networks, electronic warfare interference, etc., which make it difficult for one party to smoothly carry out air, sea or land combat operations in the area; when one's own radar is in a denied environment, that is, when it is strongly interfered by the enemy radar, our command center cannot obtain the specific location and posture of our radar.

[0003] In the prior art, the patent application "A method for spatial-temporal registration of networked radars with arbitrary motion" with publication number CN112346024A discloses a spatial registration algorithm for networked radars at low communication rates in the range-Doppler domain. After the moving target is detected, part of the signal is transmitted to the fusion center, and the correlation of the signal extracted by the local radar is constructed as the objective function. The maximization problem of the objective function is solved by a genetic algorithm to obtain the estimated values ​​of the rotation matrix and the translation vector.

[0004] Low communication volume is achieved by filtering out part of the information with larger amplitude in the signal after moving target detection. In the subsequent process, multiple coherent processing interval cycles are performed on each target, that is, several points are selected from the trajectory of a target for alignment. Moreover, due to the short coherent processing interval time, the target trajectory is approximately on a straight line, which will cause redundant and wasteful data.

[0005] In the paper "Application of SVD Algorithm in Point Cloud Data Registration of Buildings", the SVD algorithm is used in point cloud data registration to calculate the rotation matrix and translation vector to achieve registration of physical scanned point cloud data. When the prerequisites for using the quaternion array method cannot be met, the SVD algorithm can still achieve the calculation.

[0006] By using the original point set consisting of 1800 3D points and rotating them around the x, y, and z axes by known angles, the matrix to be registered is obtained. The accuracy of the algorithm is determined by comparing the calculated rotation matrix with the known rotation matrix through the SVD decomposition algorithm. The algorithm uses 1800 data pairs, and there is no translation change during the simulation setting. The data volume is large and the scene is too simple. When the point cloud data is small, the rotation matrix and translation vector cannot be accurately calculated. Summary of the invention

[0007] In order to solve the above problems existing in the prior art, the present invention provides a three-dimensional coordinate system registration method without prior information based on radar echo:

[0008] The technical problem to be solved by the present invention is achieved by the following method, comprising the following steps:

[0009] Initial scene setting steps: establish an initial scene including the target, radar A and radar B, detect the target through radar A and radar B respectively to obtain the target echo signal of radar A and the target echo signal of radar B;

[0010] Moving target detection step: performing moving target detection on the target echo signal of radar A and the target echo signal of radar B respectively to obtain the range-Doppler domain signal of radar A and the range-Doppler domain signal of radar B;

[0011] Screening and registration steps: The range-Doppler domain signal of radar A and the range-Doppler domain signal of radar B are screened to obtain the highest amplitude unit of the target echo signal of radar A and the highest amplitude unit of the target echo signal of radar B, and the range-Doppler domain index (D) of the target in the coordinate system of radar A is determined according to the highest amplitude unit of the target echo signal. A ,R A ) and the range-Doppler domain index (D B ,R B ), according to the conversion formula (D A ,R A ) and (D B ,R B ) are respectively registered to obtain the range-Doppler domain index (X B ,Y B ) and the range-Doppler domain index (X A ,Y A );

[0012] Rounding steps: Use the rounding formula to round (X B ,Y B ) and (X A ,Y A ) are rounded to obtain the rounded range-Doppler domain index (X B1 ,Y B1 ) and (X A1 ,Y A1 );

[0013] Objective function setting and solution steps: According to (X B1 ,Y B1 ) and (X A1 ,Y A1) setting the objective function; using a genetic algorithm to calculate and solve the objective function to obtain the parameters of the rotation matrix between radar A and radar B and the parameters of the translation vector between radar A and radar B;

[0014] Gradient descent optimization step: The parameters of the rotation matrix between radar A and radar B and the parameters of the translation vector between radar A and radar B are optimized by the gradient descent method to obtain the optimized parameters of the rotation matrix between radar A and radar B and the translation vector between radar A and radar B.

[0015] Furthermore, moving target detection includes:

[0016] The target echo signal of radar A and the target echo signal of radar B are divided according to the range unit respectively, and the slow time signal in the range unit is discrete Fourier transformed to obtain the first one-dimensional spectrum and the second one-dimensional spectrum; the range-Doppler domain signal of radar A is determined according to the first one-dimensional spectrum and the range unit, and the range-Doppler domain signal of radar B is determined according to the second one-dimensional spectrum and the range unit.

[0017] Further, the conversion formula includes a conversion formula for converting the target under radar B coordinates to radar A coordinates and a conversion formula for converting the target under radar A coordinates to radar B coordinates;

[0018] The conversion formula for converting the target from the radar B coordinate to the radar A coordinate is:

[0019] PA=R×PB+r

[0020] The conversion formula for converting the target from the radar A coordinate to the radar B coordinate is:

[0021] PB=R×PA+r

[0022] Among them, PA is the target in the coordinate system of radar A, PB is the target in the coordinate system of radar B, R is the 3×3 dimensional rotation matrix between radar A and radar B, and r is the 3×1 dimensional translation vector between radar A and radar B.

[0023] Furthermore, R is represented as follows:

[0024]

[0025] Among them, α, β, and γ represent the angles of rotation around the z, y, and x axes respectively.

[0026] Further, r is represented as follows:

[0027] r=[rx,ry,rz]

[0028] Among them, rx, ry, and rz represent the translation distances along the x, y, and z axes respectively.

[0029] Furthermore, the polar coordinate representations of PA and PB are:

[0030]

[0031] Among them, ρ A ,θ A , are the distance, elevation and azimuth of the target acquired by radar A in the local coordinate system; ρ B ,θ B , are the distance, elevation angle and azimuth of the target obtained by radar B in the local coordinate system.

[0032] Furthermore, the range-Doppler domain index corresponding to the target echo signal of radar A in radar B is expressed as follows:

[0033]

[0034] The range-Doppler domain index corresponding to the target echo signal of radar B in radar A is expressed as follows:

[0035]

[0036] Among them, X B is the velocity unit index corresponding to the target echo signal of radar B in radar A, Y B is the range unit index corresponding to the target echo signal of radar B in radar A; A is the velocity unit index corresponding to the target echo signal of radar A in radar B, Y A is the range unit index corresponding to the target echo signal of radar A in radar B; Δf k =v A / Δd A -v B / Δd B , Δf k is the discretization degree of the radial velocity of the target at the velocity resolution unit scale, v A 、v B are the radial velocities of the target relative to radar A and radar B, Δd A , Δd B are the velocity resolution units of radar A and radar B, ΔR A , ΔR B are the distance resolution units of radar A and B respectively, R A , D Aare the distance unit index of the target in the radar A coordinate system and the speed unit index of the target in the radar A coordinate system, R B , D B They are the distance unit index of the target in the radar A coordinate system and the speed unit index of the target in the radar A coordinate system respectively.

[0037] Furthermore, the rounding formula is:

[0038] X B1 =X B

[0039]

[0040] X A1 =X A

[0041]

[0042] Among them, X B1 Y is the velocity unit index corresponding to the rounded target echo signal of radar B in radar A, B1 is the range unit index corresponding to the rounded target echo signal of radar B in radar A; X A1 is the velocity unit index corresponding to the rounded target echo signal of radar A in radar B, Y A1 is the range unit index corresponding to the rounded target echo signal of radar A in radar B; Yes B The index value obtained by rounding down, Yes B The index value obtained by rounding up, Yes A The index value obtained by rounding down, Yes A The index value obtained by rounding up. floor is the function for rounding down, and ceil is the function for rounding up.

[0043] Further, according to (X B1 ,Y B1 ) and (X A1 ,Y A1 )Set the objective function:

[0044] According to (X B1 ,Y B1 ) and (X A1 ,Y A1 ) respectively determine a first specific unit of the target echo signal and a second specific unit of the target echo signal;

[0045] The target echo signal amplitude in the first specific unit and the target echo signal amplitude corresponding to the range-Doppler domain index of the target in the radar A coordinate system are normalized and then multiplied to obtain the correlation coefficient corr1;

[0046] The target echo signal amplitude in the second specific unit and the target echo signal amplitude corresponding to the range-Doppler domain index of the target in the radar B coordinate system are normalized and then multiplied to obtain the correlation coefficient corr2;

[0047] The target echo signal amplitude in the first specific unit and the target echo signal amplitude in the second specific unit are normalized respectively and then multiplied to obtain a correlation coefficient corr3;

[0048] According to the correlation coefficients corr1, corr2 and corr3, the objective function is set as:

[0049] -min([abs(corr1),abs(corr2),abs(corr3)])

[0050] Here, abs stands for absolute value function.

[0051] Compared with the prior art, the present invention has the following beneficial effects:

[0052] The present invention performs moving target detection on target echo signals acquired by two radars to obtain range-Doppler domain signals of the two radars, selects the unit with the highest amplitude therein to further obtain the range-Doppler domain index of the radar, aligns the range-Doppler domain index through a conversion formula, establishes an objective function according to the aligned range-Doppler domain index and solves it through a genetic algorithm, and finally obtains the parameters of the rotation matrix and the parameters of the translation vector between the radar A and the radar B through gradient descent optimization; the present invention only requires the coordinates of the target under the two radars, and aligns the maximum amplitude unit of the range-Doppler domain signal, thereby reducing the amount of alignment data and improving the parameter solution operation efficiency; at the same time, a one-dimensional correlation coefficient is considered, and the parameters obtained by the genetic algorithm are optimized by using the gradient descent method, so that the parameters are more accurate.

[0053] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 It is a flow chart of a three-dimensional coordinate system registration method without prior information based on radar echo provided by an embodiment of the present invention;

[0055] Figure 2 is an initial scene graph of the settings provided by an embodiment of the present invention;

[0056] Figure 3is a graph of the number of iterations and the corresponding objective function provided by an embodiment of the present invention;

[0057] Figure 4 is an iterative process diagram of a parameter α of a rotation matrix between radar A and radar B provided by an embodiment of the present invention;

[0058] Figure 5 is an iterative process diagram of the parameter β of the rotation matrix between the radar A and the radar B provided by an embodiment of the present invention;

[0059] Figure 6 is an iterative process diagram of the parameter γ of the rotation matrix between radar A and radar B provided by an embodiment of the present invention;

[0060] Figure 7 is an iterative process diagram of a parameter rx of a translation vector between radar A and radar B provided by an embodiment of the present invention;

[0061] Figure 8 is an iterative process diagram of the parameter ry of the translation vector between the radar A and the radar B provided by an embodiment of the present invention;

[0062] Fig. 9 is an iterative process diagram of a parameter rz of a translation vector between radar A and radar B provided by an embodiment of the present invention;

[0063] Fig.10 It is a graph of the objective function after being solved by a genetic algorithm and then optimized by gradient descent, provided by an embodiment of the present invention;

[0064] Fig.11 It is a target image of radar B registered to the radar A coordinate system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0065] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the scheme according to the present invention is described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0066] The above and other technical contents, features and effects of the present invention are clearly presented in the following detailed description of the specific implementation modes in conjunction with the accompanying drawings. Through the description of the specific implementation modes, the technical means and effects adopted by the present invention to achieve the predetermined purpose can be more deeply and specifically understood. However, the attached drawings are only for reference and explanation purposes and are not used to limit the technical solutions of the present invention.

[0067] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants are intended to cover non-exclusive inclusion, so that an article or device including a series of elements includes not only those elements, but also other elements that are not explicitly listed. In the absence of more restrictions, the elements defined by the statement "including one..." do not exclude the existence of other identical elements in the article or device including the elements.

[0068] like Figure 1 FIG. 1 is a flowchart of a method for three-dimensional coordinate system registration without prior information based on radar echoes provided by an embodiment of the present invention. The method comprises the following steps:

[0069] Step 1: Initial scene setting: Establish an initial scene including the target, radar A and radar B. Use radar A and radar B to detect the target respectively to obtain the target echo signal of radar A and the target echo signal of radar B. Then, according to the target echo signal, obtain the distance ρ of the target acquired by radar A in the local coordinate system. B , pitch angle θ A and azimuth The distance ρ of the target acquired by radar B in the local coordinate system B , pitch angle θ B and azimuth

[0070] Step 2, moving target detection: Perform moving target detection on the target echo signal of radar A and the target echo signal of radar B respectively to obtain the range-Doppler domain signal of radar A and the range-Doppler domain signal of radar B.

[0071] Specifically, moving target detection is performed by dividing the target echo signal of radar A and the target echo signal of radar B according to the distance unit, performing discrete Fourier transform calculation on the slow time signal in the distance unit to obtain the first one-dimensional spectrum and the second one-dimensional spectrum; the range-Doppler domain signal of radar A is determined according to the first one-dimensional spectrum and the distance unit, and the range-Doppler domain signal of radar B is determined according to the second one-dimensional spectrum and the distance unit. Compared with the echo signal, the range-Doppler domain signal is accumulated in the frequency domain, which is conducive to extracting target information under low signal-to-noise ratio.

[0072] Step 3: Screening and registration.

[0073] Step 3.1: The range-Doppler domain signal of radar A and the range-Doppler domain signal of radar B are screened respectively to obtain the unit with the highest amplitude of the target echo signal of radar A and the unit with the highest amplitude of the target echo signal of radar B, and the unit with the highest amplitude is selected as the unit where the target is located.

[0074] Step 3.2: Determine the range-Doppler domain index (D) of the target in the radar A coordinate system according to the unit with the highest amplitude of the target echo signal. A ,R A ) and the range-Doppler domain index (D B ,R B ).

[0075] Specifically, (D A ,R A ) includes the range unit index R of the target in the radar A coordinate system A and the target's velocity unit index D in the radar A coordinate system A , (D B ,R B ) includes the range unit index R of the target in the radar B coordinate system B and the target's velocity unit index D in radar B coordinate system B The distance unit index is obtained by measuring the delay time of the target echo signal, and the speed unit index is obtained by measuring the Doppler frequency shift of the target echo signal.

[0076] Step 3.3: According to the conversion formula, (D A ,R A ) and (D B ,R B ) are respectively registered to obtain the range-Doppler domain index (X B ,Y B ) and the range-Doppler domain index (X A ,Y A ).

[0077] Specifically, the conversion formula includes a conversion formula for converting a target under radar B coordinates to radar A coordinates and a conversion formula for converting a target under radar A coordinates to radar B coordinates;

[0078] The conversion formula for converting the target from the radar B coordinate to the radar A coordinate is:

[0079] PA=R×PB+r

[0080] The conversion formula for converting the target from the radar A coordinate to the radar B coordinate is:

[0081] PB=R×PA+r

[0082] Among them, PA is the target in the coordinate system of radar A, PB is the target in the coordinate system of radar B, R is the 3×3 dimensional rotation matrix between radar A and radar B, and r is the 3×1 dimensional translation vector between radar A and radar B.

[0083] R represents the following:

[0084]

[0085] Among them, α, β, and γ represent the angles of rotation around the z, y, and x axes respectively.

[0086] r represents the following:

[0087] r=[rx,ry,rz]

[0088] Among them, rx, ry, and rz represent the translation distances along the x, y, and z axes respectively.

[0089] The polar coordinates of PA and PB are:

[0090]

[0091] Among them, ρ A ,θ A , are the distance, elevation and azimuth of the target acquired by radar A in the local coordinate system; ρ B ,θ B , are the distance, elevation angle and azimuth of the target obtained by radar B in the local coordinate system.

[0092] The range-Doppler domain index corresponding to the target echo signal of radar A in radar B is expressed as follows:

[0093]

[0094] The range-Doppler domain index corresponding to the target echo signal of radar B in radar A is expressed as follows:

[0095]

[0096] Among them, X B is the velocity unit index corresponding to the target echo signal of radar B in radar A, Y B is the range unit index corresponding to the target echo signal of radar B in radar A; A is the velocity unit index corresponding to the target echo signal of radar A in radar B, Y A is the range unit index corresponding to the target echo signal of radar A in radar B; Δfk =v A / Δd A -v B / Δd B , Δf k is the discretization degree of the radial velocity of the target at the velocity resolution unit scale, v A 、v B are the radial velocities of the target relative to radar A and radar B, Δd A , Δd B are the velocity resolution units of radar A and radar B, ΔR A , ΔR B are the distance resolution units of radar A and B respectively, R A , D A are the distance unit index of the target in the radar A coordinate system and the speed unit index of the target in the radar A coordinate system, R B , D B They are the distance unit index of the target in the radar A coordinate system and the speed unit index of the target in the radar A coordinate system respectively.

[0097] Step 4: Since (X B ,Y B ) and (X A ,Y A ) is not necessarily an integer, so we need to use the rounding formula to round (X B ,Y B ) and (X A ,Y A ) are rounded to obtain the rounded range-Doppler domain index (X B1 ,Y B1 ) and (X A1 ,Y A1 ).

[0098] The rounding formula is:

[0099] X B1 =X B

[0100]

[0101] X A1 =X A

[0102]

[0103] Among them, X B1 Y is the velocity unit index corresponding to the rounded target echo signal of radar B in radar A, B1 is the range unit index corresponding to the rounded target echo signal of radar B in radar A; XA1 is the velocity unit index corresponding to the rounded target echo signal of radar A in radar B, Y A1 is the range unit index corresponding to the rounded target echo signal of radar A in radar B; Yes B The index value obtained by rounding down, Yes B The index value obtained by rounding up, Yes A The index value obtained by rounding down, Yes A The index value obtained by rounding up. floor is the function for rounding down, and ceil is the function for rounding up.

[0104] Step 5: Setting and solving the objective function.

[0105] Step 5.1: According to (X B1 ,Y B1 ) and (X A1 ,Y A1 )Set the objective function, specifically:

[0106] According to the rounded range-Doppler domain index (X B1 ,Y B1 ) and (X A1 ,Y A1 ) respectively determine the first specific unit of the target echo signal and the second specific unit of the target echo signal; normalize the target echo signal amplitude in the first specific unit and the target echo signal amplitude corresponding to the range-Doppler domain index of the target in the radar A coordinate system, and then multiply them to obtain the correlation coefficient corr1; normalize the target echo signal amplitude in the second specific unit and the target echo signal amplitude corresponding to the range-Doppler domain index of the target in the radar B coordinate system, and then multiply them to obtain the correlation coefficient corr2; normalize the target echo signal amplitude in the first specific unit and the target echo signal amplitude in the second specific unit, and then multiply them to obtain the correlation coefficient corr3; set the objective function according to the correlation coefficients corr1, corr2 and corr3 as follows:

[0107] -min([abs(corr1),abs(corr2),abs(corr3)])

[0108] Here, abs stands for absolute value function.

[0109] Step 5.2: Use a genetic algorithm to calculate and solve the objective function to obtain the parameters that determine the rotation matrix between radar A and radar B and the parameters of the translation vector between radar A and radar B.

[0110] Specifically, each parameter of the genetic algorithm and its upper and lower bounds are specified. The parameters are divided into parameters α, β, γ that determine the rotation matrix R between radar A and radar B and the parameter of the translation vector r between radar A and radar B. α, β, γ represent the angles of rotation around the z, y, and x axes, respectively. rx, ry, rz represent the translation distances along the x, y, and z axes, respectively. The upper and lower bounds of the first three angle parameters α, β, γ are [-pi / 2, pi / 2], and the upper and lower bounds of the last three translation distance parameters are [-3000, 3000].

[0111] 200,000 initial populations were randomly generated and the objective function value of each individual was calculated. The minimum and average values ​​were found, and the individual with the minimum objective function value was selected for crossover mutation to generate a new generation of population until the iteration termination condition was met. The iteration termination condition of the genetic algorithm was that the objective function value was less than -0.95.

[0112] Step 6, gradient descent optimization step: The parameters of the rotation matrix between radar A and radar B and the parameters of the translation vector between radar A and radar B are optimized by gradient descent method to obtain the optimized parameters of the rotation matrix between radar A and radar B and the translation vector between radar A and radar B.

[0113] Specifically, when the genetic algorithm ends and the objective function value is less than -0.95, the gradient descent method is used to optimize the parameters of the rotation matrix between radar A and radar B and the parameters of the translation vector between radar A and radar B. The gradient descent method uses the result obtained by the genetic algorithm as the initial value, calculates the gradient of the objective function for each parameter at this point, and updates the parameter size according to the gradient value to obtain the optimized parameter result. The optimized parameters are the parameters α, β, γ that determine the rotation matrix between radar A and radar B and the parameters rx, ry, rz of the translation vector between radar A and radar B.

[0114] like Figure 2 As shown, it is the initial scene diagram set by the present invention, the initial rotation angles between the radar A coordinate system and the radar B coordinate system are pi / 3, pi / 3, 0 respectively, and the coordinate system origin of radar B in the radar A coordinate system is 500, 200, 300.

[0115] like Figure 3 As shown, it is a graph of the number of iterations and the corresponding objective function of the present invention, wherein the horizontal axis is the number of iterations, and the vertical axis is the objective function -min([abs(corr1), abs(corr2), abs(corr3)]). It can be seen that the objective function value tends to be stable after 20 iterations.

[0116] like Figure 4As shown, it is an iterative process diagram of the parameter α of the rotation matrix between the radar A and the radar B of the present invention. It can be seen that α gradually tends to be stable during the iterative process of the genetic algorithm.

[0117] like Figure 5 As shown, it is an iterative process diagram of the parameter β of the rotation matrix between radar A and radar B of the present invention. It can be seen that β gradually tends to be stable during the iterative process of the genetic algorithm.

[0118] like Figure 6 As shown, it is an iterative process diagram of the parameter γ of the rotation matrix between radar A and radar B of the present invention. It can be seen that γ gradually tends to be stable during the iterative process of the genetic algorithm.

[0119] like Figure 7 As shown, it is an iterative process diagram of the parameter rx of the translation vector between the radar A and the radar B of the present invention. It can be seen that rx gradually tends to be stable during the iterative process of the genetic algorithm.

[0120] like Figure 8 As shown, it is an iterative process diagram of the parameter ry of the translation vector between the radar A and the radar B of the present invention. It can be seen that ry gradually tends to be stable during the iterative process of the genetic algorithm.

[0121] like Fig. 9 As shown, it is an iterative process diagram of the parameter rz of the translation vector between the radar A and the radar B of the present invention. It can be seen that rz gradually tends to be stable during the iterative process of the genetic algorithm.

[0122] like Fig.10 The above is the objective function value corresponding to the number of iterations after the genetic algorithm is solved and then the gradient descent optimization is performed. It can be seen that the objective function value is about -0.997752 after the gradient descent optimization.

[0123] As shown in Table 1, this is the final minimum value of the objective function value and the corresponding optimal solution of the parameters in the embodiment of the present invention.

[0124] Table 1 Minimum values ​​of target parameters and corresponding optimal solutions

[0125]

[0126] It can be seen from Table 1 that when the minimum value of the objective function is -0.9978, the parameters determining the rotation matrix parameters α, β, γ between radar A and radar B and the parameters rx, ry, rz of the translation vector between radar A and radar B are the optimal solutions.

[0127] like Fig.11As shown, the optimal solution parameters α, β, γ, rx, ry, rz in Table 1 are substituted into the rotation matrix and the translation vector to obtain the target of radar B being aligned to the radar A coordinate system. It can be seen that there is a good degree of consistency with the target in the A coordinate system. Therefore, the parameters α, β, γ of the rotation matrix between radar A and radar B and the parameters rx, ry, rz of the translation vector between radar A and radar B obtained by the present invention have high accuracy.

[0128] The present embodiment of the present invention provides a radar echo-based three-dimensional coordinate system registration method without prior information. The method performs moving target detection on target echo signals obtained by two radars to obtain range-Doppler domain signals of the two radars, selects the unit with the highest amplitude therein to further obtain the range-Doppler domain index of the radar, registers the range-Doppler domain index through a conversion formula, establishes an objective function according to the registered range-Doppler domain index and solves it through a genetic algorithm, and finally obtains the parameters of the rotation matrix and the parameters of the translation vector between radar A and radar B through gradient descent optimization. The present invention only requires the coordinates of the target under the two radars, and reduces the amount of registration data and improves the efficiency of parameter solution operation by registering the maximum amplitude unit of the range-Doppler domain signal. At the same time, the one-dimensional correlation coefficient is considered, and the parameters obtained by the genetic algorithm are optimized by the gradient descent method, so that the parameters are more accurate.

[0129] The above contents are further detailed descriptions of the present invention in combination with specific preferred embodiments, and it cannot be determined that the specific implementation of the present invention is limited to these descriptions. For ordinary technicians in the technical field to which the present invention belongs, several simple deductions or substitutions can be made without departing from the concept of the present invention, which should be regarded as falling within the protection scope of the present invention.

Claims

1. A three-dimensional coordinate system registration method without prior information based on radar echo, characterized in that: The following steps are involved: Initial scene setting steps: establish an initial scene including the target, radar A and radar B, detect the target through radar A and radar B respectively to obtain the target echo signal of radar A and the target echo signal of radar B; Moving target detection step: performing moving target detection on the target echo signal of radar A and the target echo signal of radar B respectively to obtain the range-Doppler domain signal of radar A and the range-Doppler domain signal of radar B; Screening and registration steps: The range-Doppler domain signal of radar A and the range-Doppler domain signal of radar B are screened to obtain the highest amplitude unit of the target echo signal of radar A and the highest amplitude unit of the target echo signal of radar B, and the range-Doppler domain index (D) of the target in the coordinate system of radar A is determined according to the highest amplitude unit of the target echo signal. A ,R A ) and the range-Doppler domain index (D B ,R B ), according to the conversion formula (D A ,R A ) and (D B ,R B ) are respectively registered to obtain the range-Doppler domain index (X B ,Y B ) and the range-Doppler domain index (X A ,Y A ); Rounding steps: Use the rounding formula to round (X B ,Y B ) and (X A ,Y A ) are rounded to obtain the rounded range-Doppler domain index (X B1 ,Y B1 ) and (X A1 ,Y A1 ); Objective function setting and solution steps: According to (X B1 ,Y B1 ) and (X A1 ,Y A1 ) setting the objective function; using a genetic algorithm to calculate and solve the objective function to obtain the parameters of the rotation matrix between radar A and radar B and the parameters of the translation vector between radar A and radar B; Gradient descent optimization step: The parameters of the rotation matrix between radar A and radar B and the parameters of the translation vector between radar A and radar B are optimized by the gradient descent method to obtain the optimized parameters of the rotation matrix between radar A and radar B and the translation vector between radar A and radar B.

2. The method for three-dimensional coordinate system registration without prior information based on radar echo according to claim 1, characterized in that: The moving target detection comprises: The target echo signal of radar A and the target echo signal of radar B are divided according to the range unit respectively, and the slow time signal in the range unit is discrete Fourier transformed to obtain the first one-dimensional spectrum and the second one-dimensional spectrum; the range-Doppler domain signal of radar A is determined according to the first one-dimensional spectrum and the range unit, and the range-Doppler domain signal of radar B is determined according to the second one-dimensional spectrum and the range unit.

3. The method for three-dimensional coordinate system registration without prior information based on radar echo according to claim 1, characterized in that: The conversion formula includes a conversion formula for converting a target under radar B coordinates to radar A coordinates and a conversion formula for converting a target under radar A coordinates to radar B coordinates; The conversion formula for converting the target from the radar B coordinate to the radar A coordinate is: PA=R×PB+r The conversion formula for converting the target from the radar A coordinate to the radar B coordinate is: PB=R×PA+r Among them, PA is the target in the coordinate system of radar A, PB is the target in the coordinate system of radar B, R is the 3×3 dimensional rotation matrix between radar A and radar B, and r is the 3×1 dimensional translation vector between radar A and radar B.

4. The method for three-dimensional coordinate system registration without prior information based on radar echo according to claim 3, characterized in that: The R is represented as follows: Among them, α, β, and γ represent the angles of rotation around the z, y, and x axes respectively.

5. The method for three-dimensional coordinate system registration without prior information based on radar echo according to claim 3, characterized in that: The r is represented as follows: r=[rx,ry,rz] Among them, rx, ry, and rz represent the translation distances along the x, y, and z axes respectively.

6. The method for three-dimensional coordinate system registration without prior information based on radar echo according to claim 3, characterized in that: The polar coordinates of PA and PB are: Among them, ρ A ,θ A , are the distance, elevation and azimuth of the target acquired by radar A in the local coordinate system; ρ B ,θ B , are the distance, elevation angle and azimuth of the target obtained by radar B in the local coordinate system.

7. The method for three-dimensional coordinate system registration without prior information based on radar echo according to claim 1, characterized in that: The range-Doppler domain index corresponding to the target echo signal of radar A in radar B is expressed as follows: The range-Doppler domain index corresponding to the target echo signal of radar B in radar A is expressed as follows: (X A ,AND A )= Among them, X B is the velocity unit index corresponding to the target echo signal of radar B in radar A, Y B is the range unit index corresponding to the target echo signal of radar B in radar A; A is the velocity unit index corresponding to the target echo signal of radar A in radar B, Y A is the range unit index corresponding to the target echo signal of radar A in radar B; Δf k =v A / Δd A -v B / Δd B , Δf k is the discretization degree of the radial velocity of the target at the velocity resolution unit scale, v A 、v B are the radial velocities of the target relative to radar A and radar B, Δd A , Δd B are the velocity resolution units of radar A and radar B, ΔR A , ΔR B are the distance resolution units of radar A and B respectively, R A , D A are the distance unit index of the target in the radar A coordinate system and the speed unit index of the target in the radar A coordinate system, R B , D B They are the distance unit index of the target in the radar A coordinate system and the speed unit index of the target in the radar A coordinate system respectively.

8. The method for three-dimensional coordinate system registration without prior information based on radar echo according to claim 1, characterized in that: The rounding formula is: X B1 =X B X A1 =X A Among them, X B1 Y is the velocity unit index corresponding to the rounded target echo signal of radar B in radar A, B1 is the range unit index corresponding to the rounded target echo signal of radar B in radar A; X A1 is the velocity unit index corresponding to the rounded target echo signal of radar A in radar B, Y A1 is the range unit index corresponding to the rounded target echo signal of radar A in radar B; Yes B The index value obtained by rounding down, Yes B The index value obtained by rounding up, Yes A The index value obtained by rounding down, Yes A The index value obtained by rounding up. floor is the function for rounding down, and ceil is the function for rounding up.

9. The method for three-dimensional coordinate system registration without prior information based on radar echo according to claim 1, characterized in that: The basis (X B1 ,Y B1 ) and (X A1 ,Y A1 ) Setting the objective function includes: According to (X B1 ,Y B1 ) and (X A1 ,Y A1 ) respectively determine a first specific unit of the target echo signal and a second specific unit of the target echo signal; The target echo signal amplitude in the first specific unit and the target echo signal amplitude corresponding to the range-Doppler domain index of the target in the radar A coordinate system are normalized and then multiplied to obtain the correlation coefficient corr1; The target echo signal amplitude in the second specific unit and the target echo signal amplitude corresponding to the range-Doppler domain index of the target in the radar B coordinate system are normalized and then multiplied to obtain the correlation coefficient corr2; The target echo signal amplitude in the first specific unit and the target echo signal amplitude in the second specific unit are normalized respectively and then multiplied to obtain a correlation coefficient corr3; According to the correlation coefficients corr1, corr2 and corr3, the objective function is set as: -min([abs(corr1),abs(corr2),abs(corr3)]) where abs represents the absolute value function.

Images