Two-dimensional doa estimation method and device based on perceptual compression and medium

Abstract

The application provides a two-dimensional DOA estimation method and device based on perception compression and a medium. After echo signals received by each array element in a uniform surface array are obtained, one-dimensional digital beam forming (DBF) operation is performed on echo signals received by array elements in the same direction, power spectrum peak value searching is performed on one-dimensional DBF results obtained in different directions respectively, an initial angle set of a target is obtained, the initial angle set is expanded by using a preset angle resolution, an angle set of the target is obtained, and the number of atoms in OMP operation is reduced. Thus, for the angle set of the target, OMP algorithm is used to reconstruct a sparse vector of adjusted one-dimensional echo signals, and at least according to the reconstructed sparse vector, the azimuth angle and the elevation angle of the target can be quickly and accurately obtained, the calculation amount is reduced, and the real-time demand of an application scene is met.

Description

TECHNICAL FIELD

[0001] The present application mainly relates to the field of array signal processing, and more particularly to a two-dimensional DOA estimation method based on compressed sensing, a device and a medium. BACKGROUND

[0002] In array signal processing applications, the direction of arrival (DOA) estimation technology is usually used to obtain the spatial position of the target, and in the fields of image fusion, ultra-wideband communication and underwater acoustic communication, high-resolution angle estimation can be performed in combination with the compressed sensing algorithm to realize the coherent source DOA estimation of a single fast shot.

[0003] In practical applications, a compressed sensing algorithm such as the orthogonal matching pursuit (OMP) algorithm is usually used for two-dimensional (2D) DOA estimation. In the operation process, the 2D signal needs to be adjusted to a 1D signal first, and then the OMP operation is performed. However, due to the large length of the 1D signal, the computational complexity of the OMP operation is greatly increased, and it is difficult to meet the real-time requirements in application scenarios. SUMMARY

[0004] To solve the above technical problems, the present application provides the following technical solutions:

[0005] The present application provides a two-dimensional DOA estimation method based on compressed sensing, which comprises:

[0006] Obtaining echo signals received by each array element in a rectangular array;

[0007] Performing a one-dimensional digital beam forming (DBF) operation on the echo signals received by the array elements in the same direction to obtain a one-dimensional DBF result corresponding to the direction;

[0008] Respectively performing power spectrum peak value searching on the one-dimensional DBF results in different directions to obtain an initial angle set of the target;

[0009] Expanding the initial angle set by using a preset angle resolution to obtain a target angle set; the target angle set contains a number of angles in the same direction, which is less than the number of angles contained in the one-dimensional DBF result corresponding to the direction;

[0010] Reconstructing a sparse vector by using an orthogonal matching pursuit (OMP) algorithm on the adjusted one-dimensional echo signal for the target angle set;

[0011] Obtaining a target azimuth angle and a target pitch angle according to at least the reconstructed sparse vector.

[0012] Optionally, the echo signals received by the array elements in the same direction are subjected to one-dimensional digital beam forming (DBF) operation to obtain one-dimensional DBF results corresponding to the direction, comprising:

[0013] Array echo signals of the array elements in the same angle of different directions and corresponding steering vectors are obtained;

[0014] The angle of the same direction is discretely sampled, and a steering matrix of the array elements in the direction is generated according to the obtained discrete angle set;

[0015] The one-dimensional DBF operation is performed according to the steering matrix and the array echo signals to obtain one-dimensional DBF results corresponding to the direction.

[0016] Optionally, the one-dimensional DBF results of different directions are subjected to power spectrum peak value search to obtain an initial angle set of the target, comprising:

[0017] The signal power of different frequency bands in the power spectrum of the one-dimensional DBF results of different directions is compared with a preset threshold value respectively;

[0018] The frequency band position with the signal power greater than the preset threshold value is determined as a peak position;

[0019] The angles corresponding to each peak position included in the one-dimensional DBF result of the same direction are determined to constitute an initial angle set of the target in the direction;

[0020] The initial angle set includes an intermediate angle initial set and a pitch angle initial set, and the intermediate angle can be synthesized from the corresponding pitch angle and azimuth angle of the target; the different directions include the azimuth direction and the pitch direction of the rectangular array.

[0021] Optionally, the initial angle set is expanded by using a preset angle resolution to obtain a target angle set, comprising:

[0022] Each intermediate angle initial value included in the intermediate angle initial set is subjected to increasing and decreasing operations respectively by using an intermediate angle resolution to obtain an intermediate angle target set;

[0023] Each pitch angle initial value included in the pitch angle initial set is subjected to increasing and decreasing operations respectively by using a pitch angle resolution to obtain a pitch angle target set;

[0024] The intermediate angle resolution is determined according to the number of array elements in the azimuth direction of the rectangular array, the interval distance of adjacent array elements, the average intermediate angle, and the wavelength of the target;

[0025] The elevation and depression angle resolution is determined according to the number of elements in the rectangular array in the elevation and depression direction, the interval distance of adjacent elements, the average elevation and depression angle, and the wavelength of the target.

[0026] Optionally, the adjusted one-dimensional echo signal is reconstructed into a sparse vector by using an orthogonal matching pursuit (OMP) algorithm for the target angle set, including:

[0027] The received two-dimensional echo signal of the rectangular array is adjusted into a one-dimensional echo signal;

[0028] The steering vector of the rectangular array corresponding to the one-dimensional echo signal is obtained by using the steering vector of each element in the intermediate angle target set and the elevation and depression angle target set;

[0029] A complete redundant dictionary for the OMP algorithm is obtained by using the steering vector of the rectangular array;

[0030] The sparse vector is reconstructed according to the OMP algorithm and the complete redundant dictionary.

[0031] Optionally, the target azimuth angle and the target elevation and depression angle are obtained according to the reconstructed sparse vector, including:

[0032] The position of the non-zero element in the reconstructed sparse vector is determined;

[0033] The target intermediate angle and the target elevation and depression angle corresponding to the steering vector corresponding to the position are obtained;

[0034] The target azimuth angle is obtained by performing operation on the target intermediate angle and the target elevation and depression angle according to the trigonometric function relationship among the azimuth angle, the elevation and depression angle, and the intermediate angle of the same incident signal.

[0035] Optionally, the echo signal received by each element in the rectangular array is obtained, including:

[0036] The number of elements in different directions in the rectangular array and the interval distance of adjacent elements are obtained;

[0037] The signal wavelength of each target incident on the rectangular array is determined;

[0038] The corresponding target parameters are obtained according to the signal wavelength, the interval distance, the unknown azimuth angle, and the elevation and depression angle of each target;

[0039] The echo signal model is obtained by at least according to the target parameters;

[0040] The echo signal received by each element in the rectangular array is obtained by using the echo signal model.

[0041] Optionally, the angle in the same direction is discretely sampled, including:

[0042] According to the signal wavelength of the target and the interval distance of adjacent elements in different directions, an unambiguous angle range in the corresponding direction is obtained;

[0043] The unambiguous angle range in different directions is uniformly discretely sampled to obtain a discrete angle set in the corresponding direction.

[0044] The application also provides a two-dimensional DOA estimation device based on perceptual compression, which comprises:

[0045] A return signal obtaining module is configured to obtain return signals received by each element in the rectangular array;

[0046] A one-dimensional DBF operation module is configured to perform one-dimensional digital beam forming (DBF) operation on the return signals received by the elements in the same direction to obtain one-dimensional DBF results in the corresponding direction;

[0047] A peak searching module is configured to perform power spectrum peak searching on the one-dimensional DBF results in different directions respectively to obtain an initial angle set of the target;

[0048] An angle expanding module is configured to expand the initial angle set by using a preset angle resolution to obtain a target angle set; the target angle set contains a smaller number of angles in the same direction than the number of angles contained in the one-dimensional DBF results in the corresponding direction;

[0049] A sparse vector reconstructing module is configured to reconstruct sparse vectors by using an orthogonal matching pursuit (OMP) algorithm on the adjusted one-dimensional return signals for the target angle set;

[0050] A target azimuth angle obtaining module is configured to obtain a target azimuth angle and a target elevation angle according to at least the reconstructed sparse vectors.

[0051] The application also provides a computer readable storage medium, which stores a computer program, and the computer program is loaded and executed by a processor to realize the two-dimensional DOA estimation method based on perceptual compression.

[0052] Therefore, the application provides a two-dimensional DOA estimation method based on perceptual compression, a device and a medium. After obtaining echo signals received by each array element in a rectangular array, one-dimensional digital beam forming (DBF) operation is performed on the echo signals received by the array elements in the same direction, power spectrum peak value searching is performed on the obtained one-dimensional DBF results in different directions respectively, an initial angle set of the target is obtained, the initial angle set is expanded by using a preset angle resolution, a target angle set is obtained, the number of atoms in OMP operation is reduced, for the target angle set, the adjusted one-dimensional echo signal is reconstructed by using an OMP algorithm, and the target azimuth angle and the target elevation angle can be quickly and accurately obtained according to the reconstructed sparse vector, the calculation amount is reduced, and the real-time requirement of the application scene is met. BRIEF DESCRIPTION OF DRAWINGS

[0053] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are only the embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of the provided drawings.

[0054] Figure 1 A flowchart of an optional example of the two-dimensional DOA estimation method based on perceptual compression provided by the present application;

[0055] Figure 2 An array structure diagram of a two-dimensional uniform planar array suitable for the two-dimensional DOA estimation method based on perceptual compression provided by the present application;

[0056] Figure 3 A flowchart of another optional example of the two-dimensional DOA estimation method based on perceptual compression provided by the present application;

[0057] Figure 4 A simulation parameter diagram for simulating and verifying the two-dimensional DOA estimation method based on perceptual compression provided by the present application;

[0058] Figure 5 A RMSE curve diagram of the two-dimensional DOA estimation method based on perceptual compression provided by the present application and a traditional two-dimensional OMP DOA estimation method;

[0059] Figure 6 A structure diagram of an optional example of the two-dimensional DOA estimation device provided by the present application;

[0060] Figure 7 A hardware structure diagram of an optional example of a computer device suitable for the two-dimensional DOA estimation method based on perceptual compression provided by the present application. DETAILED DESCRIPTION

[0061] For the technical problems described in the background section, the present application hopes to reduce the computational complexity of two-dimensional Direction Of Arrival (DOA) estimation to meet the real-time estimation requirements in application scenarios. It is proposed that the characteristics of the sparse target space of DOA estimation can be used to perform one-dimensional Digital Beam Forming (DBF) operation in the azimuth and elevation dimensions respectively, estimate the approximate range of the intermediate angle and the elevation angle of the sparse target, and then construct a complete redundant dictionary on this basis, use the OMP algorithm to reconstruct the sparse vector, and determine the intermediate angle and the elevation angle of the target. Since this processing method greatly reduces the number of atoms in the OMP operation, it greatly reduces the computational complexity under the premise of ensuring the performance of DOA estimation, and better meets the real-time estimation requirements.

[0062] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative labor are within the scope of protection of the present application.

[0063] Reference Figure 1 An optional flowchart of a two-dimensional DOA estimation method based on perceptual compression proposed in the present application is shown. The method can be executed by a computer device, which includes a terminal device with certain data calculation capability, such as a smart phone, a notebook computer, a vehicle-mounted terminal, etc., and can also include a server, such as a physical server or a cloud server, etc. The present application does not limit the product type of the computer device. As shown in Figure 1 The method can include:

[0064] Step S11, obtaining echo signals received by each element in the rectangular array;

[0065] In the embodiments of the present application, a rectangular array is taken as an example to construct an echo, which is a two-dimensional uniform planar array. Referring to the array structure diagram of the uniform planar array shown in Figure 2 The uniform planar array can be composed of MxN elements uniformly distributed in the xoz plane. The interval distance of adjacent elements in the x-axis direction (which can be referred to as the azimuth direction) is d x , and the interval distance of adjacent elements in the z-axis direction can be d z . It is assumed that there are K far-field targets (referred to as targets for short in the present application), the azimuth angle corresponding to the kth target is θ k , and the elevation angle is φ kIgnoring the effects of noise, these parameters can be used to model the signal echo of each element in a uniform array. This application does not limit the implementation method of signal echo modeling. It should be noted that this application does not limit the values ​​of the parameters listed above, and can be determined as appropriate.

[0066] Because the median angle α of the k-th target is directly obtained by the two-dimensional DOA estimation technique k and pitch angle φ k , where the intermediate angle α k It is the azimuth angle θ of the target. k and pitch angle φ k The intermediate angle variable is defined by combination, and the three variables have a sinα relationship. k =sinθ k cosφ k Therefore, after obtaining the two-dimensional DOA estimation results, this application can determine the azimuth angle of the corresponding target based on this trigonometric function relationship. The following will explain how to obtain the median angle α of each target. k and pitch angle φ k The implementation process is described.

[0067] Step S12: Perform one-dimensional digital beamforming (DBF) operation on the echo signals received by array elements in the same direction to obtain one-dimensional DBF results in the corresponding direction.

[0068] In the embodiments of this application, for ease of description, Figure 2 The x-axis direction is defined as the row of the uniform array, and the z-axis direction is defined as the column of the uniform array. Based on the sparse characteristics of the target space, one-dimensional digital beamforming (DBF) operations can be performed in the azimuth and elevation directions respectively to obtain the approximate range of the intermediate angle and elevation angle of the sparse target. This application can be represented by an angle set. The implementation process of the one-dimensional DBF operation in the x-axis direction and the one-dimensional DBF operation in the z-axis direction is not described in detail in this embodiment.

[0069] To facilitate computer processing, the one-dimensional DBF operation described above can be performed by uniformly discrete sampling at angles in different directions to obtain a finite set of angles in the corresponding directions. For example, the offline intermediate angle set A = {α1, α2, α3, ..., α...} can be obtained for the azimuth direction. P}, where P represents the total number of discrete samples of the intermediate angle; similarly, the offline elevation angle set in the elevation direction can be obtained as Φ={φ1,φ2,φ3,……,φ Q}, where Q represents the total number of discrete pitch angles. This application does not impose any restrictions on the values ​​of P and Q.

[0070] Step S13: Perform power spectrum peak search on the one-dimensional DBF results in different directions to obtain the initial angle set of the target;

[0071] In order to reduce the number of atoms of the two-dimensional OMP operation, the application can perform dimension reduction processing on the one-dimensional DBF results in different directions, select the angles corresponding to the signal power peaks on the power spectrum, and complete the subsequent two-dimensional DOA estimation. Therefore, the power spectrum of the one-dimensional DBF result (denoted as DBF αz (α p ), p = 1, 2, 3,..., P) in the x-axis direction and the one-dimensional DBF result (denoted as DBF el (α q ), q = 1, 2, 3,..., Q) in the z-axis direction are normalized, and then the signal power peak search is performed, such as searching for all peaks with a normalized signal power of-12dB or more.

[0072] For example, according to the above method, K1 (the value can be determined according to the situation) peaks are searched in the x-axis direction, and the middle angle set U corresponding to the positions of the peaks can be denoted as: U = {u k1 ∈ A | k1 = 1, 2, 3,..., K1}, u k1 may represent the middle angle corresponding to the position of the k1th peak; K2 (the value can be determined according to the situation) peaks are searched in the z-axis direction, and the middle angle set W corresponding to the positions of the peaks can be denoted as: W = {w k2 ∈ Φ | k2 = 1, 2, 3,..., K2}, w k2 may represent the pitch angle corresponding to the position of the k2th peak.

[0073] It can be seen that after the peak search in the x-axis direction and the z-axis direction, the number of angles K1 contained in the initial middle angle set U is much larger than the number of angles P contained in the above discrete middle angle set A; the number of angles K2 contained in the initial pitch angle set W is much larger than the number of angles Q contained in the above discrete pitch angle set Φ, which reduces the amount of calculation to a certain extent.

[0074] Step S14: expanding the initial angle set by using a preset angle resolution to obtain a target angle set;

[0075] In actual application, since the DBF does not have super-resolution capability, when the difference between the middle angles of two targets or the difference between the pitch angles of two targets is less than the corresponding resolution, the one-dimensional DBF result cannot distinguish the two targets, and in this case, the initial angle set searched by the peak search on the one-dimensional DBF result is inaccurate. Therefore, the application proposes to expand each angle in the initial angle set U and W by one resolution unit on the left and right, and then perform the subsequent orthogonal matching pursuit OMP operation according to the expansion, so as to ensure that the two-dimensional DOA estimation result is accurate and reliable. The expansion implementation method of the initial angle set is not described in detail.

[0076] It should be understood that even if the initial angle sets U and W are angle-expanded according to the method described above, so that each element in the set can be an interval range value, the number of elements is still much smaller than the number of angles of the discrete angle sets A and Φ, that is, the number of angles in the same direction contained in the target angle set is less than the number of angles contained in the one-dimensional DBF result of the corresponding direction, and the subsequent complete redundant dictionary of the two-dimensional OMP is generated according to this, which can greatly reduce the number of atoms of the dictionary in the two-dimensional OMP, and further greatly reduce the operation amount when the OMP algorithm is applied.

[0077] Step S15, for the target angle set, using the OMP algorithm to perform sparse vector reconstruction on the adjusted one-dimensional echo signal;

[0078] Step S16, obtaining the target azimuth angle and the target elevation angle according to at least the reconstructed sparse vector.

[0079] Before performing the OMP operation, the two-dimensional echo signal of each array element can be adjusted (reshaped) to obtain a one-dimensional echo signal, and then the OMP algorithm is applied to reconstruct the sparse vector to obtain the target intermediate angle and the target elevation angle corresponding to each target. Then, the sinα k = sinθ k cosφ k This trigonometric function relationship obtains the target azimuth angle corresponding to each target.

[0080] In combination with the above analysis, the number of atoms of the complete redundant dictionary in the original two-dimensional OMP DOA is reduced from PQ to K3K4 (that is, the number of angles contained in the target angle set described above), which greatly reduces the operation amount, thereby reducing the requirement for the hardware of the computer device, improving the calculation efficiency, and better meeting the real-time demand in the application scenario.

[0081] Referring to Figure 3 The flowchart of another optional example of the two-dimensional DOA estimation method based on perceptual compression proposed in the present application can describe an optional detailed implementation process of the two-dimensional DOA estimation method based on perceptual compression proposed in the above, as shown in Figure 3 The method can include the following steps:

[0082] Step S31, modeling the signal echo of each array element in the rectangular array to obtain an echo signal model;

[0083] Step S32, obtaining the echo signal received by each array element in the rectangular array for the target by using the echo signal model;

[0084] Still referring to the above description, as Figure 2The two-dimensional uniform planar array (rectangular array) shown is taken as an example to illustrate the echo modeling, and in combination with the above description of the array structure of the two-dimensional uniform planar array, the number of array elements in different directions (such as the azimuth direction and the elevation direction, i.e., the x-axis direction and the z-axis direction) and the interval distance d of adjacent array elements in the uniform planar array are obtained x and d z After determining the signal wavelength λ of each target incident on the uniform planar array, the corresponding target parameters can be obtained according to the signal wavelength, the interval distance, and the unknown azimuth angle and the elevation angle of each target, and then the signal echo modeling can be performed at least according to the target parameters to obtain an echo signal model S mn , as shown in the following formula:

[0085]

[0086] In the above formula (1), 1≤m≤M, 1≤n≤N, m and n are positive integers. exp() represents the exponential function with the natural constant e as the base. Since the intermediate angle α k , the azimuth angle θ k and the elevation angle φ k of the same target have the relationship sinα k =sinθ k cosφ k , which is substituted into the above formula (1), to obtain:

[0087]

[0088] In combination with the idea of the two-dimensional DOA estimation technology, the echo signal model shown in formula (2) can be used to obtain the echo signals received by each array element in the uniform planar array, i.e., to obtain the echo signal expression received by each array element, and the intermediate angle α k and the elevation angle φ k therein are calculated according to the method described below.

[0089] Step S33, obtaining the array echo signals of each array element at the same angle in different directions and the corresponding steering vectors;

[0090] In combination with the above description of the uniform planar array, the echo signals received by any row of array elements and any column of array elements can be selected for one-dimensional DBF operation in the x-axis direction and the z-axis direction, respectively. In order to explain the operation process, the echo signals of the first row of array elements and the first column of array elements are taken as an example to illustrate the one-dimensional DBF operation, and the one-dimensional DBF operation implementation process for the echo signals of the array elements of other rows and other columns is similar, which will not be described one by one.

[0091] Wherein, in combination with the above description of the echo signal model, in the process of one-dimensional DBF operation in the x-axis direction, the echo signal of the first row of elements can be obtained, which can be represented as:

[0092]

[0093] The steering vector v az (α k ) corresponding to the first row of elements can also be obtained, which can be represented as:

[0094]

[0095] Similarly, in the process of one-dimensional DBF operation in the z-axis direction, the echo signal of the first column of elements can be obtained, which can be represented as:

[0096]

[0097] Correspondingly, the steering vector v el (φ k ) corresponding to the first column of elements can also be obtained, which can be represented as:

[0098]

[0099] The calculation process of the steering vector of the above-mentioned elements will not be described in detail.

[0100] Step S34: Discretely sampling the angle in the same direction, and generating the steering matrix of each element in the direction according to the obtained discrete angle set;

[0101] In order to facilitate the operation of the computer device, the present application can uniformly discretely sample the intermediate angle and the pitch angle respectively. Assuming that the total number of discrete intermediate angles P and the total number of discrete pitch angles Q are both even numbers, the discrete intermediate angle α p and the discrete pitch angle φ q obtained by sampling can be represented as:

[0102]

[0103]

[0104] Wherein, in the above formula (7), α max = arcsin (λ / 2d x ) can be defined as the maximum unambiguous angle in the x-axis direction. In this way, the unambiguous angle range in the x-axis direction can be [-α max , α maxIf so, the set corresponding to the discrete intermediate angles, i.e., the discrete intermediate angle set, can be represented as A = {α1, α2, α3, …, α P}。 In formula (8), φ max = arcsin (λ / 2d z ) is defined as the maximum unambiguous angle in the z-axis direction, so the unambiguous angle range in the z-axis direction can be [-φ max , φ max ], and then the set corresponding to the discrete elevation angles, i.e., the discrete elevation angle set, can be represented as Φ = {φ1, φ2, φ3, …, φ Q}.

[0105] It can be seen that the present application can obtain the unambiguous angle range in the corresponding direction according to the signal wavelength of the target and the interval distance of adjacent elements in different directions, and then uniformly discretely sample the unambiguous angle range in different directions to obtain the discrete angle set in the corresponding direction, such as the discrete intermediate angle set and the discrete elevation angle set described above, but is not limited to the discretely sampling implementation method described above.

[0106] Then, based on the angles contained in the discrete angle set and in combination with formula (4) and formula (6), the steering matrix V αz and V el of the corresponding direction element can be obtained, where:

[0107] V αz = [v αz (α1), v αz (α2), …, v αz (α P )] M×P (9)

[0108] V el = [v el (φ1), v el (φ2), …, v el (φ q )] N×Q (10)

[0109] Step S35, performing one-dimensional DBF operation according to the steering matrix and the array echo signal to obtain one-dimensional DBF result in the corresponding direction;

[0110] Continuing the above analysis, the present application can use the steering array to perform one-dimensional DBF operation to obtain one-dimensional DBF result in the x-axis direction, denoted as DBF αz (α p ), and one-dimensional DBF result in the z-axis direction, denoted as DBF el (α q ), where:

[0111] DBF αz (α p )=V αz H S αz (11)

[0112] DBF el (α q )=V el H S el (12)

[0113] Step S36, compare the signal power of different frequency bands in the power spectrum of the one-dimensional DBF result of different directions respectively with a preset threshold value;

[0114] Step S37, determine the frequency band position with signal power greater than the preset threshold value as the peak position;

[0115] Step S38, determine the angle corresponding to each peak position contained in the one-dimensional DBF result of the same direction, to constitute the initial angle set of the target in the direction;

[0116] In combination with the above description of peak search, the present application can perform peak search on the one-dimensional DBF result DBF αz (α p ) of the x-axis direction, and the one-dimensional DBF result DBF el (α q ) of the z-axis direction respectively, as described in the above steps, assuming that K1 peaks are found in the x-axis direction, to obtain the intermediate angle initial set U={u k1 ∈A|k1=1,2,3,……K1} corresponding to the peak positions; K2 peaks are found in the z-axis direction, to obtain the pitch angle initial set W={w k2 ∈Φ|k2=1,2,3,……K2} corresponding to the peak positions.

[0117] It can be seen that the initial angle set constituted by step S38 can include the intermediate angle initial set U and the pitch angle initial set W, as analyzed above, the intermediate angle can be synthesized by the corresponding pitch angle and azimuth angle of the target; different directions include the azimuth direction and the pitch direction representing the uniform planar array.

[0118] Step S39, increase and decrease operations are respectively performed on each intermediate angle initial value contained in the intermediate angle initial set by using the intermediate angle resolution, to obtain the intermediate angle target set;

[0119] Step S310, increase and decrease operations are respectively performed on each pitch angle initial value contained in the pitch angle initial set by using the pitch angle resolution, to obtain the pitch angle target set;

[0120] In order to improve the DBF angle measurement accuracy and reliability, the initial angle set obtained by peak value search can be expanded. According to the above method, the intermediate angle resolution can be determined according to the number of elements in the azimuth direction of the uniform planar array, the interval distance of adjacent elements, the average intermediate angle, and the wavelength of the target; the elevation angle resolution can be determined according to the number of elements in the elevation direction of the uniform planar array, the interval distance of adjacent elements, the average elevation angle, and the wavelength of the target. Then, the two resolutions can be used to expand the elements in the corresponding initial angle set.

[0121] Therefore, the initial intermediate angle values contained in the initial intermediate angle set U are expanded, and the target intermediate angle set D={d∈A||d-u1|≤α res ,|d-u2|≤α res ,……,|d-u k1 |≤α res} can be obtained; similarly, the initial elevation angle values contained in the initial elevation angle set W are expanded, and the target intermediate angle set E={e∈Φ||e-w1|≤φ res ,|e-w2|≤φ res ,……,|e-w k2 |≤φ res} can be obtained, wherein the intermediate angle resolution α res =λ / Md x cos(α), the elevation angle resolution α represents the value of the intermediate angle where the target is located, φ represents the value of the elevation angle where the target is located, and the angle resolution of the rectangular array is related to the angle where the target is located. When the angle where the target is located is equal to zero, the rectangular array can obtain the highest angle resolution.

[0122] In order to facilitate subsequent description and analysis of technical effects, it is assumed that the total number of elements in the above set D is K3; the total number of elements in the set E is K4, so the above set D and set E can be simplified as: D={d1,d2,d3,……,d k3}, E={e1,e2,e3,……,e k4}.

[0123] Step S311, adjusting the received two-dimensional echo signal of the rectangular array to a one-dimensional echo signal;

[0124] Step S312, obtaining the steering vector of the rectangular array corresponding to the one-dimensional echo signal by using the steering vectors of the elements in the target intermediate angle set and the target elevation angle set at each angle;

[0125] Since the targets are sparse, the total number of elements K3 in set D and the total number of elements K4 in set E are much smaller than the number of elements in set A and set Φ, and the number of atoms in the dictionary of two-dimensional OMP can be greatly reduced by replacing set A with set D and replacing Φ with E to generate a complete redundant dictionary of two-dimensional OMP.

[0126] Before performing the OMP algorithm, the uniform planar array two-dimensional echo signal can be arranged into a one-dimensional echo signal S all , that is,

[0127]

[0128] Based on this, the one-dimensional echo signal S all The steering vector corresponding to the array can be expressed as:

[0129] v all (α k ,φ k )=[kron(v el (φ k ),vα z (α k ))] MN×1 (14)

[0130] In the above formula (14), kron() can represent the Kronecker product.

[0131] In step S313, a complete redundant dictionary for the orthogonal matching pursuit (OMP) algorithm is obtained by using the steering vector of the rectangular array.

[0132] In the embodiments of the present application, a compressed sensing model can be constructed according to the one-dimensional echo signal of the uniform planar array and the steering vector, and the model construction process is not limited in the present application. Wherein, the steering matrix of the uniform planar array under the middle angle target set D and the pitch angle target set E is obtained according to the steering vector v all (α k ,φ k ).

[0133] V all =[v all (d1,e1),v all (d2,e1),…,v all (d K3 ,e1),

[0134] v all (d1,e2),v all (d2,e2),…v all (d K3 ,e2),

[0135] …,

[0136] v all (d1,e K4 ),v all (d2,e K4 ),…,v all (d K3 ,e K4 )] MN×K3K4 (15)

[0137] The above steering matrix V all is denoted as a complete redundant dictionary in the OMP operation, assuming S all represents an observation signal, x is a corresponding sparse vector (the length of x can be K3K4, wherein the position of the non-zero element corresponds to the corresponding angle of the steering vector, which represents the intermediate angle and the pitch angle of the incident signal, and the amplitude of the non-zero element is the amplitude of the incident signal), and the sparsity of x is the total number K of templates. On the basis of the above, a sparse linear model, i.e., a compressed sensing model, can be constructed:

[0138] S all =V all x (16)

[0139] In step S314, the sparse vector is reconstructed according to the OMP algorithm and the complete redundant dictionary.

[0140] In step S315, the target azimuth angle and the target pitch angle are obtained according to at least the reconstructed sparse vector.

[0141] In combination with the above description of the sparse vector, the position of the non-zero element in the reconstructed sparse vector can be determined, the target intermediate angle and the target pitch angle corresponding to the position of the steering vector are obtained, and then the target azimuth angle can be obtained by performing operations on the target intermediate angle and the target pitch angle according to the trigonometric relationship among the azimuth angle, the pitch angle and the intermediate angle of the same incident signal, as described above.

[0142] The two-dimensional DOA estimation method based on perception compression described in the above embodiment is simulated and compared with the traditional two-dimensional OMP DOA estimation method under the simulation parameters as shown in FIG. 6, to verify the effectiveness of the present application. The effectiveness can be evaluated by the root mean squares error (RMSE) under different signal noise ratios (SNR), and the RMSE evaluation formula is as follows: Figure 4

[0143]

[0144] In the above formula (17), N mc ​​N can represent the number of Monte Carlo simulations, and N = 5000 in the simulation mc K can represent the number of targets, and K = 3 in the simulation, and the angles of the targets can be randomly generated in each Monte Carlo simulation.

[0145] Thus, for a given SNR, the RMSE curves of the two-dimensional DOA estimation method of the perceptual compression and the traditional two-dimensional OMP DOA estimation method are as shown in Figure 5 Under 5000 times of Monte Carlo simulation, the average time consumed by the two two-dimensional DOA estimation methods is 0.0244s and 0.3934s respectively. It can be seen that the two-dimensional DOA estimation method of the perceptual compression greatly reduces the running time of the two-dimensional OMP algorithm and improves the operation efficiency.

[0146] Referring to Figure 6 , a structure schematic diagram of an optional example of the two-dimensional DOA estimation device based on perceptual compression is shown, which can include:

[0147] The echo signal obtaining module 61 is configured to obtain echo signals received by each array element in the rectangular array;

[0148] The one-dimensional DBF operation module 62 is configured to perform one-dimensional digital beam forming (DBF) operation on the echo signals received by the array elements in the same direction to obtain one-dimensional DBF results corresponding to the direction;

[0149] The peak search module 63 is configured to perform power spectrum peak search on the one-dimensional DBF results in different directions respectively to obtain an initial angle set of the target;

[0150] The angle expansion module 64 is configured to expand the initial angle set by using a preset angle resolution to obtain a target angle set; the target angle set contains a number of angles in the same direction, which is less than the number of angles contained in the one-dimensional DBF result corresponding to the direction;

[0151] The sparse vector reconstruction module 65 is configured to reconstruct a sparse vector by using an orthogonal matching pursuit (OMP) algorithm on the adjusted one-dimensional echo signal according to the target angle set;

[0152] The target azimuth angle obtaining module 66 is configured to obtain a target azimuth angle and a target elevation angle according to at least the reconstructed sparse vector.

[0153] Optionally, the echo signal obtaining module 61 can include:

[0154] The interval distance obtaining unit is configured to obtain the number of array elements in different directions in the rectangular array and the interval distance between adjacent array elements;

[0155] a signal wavelength determination unit, configured to determine a signal wavelength of each target incident on the rectangular array;

[0156] a target parameter obtaining unit, configured to obtain a corresponding target parameter according to the signal wavelength, the interval distance, and unknown azimuth and elevation angles of each target;

[0157] a modeling unit, configured to model a signal echo according to at least the target parameter, to obtain an echo signal model;

[0158] an echo signal obtaining unit, configured to obtain echo signals received by each array element in the rectangular array by using the echo signal model.

[0159] Optionally, the one-dimensional DBF operation module 62 can include:

[0160] a first obtaining unit, configured to obtain array echo signals of each array element at the same angle in different directions and corresponding steering vectors;

[0161] a steering matrix generating unit, configured to discretely sample the angle in the same direction, and generate a steering matrix of each array element in the direction according to a discrete angle set obtained;

[0162] the one-dimensional DBF operation unit is configured to perform one-dimensional digital beam forming (DBF) operation according to the steering matrix and the array echo signals, to obtain a one-dimensional DBF result of the corresponding direction.

[0163] Optionally, the steering matrix generating unit can include:

[0164] a non-ambiguous angle range obtaining unit, configured to obtain a non-ambiguous angle range in the corresponding direction according to the signal wavelength of the target and the interval distance of adjacent array elements in different directions;

[0165] a discrete angle set obtaining unit, configured to uniformly discretely sample the non-ambiguous angle range in different directions, to obtain a discrete angle set in the corresponding direction.

[0166] Optionally, the peak searching module 63 can include:

[0167] a comparing unit, configured to compare signal powers of different frequency bands in a power spectrum of the one-dimensional DBF result of different directions with a preset threshold respectively;

[0168] a peak position determining unit, configured to determine a frequency band position with the signal power greater than the preset threshold as a peak position;

[0169] The initial angle set construction unit is configured to determine angles corresponding to each peak position included in the one-dimensional DBF result of the same direction, and construct an initial angle set of the target in the direction;

[0170] The initial angle set includes an intermediate angle initial set and a pitch angle initial set, and the intermediate angle can be synthesized from the corresponding pitch angle and azimuth angle of the target; and the different directions include an azimuth direction and a pitch direction of the uniform planar array.

[0171] Optionally, the angle extension module 64 can include:

[0172] The intermediate angle target set obtaining unit is configured to increase and decrease each intermediate angle initial value included in the intermediate angle initial set respectively by using an intermediate angle resolution, and obtain an intermediate angle target set;

[0173] The pitch angle target set obtaining unit is configured to increase and decrease each pitch angle initial value included in the pitch angle initial set respectively by using a pitch angle resolution, and obtain a pitch angle target set;

[0174] The intermediate angle resolution is determined according to the number of elements in the azimuth direction of the rectangular array, the interval distance of adjacent elements, the average intermediate angle, and the wavelength of the target;

[0175] The pitch angle resolution is determined according to the number of elements in the pitch direction of the rectangular array, the interval distance of adjacent elements, the average pitch angle, and the wavelength of the target.

[0176] Optionally, the sparse vector reconstruction module 65 can include:

[0177] The adjustment unit is configured to adjust the received two-dimensional echo signal of the rectangular array into a one-dimensional echo signal;

[0178] The steering vector obtaining unit is configured to obtain the steering vector of the rectangular array corresponding to the one-dimensional echo signal by using the steering vector of each angle in the intermediate angle target set and the pitch angle target set;

[0179] The complete redundant dictionary obtaining unit is configured to obtain a complete redundant dictionary for an orthogonal matching pursuit (OMP) algorithm by using the steering vector of the rectangular array;

[0180] The reconstruction unit is configured to reconstruct a sparse vector according to the OMP algorithm and the complete redundant dictionary.

[0181] Optionally, the target azimuth angle obtaining module 66 can include:

[0182] The position determination unit is configured to determine the position of a non-zero element in the reconstructed sparse vector;

[0183] an object angle obtaining unit, configured to obtain a target intermediate angle and a target pitch angle corresponding to the position corresponding steering vector;

[0184] an object azimuth angle obtaining unit, configured to perform operation on the target intermediate angle and the target pitch angle according to a trigonometric function relationship among the azimuth angle, the pitch angle and the intermediate angle of the same incident signal, to obtain a target azimuth angle.

[0185] It should be noted that, as to various modules, units and the like in the above-mentioned device embodiments, they can be stored in the memory as program modules, and the processor executes the above-mentioned program modules stored in the memory to realize the corresponding functions. As to the functions realized by each program module and its combination, and the technical effects achieved, reference can be made to the description of the corresponding part of the above-mentioned method embodiments, and the present embodiment will not be described herein.

[0186] The present application also provides a computer readable storage medium, which can store a computer program, the computer program can be called and loaded by a processor to realize each step of the above-mentioned two-dimensional DOA estimation method based on perceptual compression.

[0187] Reference Figure 7 For the hardware structure schematic diagram of an optional example of the computer device suitable for the two-dimensional DOA estimation method based on perceptual compression proposed in the present application, as Figure 7 shown, the computer device can include at least one memory 71, at least one processor 72 and at least one communication module 73, wherein:

[0188] The memory 71 can be used to store the program of the above-mentioned two-dimensional DOA estimation method based on perceptual compression; the processor 72 can load and execute the program stored in the memory to realize each step of the above-mentioned two-dimensional DOA estimation method based on perceptual compression. The device type of the memory 71 and the processor 72 is not limited in the present application, and can be determined as appropriate.

[0189] The above-mentioned communication module 73 can include but is not limited to a communication module for realizing data interaction by using a wireless communication network, such as a WIFI module, a 5G / 6G (fifth generation mobile communication network / sixth generation mobile communication network) module, a GPRS module, an antenna, etc., to realize data communication between the computer device and other devices.

[0190] It should be understood that the structure of the computer device described in the above embodiments does not constitute a limitation on the computer device in the embodiments of the present application, and in actual application, the computer device can include more components than those described above, or some components can be combined, such as a power supply, various sensor modules, etc., which are not enumerated herein.

[0191] Finally, it is to be understood that the phraseology or terminology employed herein, such as "first," "second," "upwardly," "downwardly," and the like, is for the purpose of common and consistent use only and

[0192] The various embodiments in the specification are described with reference to progressive or parallel numberings. Each embodiment focuses on the differences from other embodiments, and the same or similar parts among the embodiments can be referred to each other. For the apparatuses and computer devices disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, they are described simply. The relevant parts can be referred to the description of the methods.

[0193] The above description of disclosed embodiments provides enabling disclosure sufficient for one of ordinary skill in the art to practice the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein can be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for 2D DOA estimation based on perceptual compression, characterized in that, The method comprises: obtaining echo signals received by each element in a rectangular array; performing one-dimensional digital beam forming (DBF) operation on the echo signals received by elements in the same direction: obtaining array echo signals of each element at the same angle in different directions and corresponding steering vectors; discretely sampling angles in the same direction, generating a steering matrix of each element in the direction according to the obtained discrete angle set; performing one-dimensional digital beam forming (DBF) operation according to the steering matrix and the array echo signals to obtain one-dimensional DBF results corresponding to the direction; respectively searching for power spectrum peaks of one-dimensional DBF results in different directions: comparing signal powers of different frequency bands in the power spectrum of one-dimensional DBF results in different directions with a preset threshold; determining the frequency band position with the signal power greater than the preset threshold as a peak position; determining the angles corresponding to each peak position included in one-dimensional DBF results in the same direction to form an initial angle set of the target in the direction; wherein the initial angle set includes an intermediate angle initial set and a pitch angle initial set, and the intermediate angle can be synthesized from the corresponding pitch angle and azimuth angle of the target; the different directions include the azimuth direction and the pitch direction of the rectangular array; using a preset angle resolution to expand the initial angle set to obtain a target angle set; the number of angles in the same direction included in the target angle set is less than the number of angles included in the one-dimensional DBF result corresponding to the direction; using an orthogonal matching pursuit (OMP) algorithm to reconstruct a sparse vector from the adjusted one-dimensional echo signal for the target angle set; obtaining a target azimuth angle and a target pitch angle according to at least the reconstructed sparse vector; wherein the using a preset angle resolution to expand the initial angle set to obtain a target angle set comprises: using an intermediate angle resolution to respectively increase and decrease each intermediate angle initial value included in the intermediate angle initial set to obtain an intermediate angle target set; the intermediate angle resolution is determined according to the number of elements in the azimuth direction of the rectangular array, the interval distance between adjacent elements, the average intermediate angle, and the wavelength of the target; using a pitch angle resolution to respectively increase and decrease each pitch angle initial value included in the pitch angle initial set to obtain a pitch angle target set; the pitch angle resolution is determined according to the number of elements in the pitch direction of the rectangular array, the interval distance between adjacent elements, the average pitch angle, and the wavelength of the target.

2. The method of claim 1, wherein, The using an orthogonal matching pursuit (OMP) algorithm to reconstruct a sparse vector from the adjusted one-dimensional echo signal for the target angle set comprises: adjusting the received two-dimensional echo signals of the rectangular array into one-dimensional echo signals; obtaining steering vectors of the rectangular array corresponding to the one-dimensional echo signals using the steering vectors of elements at each angle in the intermediate angle target set and the pitch angle target set; obtaining a complete redundant dictionary for the orthogonal matching pursuit (OMP) algorithm using the steering vectors of the rectangular array; reconstructing a sparse vector according to the OMP algorithm and the complete redundant dictionary.

3. The method of claim 1, wherein, The target azimuth angle and the target elevation angle are obtained according to the reconstructed sparse vector, and the method comprises the following steps: determining the position of the non-zero element in the reconstructed sparse vector; obtaining the target intermediate angle and the target elevation angle corresponding to the steering vector corresponding to the position; performing operation on the target intermediate angle and the target elevation angle according to the trigonometric function relationship among the azimuth angle, the elevation angle and the intermediate angle of the same incident signal to obtain the target azimuth angle.

4. The method of claim 1, wherein, The method comprises the following steps: obtaining the number of array elements in different directions of the rectangular array and the interval distance of adjacent array elements; determining the signal wavelength of each target incident on the rectangular array; obtaining the target parameter corresponding to each target according to the signal wavelength, the interval distance and the unknown azimuth angle and elevation angle of each target; performing signal echo modeling according to the target parameter to obtain an echo signal model; obtaining the echo signal received by each array element in the rectangular array by using the echo signal model.

5. The method of claim 1, wherein, The method comprises the following steps: obtaining the non-ambiguous angle range corresponding to each direction according to the signal wavelength of the target and the interval distance of adjacent array elements in different directions; performing uniform discrete sampling on the non-ambiguous angle range in different directions to obtain a discrete angle set corresponding to each direction.

6. A two-dimensional DOA estimation apparatus based on perceptual compression, characterized in that, The device comprises: an echo signal obtaining module, configured to obtain the echo signal received by each array element in the rectangular array; a one-dimensional DBF operation module, configured to perform one-dimensional digital beam forming (DBF) operation on the echo signal received by the array element in the same direction to obtain a one-dimensional DBF result corresponding to the direction; a peak searching module, configured to perform power spectrum peak searching on the one-dimensional DBF result in different directions respectively to obtain an initial angle set of the target; an angle expanding module, configured to expand the initial angle set by using a preset angle resolution to obtain a target angle set; the target angle set contains a smaller number of angles in the same direction than the number of angles contained in the one-dimensional DBF result corresponding to the direction; a sparse vector reconstruction module, configured to reconstruct a sparse vector by using an orthogonal matching pursuit (OMP) algorithm for the adjusted one-dimensional echo signal according to the target angle set; a target azimuth angle obtaining module, configured to obtain a target azimuth angle and a target elevation angle according to the reconstructed sparse vector; The one-dimensional DBF operation module comprises: a first obtaining unit, configured to obtain the array echo signal of each array element at the same angle in different directions and the corresponding steering vector; a steering matrix generating unit, configured to perform discrete sampling on the angle in the same direction, and generate a steering matrix of the array element in the direction according to the obtained discrete angle set; the one-dimensional DBF operation unit, configured to perform one-dimensional digital beam forming (DBF) operation according to the steering matrix and the array echo signal to obtain a one-dimensional DBF result corresponding to the direction; the peak searching module comprises: a comparison unit, configured to compare the signal power of different frequency bands in the power spectrum of the one-dimensional DBF result in different directions with a preset threshold respectively; a peak position determination unit configured to determine a frequency band position, at which the signal power is greater than the preset threshold, as a peak position; an initial angle set construction unit configured to determine angles corresponding to each of the peak positions included in a one-dimensional DBF result of a same direction, and to construct an initial angle set of the target in the direction; wherein the initial angle set includes an intermediate angle initial set and a pitch angle initial set, and the intermediate angle can be synthesized from a corresponding pitch angle and an azimuth angle of the target; and the different directions include an azimuth direction and a pitch direction representing a uniform planar array; the angle extension module includes: an intermediate angle target set obtaining unit configured to increase and decrease each intermediate angle initial value included in the intermediate angle initial set respectively by using an intermediate angle resolution, and to obtain an intermediate angle target set; the intermediate angle resolution is determined according to a number of array elements in the azimuth direction of the rectangular array, a spacing distance between adjacent array elements, an average intermediate angle, and a wavelength of the target; a pitch angle target set obtaining unit configured to increase and decrease each pitch angle initial value included in the pitch angle initial set respectively by using a pitch angle resolution, and to obtain a pitch angle target set; the pitch angle resolution is determined according to a number of array elements in the pitch direction of the rectangular array, a spacing distance between adjacent array elements, an average pitch angle, and a wavelength of the target.

7. A computer-readable storage medium, characterized in that, a computer program is stored thereon, and the computer program is loaded and executed by a processor to implement the two-dimensional DOA estimation method based on perception compression according to any one of claims 1-5.

Images