Real-valued super-resolution direction-of-arrival estimation method for high-order acoustic field sensor array

Abstract

The application relates to a real-valued super-resolution direction estimation method of a high-order acoustic field sensor array, which uses a received signal covariance matrix to approximately calculate an estimated value of noise power, avoids the iterative operation process of noise power in the prior art, improves the calculation efficiency, and reduces the floating-point operation amount. By constructing an augmented matrix, the array receiving data matrix and the array manifold matrix with a multi-dimensional structure of an element are made into Hermitian matrices, so that the unitary transformation processing of the related parameters of the high-order acoustic field sensor array is realized. On the basis of obtaining the array receiving data matrix and the array manifold matrix in the real number domain, the array receiving data matrix and the array manifold matrix are applied to a sparse approximate minimum variance method with a variable exponential factor, and the completely real-valued sparse approximate minimum variance direction estimation with a variable exponential factor is realized.

Description

Technical Field

[0001] This invention relates to the field of signal processing technology, and specifically to a real-valued super-resolution azimuth estimation method for a high-order acoustic field sensor array. Background Technology

[0002] In related technologies, sparse signal processing-based orientation estimation algorithms are DOA (Direction of Arrival) estimation algorithms based on sparse signal models that have been developed in the last decade. Sparse orientation estimation methods discretize the spatial scan orientation and assume that the signal is distributed only at a finite number of scan orientation positions, with zero signal parameters at scan orientations where no signal is detected. Therefore, the target spatial orientation distribution model possesses sparsity, and utilizing the sparse information of the signal can improve DOA estimation performance. Among these, non-regular parameter algorithms such as SPICE (Simulation Program with Integrated Circuit Emphasis) and SAMV (Sparse Asymptotic Minimum Variance) provide signal parameter estimation criteria from the perspective of maximum likelihood estimation, using the relationship between the sampling covariance and the desired signal model covariance to obtain the signal power spectrum estimate at the scan grid points. These algorithms do not require any regularization parameters.

[0003] Traditional sparse orientation estimation methods all involve complex number operations. In practical engineering applications, these operations require separating the real and imaginary parts, significantly increasing the computational burden of orientation estimation. Unitary transform can convert complex-domain signals to the real-domain, drastically reducing the computational load while preserving the orientation information of the incident signal. However, unitary transform requires the transformed matrix to be a Hermitian matrix. Therefore, existing unitary transform-based real-array orientation estimation methods are only applicable to uniform linear arrays. For high-order acoustic field sensor arrays or vector sensor arrays, since the array elements themselves have multi-channel signal outputs and do not possess the characteristics of a Hermitian matrix, existing unitary transform techniques cannot be directly applied.

[0004] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of the present invention, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] This invention provides a real-valued super-resolution azimuth estimation method for a high-order sound field sensor array, a computer program product, and an electronic device, which can effectively overcome the defects existing in the prior art.

[0006] Other features and advantages of the invention will become apparent from the following detailed description, or may be learned in part by practice of the invention.

[0007] According to a first aspect of the present invention, a real-valued super-resolution azimuth estimation method for a high-order sound field sensor array is provided, the method comprising:

[0008] A high-order sound field sensor array is used to receive spatial sound field signals, and the response vectors of each high-order sound field sensor are used to... Constructing an array manifold matrix ; and based on array manifold matrix Determine the array received signal ; and, based on array received signals Define the array received signal data matrix Y for multi-shot scenarios;

[0009] Determine the covariance matrix of the array received signal data matrix, and estimate the noise power by combining the number of array elements of the high-order sound field sensor array and the number of modal signal channels output by the high-order sound field sensor, and configure the upper bound of the noise power.

[0010] Construct the augmented matrix Y corresponding to the array received signal data matrix Y. aug ; and for the augmented matrix Y aug Perform a unitary transform to obtain the array received signal data matrix in the real number field. ; Array received signal data matrix in the real number field Calculate the signal covariance matrix in the real domain , and configured as the initial value of the array received signal data covariance matrix;

[0011] Constructing an array manifold matrix The corresponding augmented array manifold matrix A aug (Θ Ω ); and for the augmented array manifold matrix A aug (Θ Ω Perform a unitary transformation to obtain the array manifold matrix A in the real number field. r (Θ Ω Based on the real-field array manifold matrix A r (Θ Ω Obtain the initial value of the spatial orientation spectrum, and configure the initial value of sparsity based on the initial value of the spatial orientation spectrum;

[0012] The sparse high-resolution azimuth estimation method is used to iterate until a preset threshold is met based on the initial values ​​of the spatial azimuth spectrum and the initial values ​​of the covariance matrix of the array received signal data. The azimuth estimation result is determined according to the sparsity of the azimuth spectrum calculated in the j-th iteration; where j is a positive integer.

[0013] In some exemplary embodiments, the high-order sound field sensor array includes a linear array constructed based on N high-order sound field sensors; wherein, the high-order sound field sensor includes M sound pressure sensors, which are uniformly distributed to form a small-scale circular array, and the diameter of the aperture of the circular array is d; wherein, N and M are positive integers.

[0014] In some exemplary embodiments, based on the response vectors of each higher-order sound field sensor Constructing an array manifold matrix ,include:

[0015] Based on the sound pressure response vectors received by each high-order sound field sensor and based on N max The modal response vectors of the first-order modal signals are subjected to Kronecker product operation to obtain the response vectors corresponding to each higher-order sound field sensor. ;in, N max ≤M / 2;

[0016] Based on the response vectors of each high-order sound field sensor Combined with the set of signal incident azimuth angles Constructing the array manifold matrix corresponding to a high-order sound field sensor array .

[0017] In some exemplary implementations, based on array manifold matrix Determine the array received signal ; and, based on array received signals Define the array received signal data matrix Y for multi-shot scenarios, including:

[0018] By combining the array manifold matrix, the incident signal vector, the set of signal incident azimuth angles, and Gaussian white noise, a high-order sound field sensor array array receiving signal model is configured, including:

[0019]

[0020] in, This represents the array's received signal model; Represents the incident signal vector; This represents Gaussian white noise with a mean of 0, which is uncorrelated with the incident signal. ; Represents an array manifold matrix;

[0021] Angle scan grid based on observation space Configure signal incident azimuth angle set And reconstruct the array received signal model, including:

[0022]

[0023] in, Represents an array manifold matrix; Let be a matrix of sparse incident signal vectors, and respectively with the signal... correspond; An angular scanning grid for the observation space;

[0024] Based on the reconstructed array received signal, define the array received signal data matrix Y in the multi-shot case.

[0025] In some exemplary embodiments, the process iterates based on the initial values ​​of the spatial azimuth spectrum and the initial values ​​of the array received signal data covariance matrix until a preset threshold is met. The azimuth estimation result is then determined based on the sparsity of the azimuth spectrum obtained from the j-th iteration, including:

[0026] Based on the signal covariance matrix in the real number field , Noise power upper bound configuration for noise power estimation;

[0027] Configure the variable exponential factor corresponding to the current iteration cycle;

[0028] The spatial orientation spectrum is updated based on the covariance matrix in the real number field and the variable exponential factor corresponding to the current iteration period.

[0029] The covariance matrix in the real domain is updated based on the noise power estimate, spatial orientation spectrum, and array manifold matrix in the real domain.

[0030] Determine the directional spectrum sparsity based on the variable exponential factor corresponding to the previous iteration period;

[0031] When the updated azimuth spectrum sparsity in the current iteration cycle meets the preset iteration termination threshold, the iteration stops, and the azimuth estimation result is determined based on the azimuth spectrum sparsity in the current iteration cycle.

[0032] In some exemplary implementations, configuring the variable exponential factor corresponding to the current iteration period includes:

[0033] Configure the corresponding value range for the index factor. ;

[0034] A discrete search is performed based on the range of values ​​for the exponential factor to determine the negative log-likelihood function. ;

[0035] Based on the negative log-likelihood function Determine the optimal variable exponential factor for the current iteration period. ,include:

[0036]

[0037] in, Represents the negative log-likelihood function The inverse function; , which represents a compromise parameter.

[0038] In some exemplary embodiments, determining that the updated azimuth spectrum sparsity meets a preset iteration termination threshold includes:

[0039] The sparsity change value is calculated based on the azimuth spectrum sparsity of two adjacent iteration cycles;

[0040] The ratio of the sparsity change value to that of the previous iteration cycle is compared with a preset threshold.

[0041] If the ratio is less than or equal to the threshold value, the iteration termination condition is determined to be met; otherwise, if the ratio is greater than the threshold value, the iteration termination condition is determined to be not met.

[0042] In some exemplary implementations, based on the signal covariance matrix in the real number field Noise power upper bound configuration noise power estimation, including

[0043]

[0044] in, N This indicates the number of elements in a high-order acoustic field sensor array; N c This indicates the number of modal signal channels output by the high-order sound field sensor.

[0045] In some exemplary embodiments, the spatial orientation spectrum is updated based on the covariance matrix in the real field and the variable exponential factor corresponding to the current iteration period, including:

[0046]

[0047] in, α Indicates a variable exponential factor; Represents an array manifold matrix of the real number field.

[0048] In some exemplary embodiments, the covariance matrix in the real domain is updated based on the noise power estimate, the spatial orientation spectrum, and the array manifold matrix in the real domain, including:

[0049]

[0050] in, Represents the identity matrix; Represents an array manifold matrix of the real number field; Represents the spatial orientation spectrum; This indicates a noise power estimate.

[0051] According to a second aspect of the present invention, a storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the above-described real-valued super-resolution azimuth estimation method for a high-order acoustic field sensor array.

[0052] According to a third aspect of the present invention, a computer program product is provided, on which a computer program is stored, wherein when the computer program is executed by a processor, the above-described real-valued super-resolution azimuth estimation method for a high-order sound field sensor array is implemented.

[0053] According to a fourth aspect of the present invention, an electronic device is provided, comprising:

[0054] Processor; and

[0055] Memory for storing the executable instructions of the processor;

[0056] The processor is configured to implement the above-described real-valued super-resolution azimuth estimation method for a high-order acoustic field sensor array when executing the executable instructions.

[0057] The real-valued super-resolution azimuth estimation method for high-order sound field sensor arrays provided in the embodiments of the present invention utilizes the received signals from multiple high-order sound field sensors in the high-order sound field sensor array. An array manifold matrix is ​​constructed based on the response of the high-order sound field sensor array, thereby constructing a multi-shot array received signal data matrix Y based on the array manifold matrix. By constructing corresponding augmented matrices for the array manifold matrix and the array received signal data matrix, and performing unitary transformation on the augmented matrices, the corresponding real-domain array manifold matrix and array received signal data matrix are obtained. This achieves a transformation from the complex domain to the real domain, making the method applicable to the real-valued variable exponential factor sparse approximate minimum variance method for high-order sound field sensor arrays. By utilizing the covariance matrix of the array received signal data matrix and combining it with the number of array elements and modal signal channels of the high-order sound field sensor array, noise power estimation can be performed. This allows subsequent iterations to calculate noise power only once, saving the iterative update process during noise power calculations. While maintaining noise power estimation performance, this simplifies the calculation process of the sparse approximate minimum variance method with variable exponential factors. Consequently, it achieves effective estimation and accurate resolution of nearby multiple targets while significantly reducing computational load and time, thus improving data processing efficiency.

[0058] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit the invention. Attached Figure Description

[0059] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention. It is obvious that the drawings described below are merely some embodiments of the invention, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0060] Figure 1 The illustration shows a schematic diagram of a real-valued super-resolution azimuth estimation method for a high-order acoustic field sensor array according to an exemplary embodiment of the present invention;

[0061] Figure 2 This diagram schematically illustrates a high-order acoustic field sensor array distribution according to an exemplary embodiment of the present invention.

[0062] Figure 3 The diagram illustrates the azimuth spectrum results of different methods for using a high-order acoustic field sensor array, as an exemplary embodiment of the present invention.

[0063] Figure 4 The diagram illustrates the composition of an electronic device according to an exemplary embodiment of the present invention. Detailed Implementation

[0064] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the invention will be more comprehensive and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0065] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0066] In related technologies, the Sparse Asymtotic Minimum Variance (SAMV) algorithm is a typical non-regular parameter-based algorithm. In this method, noise variance is treated as an unknown parameter for iterative estimation; however, the more unknown parameters there are in the sparse recovery iteration, the worse the robustness of the method. High-order sound field sensors have multi-dimensional output characteristics, offering better performance advantages compared to traditional sound pressure sensors. Further arraying them can achieve superior azimuth estimation performance. However, since the array elements themselves have multi-channel signal outputs and do not possess the characteristics of Hermitian matrices, existing unitary transform techniques cannot be directly applied.

[0067] To address the shortcomings and deficiencies of existing technologies, this example embodiment provides a real-valued super-resolution azimuth estimation method for a high-order sound field sensor array, referencing... Figure 1 As shown, the method includes:

[0068] Step S11: Receive spatial sound field signals using a high-order sound field sensor array, and determine the response vectors of each high-order sound field sensor... Constructing an array manifold matrix ; and based on array manifold matrix Determine the array received signal ; and, based on array received signals Define the array received signal data matrix Y for multi-shot scenarios;

[0069] Step S12: Determine the covariance matrix of the array received signal data matrix, and estimate the noise power by combining the number of array elements of the high-order sound field sensor array and the number of modal signal channels output by the high-order sound field sensor, and configure the upper bound of the noise power.

[0070] Step S13: Construct the augmented matrix Y corresponding to the array received signal data matrix Y. aug ; and for the augmented matrix Y aug Perform a unitary transform to obtain the array received signal data matrix in the real number field. ; Array received signal data matrix in the real number field Calculate the signal covariance matrix in the real domain , and configured as the initial value of the array received signal data covariance matrix;

[0071] Step S14: Construct the array manifold matrix The corresponding augmented array manifold matrix A aug (Θ Ω ); and for the augmented array manifold matrix A aug (Θ Ω Perform a unitary transformation to obtain the array manifold matrix A in the real number field. r(Θ Ω Based on the real-field array manifold matrix A r (Θ Ω Obtain the initial value of the spatial orientation spectrum, and configure the initial value of sparsity based on the initial value of the spatial orientation spectrum;

[0072] Step S15: Using the sparse high-resolution azimuth estimation method, the initial values ​​of the spatial azimuth spectrum and the initial values ​​of the array received signal data covariance matrix are iterated until a preset threshold is met, and the azimuth estimation result is determined according to the sparsity of the azimuth spectrum calculated in the j-th iteration; where j is a positive integer.

[0073] The following will describe in more detail each step of the real-valued super-resolution azimuth estimation method for a high-order sound field sensor array in this exemplary embodiment, with reference to the accompanying drawings and embodiments.

[0074] In step S11, a high-order sound field sensor array is used to receive spatial sound field signals, and the response vectors of each high-order sound field sensor are used to... Constructing an array manifold matrix ; and based on array manifold matrix Determine the array received signal ; and, based on array received signals Define the array received signal data matrix Y for multi-shot scenarios.

[0075] For example, the high-order sound field sensor array includes a linear array constructed based on N high-order sound field sensors; wherein, the high-order sound field sensors include M sound pressure sensors, which are uniformly distributed to form a small-scale circular array, and the diameter of the aperture of the circular array is d; where N and M are positive integers. The range of values ​​for d can be configured according to the signal wavelength; specifically, the range of values ​​for d is... λ is the signal wavelength.

[0076] Specifically, assuming that there exists in the space K A far-field signal ,in, . No. k The horizontal azimuth angle of the incident signal is The unit vector of its propagation direction is defined as:

[0077] (1)

[0078] Where T is the number of snapshots of the received signal.

[0079] Specifically, a high-order sound field sensor is essentially a small-scale circular array composed of multiple uniformly distributed sound pressure sensors. By employing a high-order sound field information extraction method using real-number weighting, it is possible to acquire sound field information of different orders.

[0080] Consider a high-order sound field sensor made of M It consists of several sound pressure sensors, evenly distributed in xoy Diameter on the plane d The diameter of the annular aperture, after real-number weighted processing, can be obtained at most 0 ~ N max First-order modal signal output, N max It is the highest order, and N max ≤M / 2, meaning there are a total of The first modal signal is output. n c The expression for the modal response of a higher-order sound field sensor is:

[0081] (2)

[0082] In the formula, the horizontal azimuth angle in formula (2) is ignored. θ The irrelevant magnitude term yields:

[0083] .

[0084] The modal response vector of a high-order sound field sensor is then defined as follows:

[0085] (3)

[0086] refer to Figure 2 As shown, considering the receiving array as a linear array with N high-order sound field sensors, and taking the center point of each high-order sound field sensor as its location, the position coordinates of the nth high-order sound field sensor can be expressed as:

[0087] (4)

[0088] in, Indicates the first n The line connecting each array element and the origin of the coordinate system. x The angle between the positive and negative axes. Indicates transpose. For the first n The distance between each array element and the origin .

[0089] Of course, in other exemplary embodiments of this disclosure, the high-order sound field sensor array may also be other array forms, such as a ring array, an array with a regular polygonal layout, etc.

[0090] For example, in step S11, based on the response vectors of each higher-order sound field sensor... Constructing an array manifold matrix ,include:

[0091] Step S21, based on the sound pressure response vector received by each high-order sound field sensor and the data based on... N max The modal response vectors of the first-order modal signals are subjected to Kronecker product operation to obtain the response vectors corresponding to each higher-order sound field sensor. ;in, N max ≤M / 2;

[0092] Step S22, based on the response vectors of each higher-order sound field sensor Combined with the set of signal incident azimuth angles Constructing the array manifold matrix corresponding to a high-order sound field sensor array .

[0093] For example, in step S11, based on the array manifold matrix Determine the array received signal ; and, based on array received signals Define the array received signal data matrix Y for multi-shot scenarios, including:

[0094] Step S31, combining the array manifold matrix, the incident signal vector, the set of signal incident azimuth angles, and Gaussian white noise, configure the array receiving signal model of the high-order sound field sensor array, including:

[0095]

[0096] in, This represents the array's received signal model; Represents the incident signal vector; This represents Gaussian white noise with a mean of 0, which is uncorrelated with the incident signal. ; Represents an array manifold matrix;

[0097] Step S32, scan the grid based on the angle of the observation space. Configure signal incident azimuth angle set And reconstruct the array received signal model, including:

[0098]

[0099] in, Represents an array manifold matrix; Let be a matrix of sparse incident signal vectors, and respectively with the signal... correspond; An angular scanning grid for the observation space;

[0100] Step S33: Based on the reconstructed array received signal, define the array received signal data matrix Y for the multi-shot case.

[0101] Specifically, the output signal of a high-order sound field sensor has a multi-dimensional structure, leading to the construction of a long vector signal model for the high-order sound field sensor array. This model treats each output of the high-order sound field sensor as equivalent to the output of a single independent array element. Using the product theorem, the response vector of the high-order sound field sensor array consists of two parts: first, the sound pressure response vector at different positions within the array of high-order sound field sensors; and second, the modal response vectors of the high-order sound field sensors themselves, representing different orders of sound field information. An array of multiple high-order sound field sensors is formed, and the sound pressure response of the nth high-order sound field sensor is expressed as:

[0102] (5)

[0103] in, k Indicates wave number, It is the imaginary unit.

[0104] The sound pressure response vector of the array is defined as follows: .

[0105] According to the product theorem, the array response vector of the high-order sound field sensor is obtained. Also known as an array manifold vector, its expression is as follows:

[0106] (6)

[0107] in, This represents the Kronecker product of vectors.

[0108] Therefore, the received signal of the high-order sound field sensor array can be represented as:

[0109] (7)

[0110] in, , representing the incident signal vector; , represents the array manifold matrix; The set of incident azimuth angles of the signal, i.e. ; It is Gaussian white noise with a mean of 0 and is uncorrelated with the incident signal.

[0111] Define the angle scan grid of the observation space as Its dimension is Obviously, the set of signal incident azimuth angles For scanning grid A subset of, i.e. .generally That is, the number of scanning grids is much greater than the number of signals, so the array receiving signal model in equation (7) is reconstructed:

[0112] (8)

[0113] in, It has sparsity, only K Each of the non-zero elements is associated with the signal. correspond.

[0114] Define the array received signal data matrix for multi-shot scenarios: .

[0115] In step S12, the covariance matrix of the array received signal data matrix is ​​determined, and noise power is estimated by combining the number of array elements of the high-order sound field sensor array and the number of modal signal channels output by the high-order sound field sensor, and an upper bound for noise power is configured.

[0116] Specifically, assuming the array is located in a Gaussian white noise field, let the noise power be... The covariance matrix of the array received signal can be expressed as:

[0117] (9)

[0118] in, Indicates the expectation. This indicates the complex conjugate transpose. , , , , The dimension is The identity matrix.

[0119] According to equation (9), the inverse of the covariance matrix R of the received signal can be expressed as:

[0120] (10)

[0121] in, This represents the matrix inversion operation.

[0122] For matrix By performing eigenvalue decomposition, we can obtain:

[0123] (11)

[0124] in, , representing a matrix The feature matrix is ​​composed of the eigenvectors. It is a diagonal matrix, where the elements on the diagonal are the corresponding eigenvalues, i.e. . and for The eigenvalues ​​and corresponding eigenvectors, when hour, .

[0125] Substituting equation (11) into equation (10), we get the following result: The expression is:

[0126] (12)

[0127] in, .

[0128] According to the Woodbury matrix identity, the above equation can be further simplified to:

[0129] (13)

[0130] Among them, the feature vectors are .

[0131] definition ,use Equation (13) can be rewritten as:

[0132] (14)

[0133] in, .

[0134] By exponentiation on both sides of equation (14), we can further derive the following:

[0135] (15)

[0136] Therefore, when the parameter n When it is big enough, there is .and At that time, the following relationship exists:

[0137] (16)

[0138] Furthermore, it can be deduced that:

[0139] (17)

[0140] Therefore, based on equation (17), the upper bound of the noise power can be obtained, i.e. .

[0141] in, N This represents the number of elements in a high-order acoustic field sensor array. This refers to the number of modal signal channels output by the high-order sound field sensor.K This represents the number of sound sources in the space. However, in practical applications, the number of sound sources in the space is often unknown. K It is an unknown.

[0142] Therefore, based on the above, this invention proposes equation (18) as an approximate estimate of the noise power:

[0143] (18)

[0144] In step S13, the augmented matrix Y corresponding to the array received signal data matrix Y is constructed. aug ; and for the augmented matrix Y aug Perform a unitary transform to obtain the array received signal data matrix in the real number field. ; Array received signal data matrix in the real number field Calculate the signal covariance matrix in the real domain And configured as the initial value of the array received signal data covariance matrix.

[0145] Specifically, unitary transform requires the transformed matrix to be a Hermitian matrix, making it only applicable to linear arrays. High-order sound field sensors have multi-output structures, and their data matrices do not satisfy the Hermitian matrix. Therefore, existing unitary transform methods cannot be directly applied to high-order sound field sensor arrays.

[0146] This invention extends real-valued processing techniques based on unitary transform to arrays with multi-output structures for their array elements. Before implementing the transformation from the complex domain to the real domain, the data matrix Y of the array's received signal can be processed by constructing an augmented matrix Y. aug This makes it a Hermitian matrix:

[0147] (19)

[0148] in, for M A 3D inverse identity matrix, i.e., a matrix whose secondary diagonal elements are all 1s and all other elements are 0s. phalanx, for T 3D inverse identity matrix This indicates the complex conjugate operation.

[0149] Extended augmented matrix Y aug By performing a unitary transform, the array received signal data matrix in the real number field can be obtained. This process preserves the azimuth information of the signal. The main principle of the unitary transform is to utilize the properties of the central Hermitian matrix to receive the signal data matrix in the real number field. The specific expression is:

[0150] (20)

[0151] in, Represents the matrix Y after unitary transformation aug Realization result, and All are unitary matrices.

[0152] unitary matrix For example, the representation of a unitary matrix varies depending on its dimension. M When the number is even, we have:

[0153] (twenty one)

[0154] when M When the number is odd, we have:

[0155] (twenty two)

[0156] in, for M / 2D identity matrix, for( M -1) / 2D identity matrix.

[0157] Based on the array received signal data matrix in the real number field, the signal covariance matrix in the real number field can be obtained and used as the initial value of the array received signal data covariance matrix. The specific formula is as follows:

[0158] (twenty three)

[0159] In step S14, the array manifold matrix is ​​constructed. The corresponding augmented array manifold matrix A aug (Θ Ω ); and for the augmented array manifold matrix A aug (Θ Ω Perform a unitary transformation to obtain the array manifold matrix A in the real number field. r (Θ Ω Based on the real-field array manifold matrix A r (Θ Ω Obtain the initial value of the spatial orientation spectrum and configure the initial value of sparsity based on the initial value of the spatial orientation spectrum.

[0160] Specifically, considering that the array manifold matrix of a high-order sound field sensor array with multi-output elements does not satisfy the properties of the Hermitian matrix, the same processing procedure described above is used to first obtain the augmented array manifold matrix. And then Performing a unitary transformation yields a real-field array manifold. To realize the transformation of complex field array manifold Transform to a real-field array manifold It still retains angle-related features.

[0161] The initial value of the spatial azimuth spectrum can be calculated using conventional beam scanning azimuth estimation methods. The calculation expression is as follows:

[0162] (twenty four)

[0163] The initial value expression for sparsity is defined as follows:

[0164] (25)

[0165] in, This is used to calculate the 1-norm of a vector. The larger the sparsity value, the worse the sparsity of the azimuth spectrum; conversely, the smaller the sparsity value, the higher the sparsity of the azimuth spectrum.

[0166] Specifically, the real-valued transformation is achieved through unitary transformation, which is applied to both the received signal matrix and the array manifold matrix of the high-order sound field sensor array. However, traditional unitary transformation requires the matrix to be Hermitian, which is not met by the received signal matrix and array manifold matrix of the high-order sound field sensor array. Therefore, this invention constructs augmented matrices of the received data and array manifold matrix to make them Hermitian matrices. Then, unitary transformation is applied to convert the array received signal and array manifold matrix from the complex domain to the real domain.

[0167] In step S15, the sparse high-resolution azimuth estimation method is used to iterate based on the initial value of the spatial azimuth spectrum and the initial value of the covariance matrix of the array received signal data until a preset threshold is met. The azimuth estimation result is determined according to the sparsity of the azimuth spectrum calculated in the j-th iteration; where j is a positive integer.

[0168] For example, step S15 described above may include:

[0169] Step S41, based on the signal covariance matrix in the real domain , Noise power upper bound configuration for noise power estimation;

[0170] Step S42: Configure the variable exponential factor corresponding to the current iteration cycle;

[0171] Step S43: Update the spatial orientation spectrum based on the covariance matrix of the real number field and the variable exponential factor corresponding to the current iteration period;

[0172] Step S44: Update the covariance matrix in the real domain based on the noise power estimate, spatial orientation spectrum, and array manifold matrix in the real domain.

[0173] Step S45, and determine the azimuth spectrum sparsity based on the variable exponential factor corresponding to the previous iteration period;

[0174] Step S46: When the updated azimuth spectrum sparsity in the current iteration cycle meets the preset iteration termination threshold, stop the iteration and determine the azimuth estimation result based on the azimuth spectrum sparsity in the current iteration cycle.

[0175] For example, in step S42 above, configuring the variable exponential factor corresponding to the current iteration cycle includes:

[0176] Configure the corresponding value range for the index factor. ;

[0177] A discrete search is performed based on the range of values ​​for the exponential factor to determine the negative log-likelihood function. ;

[0178] Based on the negative log-likelihood function Determine the optimal variable exponential factor for the current iteration period. .

[0179] For example, in step S46 above, determining whether the updated azimuth spectrum sparsity satisfies a preset iteration termination threshold includes:

[0180] The sparsity change value is calculated based on the azimuth spectrum sparsity of two adjacent iteration cycles;

[0181] The ratio of the sparsity change value to that of the previous iteration cycle is compared with a preset threshold.

[0182] If the ratio is less than or equal to the threshold value, the iteration termination condition is determined to be met; otherwise, if the ratio is greater than the threshold value, the iteration termination condition is determined to be not met.

[0183] Specifically, the SAMV algorithm based on the sparse approximate minimum variance criterion can be viewed as a sparse approximate minimum variance algorithm with a fixed exponent factor. Choosing an appropriate exponent factor can effectively improve the sparsity of the azimuth spectrum, achieving super-resolution azimuth estimation. Based on this, this invention proposes a real-valued sparse high-resolution azimuth estimation method suitable for high-order sound field sensor arrays. The iterative calculation process of the spatial azimuth spectrum and covariance matrix is ​​as follows:

[0184] The calculation expression for the noise power estimate is obtained from equations (18) and (23):

[0185] (26)

[0186] Let the variable exponential factor be α To ensure that all variable exponential factors are positive, their value range is set to... .

[0187] Update the spatial orientation spectrum, number j Spatial orientation spectrum of different index factors obtained in the next iteration The calculation formula is:

[0188] (27)

[0189] Update the covariance matrix, the first... j The real-field covariance matrix obtained in the next iteration The calculation formula is:

[0190] (28)

[0191] in, Represents the M-dimensional identity matrix; Represents an array manifold matrix of the real number field; Represents the spatial orientation spectrum; This indicates a noise power estimate.

[0192] Using the maximum likelihood estimation method, a discrete search is performed within the range of values ​​for the variable exponential factor to calculate the negative log-likelihood function. The expression is as follows:

[0193] (29)

[0194] Define the optimal variable exponential factor The calculation formula is as follows:

[0195] (30)

[0196] in, Represents the negative log-likelihood function inverse function, It is a trade-off parameter, which can be considered as a trade-off factor between maximum likelihood estimation and azimuth spectrum sparsity, that is, a trade-off between estimation accuracy and sparsity.

[0197] Calculate the first j The value of azimuth spectrum sparsity after the next iteration under the optimal variable exponential factor:

[0198] (31)

[0199] Iterative calculations of spatial orientation spectrum and covariance matrix require a given iteration termination threshold. The relative change in azimuth spectrum sparsity obtained from two iterations is used as the iteration termination condition, i.e.:

[0200] (32)

[0201] For example, after obtaining the azimuth spectrum sparsity of the current j-th iteration, the ratio between it and the azimuth spectrum sparsity of the (j-1)-th iteration is calculated using the formula (32) above. If the ratio is less than or equal to the termination threshold, the iteration is terminated, and the azimuth of the sound source is calculated based on the azimuth spectrum sparsity of the current j-th iteration. For example, the target azimuth can be obtained through spectral peak search.

[0202] Specifically, the improved noise power calculation method provided by this invention can replace the iterative update process of noise power in the original sparse approximate minimum variance method with variable exponential factors by calculating the noise power approximation value once. While ensuring the estimation performance, it simplifies the calculation process of the sparse approximate minimum variance method with variable exponential factors.

[0203] For example, in one embodiment, the receiving array is a high-order sound field sensor linear array with 6 elements. The elements are spaced at half the wavelength of the center frequency, which is set to 1 kHz. Two uncorrelated far-field complex Gaussian sound sources of equal intensity exist in a two-dimensional plane. The number of snapshots is 200, and the sound sources are located at 90° and 100° directions, respectively. First, the signal-to-noise ratio of the incident signals from both complex Gaussian sound sources is set to 3 dB. Each high-order sound field sensor is a small-scale circular array composed of 8 sound pressure sensors with an array radius of 0.08 meters. Signals from the 0th to 4th orders are extracted and processed. In the simulation, parameter n is set to 20, the compromise factor to 0.05, and the iteration termination threshold to 0.02. The number of Monte Carlo trials is set to 200. Figure 3 The azimuth spectrum results of different methods for high-order sound field sensor arrays are shown. It can be seen that the method proposed in this invention has a narrower main lobe width and lower side lobes compared with CBF, MVDR and MUSIC methods, indicating that the method proposed in this invention has better resolution of nearby multiple targets and better azimuth estimation performance.

[0204] Furthermore, as shown in Table 1, the average computation time of the method proposed in this invention and the existing SAMV-α method can be compared. It can be seen that the present invention can effectively improve computational efficiency, and the computation time can be shortened to 57% of that of the SAMV-α method.

[0205] Table 1

[0206]

[0207] The method provided in this invention proposes a real-valued sparse high-resolution azimuth estimation method applicable to arbitrary arrays. First, the noise power estimate is approximated using the received signal covariance matrix, avoiding the iterative noise power calculation process in existing sparse approximation minimum variance methods with variable exponential factors. This improves the computational efficiency of the method and reduces the amount of floating-point operations. .

[0208] Furthermore, traditional real-valued transformation based on unitary transform requires the transformation matrix to satisfy the properties of Hermitian matrices, thus only applicable to linear arrays. This invention, however, constructs an augmented matrix that makes the array's received data matrix and array manifold matrix, which have multi-dimensional structures, Hermitian matrices, thereby enabling unitary transform processing of relevant parameters of high-order sound field sensor arrays.

[0209] Furthermore, based on the array received signal data matrix and array manifold matrix obtained in the real domain, these are applied to the sparse approximate minimum variance method for variable exponential factors, achieving fully real-valued sparse approximate minimum variance azimuth estimation of variable exponential factors. Compared to existing methods, this invention employs a fully real-valued processing procedure, which significantly improves computational efficiency. The computation time of the proposed method is significantly shorter than that of existing sparse approximate minimum variance methods for variable exponential factors, and it is applicable to arrays with multi-dimensional array elements, including high-order acoustic field sensor arrays and vector sensor arrays, thus having a wide range of applications.

[0210] It should be noted that the above figures are merely illustrative of the processes included in the method according to exemplary embodiments of the present invention, and are not intended to be limiting. It is readily understood that the processes shown in the above figures do not indicate or limit the temporal order of these processes. Furthermore, it is readily understood that these processes may, for example, be executed synchronously or asynchronously in multiple modules.

[0211] It should be noted that although several modules or units of the device for performing actions have been mentioned in the detailed description above, this division is not mandatory. In fact, according to embodiments of the present invention, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0212] Figure 4 A schematic diagram of an electronic device suitable for implementing embodiments of the present invention is shown.

[0213] It should be noted that, Figure 4 The electronic device 1000 shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments of the present invention.

[0214] like Figure 4 As shown, the electronic device 1000 includes a Central Processing Unit (CPU) 1001, which can perform various appropriate actions and processes based on programs stored in Read-Only Memory (ROM) 1002 or programs loaded from storage section 1008 into Random Access Memory (RAM) 1003. The RAM 1003 also stores various programs and data required for system operation. The CPU 1001, ROM 1002, and RAM 1003 are interconnected via a bus 1004. An Input / Output (I / O) interface 1005 is also connected to the bus 1004. Furthermore, the electronic device 1000 also includes an FPGA device and a System-on-a-Chip (SoC) device.

[0215] The following components are connected to I / O interface 1005: an input section 1006 including a keyboard, mouse, etc.; an output section 1007 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 1008 including a hard disk, etc.; and a communication section 1009 including a network interface card such as a LAN (Local Area Network) card, modem, etc. The communication section 1009 performs communication processing via a network such as the Internet. A drive 1010 is also connected to I / O interface 1005 as needed. Removable media 1011, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., are installed on drive 1010 as needed so that computer programs read from them can be installed into storage section 1008 as needed.

[0216] In particular, according to embodiments of the present invention, the processes described below with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a storage medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 1009, and / or installed from removable medium 1011. When the computer program is executed by central processing unit (CPU) 1001, it performs various functions defined in the system of this application.

[0217] Specifically, the aforementioned electronic devices can be airborne intelligent electronic devices, such as airborne video processing equipment.

[0218] It should be noted that the storage medium shown in the embodiments of the present invention can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, optical fiber, portable compact disc read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In the present invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In the present invention, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, wherein computer-readable program code is carried. Such transmitted data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The computer-readable signal medium can also be any storage medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the storage medium can be transmitted using any suitable medium, including but not limited to wireless, wired, etc., or any suitable combination thereof.

[0219] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0220] The units described in the embodiments of the present invention can be implemented in software or hardware, and the described units can also be located in a processor. The names of these units do not necessarily limit the specific unit itself.

[0221] It should be noted that, as another aspect, this application also provides a storage medium, which may be included in an electronic device or may exist independently without being assembled into the electronic device. The aforementioned storage medium carries one or more programs, which, when executed by an electronic device, cause the electronic device to perform the methods described in the following embodiments. For example, the electronic device may perform... Figure 1 The steps of the method shown.

[0222] In one embodiment, this application provides a computer program product including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0223] Furthermore, the above figures are merely illustrative of the processes included in the method according to exemplary embodiments of the present invention, and are not intended to be limiting. It is readily understood that the processes shown in the above figures do not indicate or limit the temporal order of these processes. Additionally, it is readily understood that these processes may be executed synchronously or asynchronously, for example, in multiple modules.

[0224] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention herein. This application is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the claims.

[0225] It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A real-valued super-resolution azimuth estimation method for a high-order sound field sensor array, characterized in that, The method includes: A high-order sound field sensor array is used to receive spatial sound field signals, and the response vectors of each high-order sound field sensor are used to... Constructing an array manifold matrix ; and based on array manifold matrix Determine the array received signal ; and, based on array received signals Define the array received signal data matrix Y for multi-shot scenarios; Determine the covariance matrix of the array received signal data matrix, and estimate the noise power by combining the number of array elements of the high-order sound field sensor array and the number of modal signal channels output by the high-order sound field sensor, and configure the upper bound of the noise power. Construct the augmented matrix Y corresponding to the array received signal data matrix Y. aug ; and for the augmented matrix Y aug Perform a unitary transform to obtain the array received signal data matrix in the real number field. ; Array received signal data matrix in the real number field Calculate the signal covariance matrix in the real domain , and configured as the initial value of the array received signal data covariance matrix; Constructing an array manifold matrix The corresponding augmented array manifold matrix A aug (Θ Ω ); and for the augmented array manifold matrix A aug (Θ Ω Perform a unitary transformation to obtain the array manifold matrix A in the real number field. r (Θ Ω Based on the real-field array manifold matrix A r (Θ Ω Obtain the initial value of the spatial orientation spectrum, and configure the initial value of sparsity based on the initial value of the spatial orientation spectrum; The sparse high-resolution azimuth estimation method is used to iterate until a preset threshold is met based on the initial values ​​of the spatial azimuth spectrum and the initial values ​​of the covariance matrix of the array received signal data. The azimuth estimation result is determined according to the sparsity of the azimuth spectrum calculated in the j-th iteration; where j is a positive integer.

2. The method according to claim 1, characterized in that, The high-order sound field sensor array includes a linear array constructed based on N high-order sound field sensors; wherein, the high-order sound field sensors include M sound pressure sensors, which are uniformly distributed to form a small-scale circular array, and the diameter of the aperture of the circular array is d; where N and M are positive integers.

3. The method according to claim 1, characterized in that, Based on the response vectors of each high-order sound field sensor Constructing an array manifold matrix ,include: Based on the sound pressure response vectors received by each high-order sound field sensor and based on N max The modal response vectors of the first-order modal signals are subjected to Kronecker product operation to obtain the response vectors corresponding to each higher-order sound field sensor. ;in, N max ≤M / 2; Based on the response vectors of each high-order sound field sensor Combined with the set of signal incident azimuth angles Constructing the array manifold matrix corresponding to a high-order sound field sensor array .

4. The method according to claim 1 or 3, characterized in that, Based on array manifold matrix Determine the array received signal ; And, based on array received signals Define the array received signal data matrix Y for multi-shot scenarios, including: By combining the array manifold matrix, the incident signal vector, the set of signal incident azimuth angles, and Gaussian white noise, a high-order sound field sensor array array receiving signal model is configured, including: in, This represents the array's received signal model; Represents the incident signal vector; This represents Gaussian white noise with a mean of 0, which is uncorrelated with the incident signal. ; Represents an array manifold matrix; Angle scan grid based on observation space Configure signal incident azimuth angle set And reconstruct the array received signal model, including: in, Represents an array manifold matrix; Let be a matrix of sparse incident signal vectors, and respectively with the signal... correspond; An angular scanning grid for the observation space; Based on the reconstructed array received signal, the array received signal data matrix Y is defined in the multi-shot case.

5. The method according to claim 1, characterized in that, The process iterates based on the initial values ​​of the spatial azimuth spectrum and the initial values ​​of the array received signal data covariance matrix until a preset threshold is met. The azimuth estimation result is then determined based on the sparsity of the azimuth spectrum obtained from the j-th iteration, including: Based on the signal covariance matrix in the real number field , Noise power upper bound configuration for noise power estimation; Configure the variable exponential factor corresponding to the current iteration cycle; The spatial orientation spectrum is updated based on the covariance matrix in the real number field and the variable exponential factor corresponding to the current iteration period. The covariance matrix in the real domain is updated based on the noise power estimate, spatial orientation spectrum, and array manifold matrix in the real domain. Determine the directional spectrum sparsity based on the variable exponential factor corresponding to the previous iteration period; When the updated azimuth spectrum sparsity in the current iteration cycle meets the preset iteration termination threshold, the iteration stops, and the azimuth estimation result is determined based on the azimuth spectrum sparsity in the current iteration cycle.

6. The method according to claim 5, characterized in that, Configure the variable exponential factor corresponding to the current iteration cycle, including: Configure the corresponding value range for the variable exponential factor. ; A discrete search is performed based on the range of values ​​for the variable exponential factor to determine the negative log-likelihood function. ; Based on the negative log-likelihood function Determine the optimal variable exponential factor for the current iteration period. ,include: in, Represents the negative log-likelihood function The inverse function; , which represents a compromise parameter.

7. The method according to claim 5, characterized in that, Determine if the updated azimuth spectrum sparsity meets the preset iteration termination threshold, including: The sparsity change value is calculated based on the azimuth spectrum sparsity of two adjacent iteration cycles; The ratio of the sparsity change value to that of the previous iteration cycle is compared with a preset threshold. If the ratio is less than or equal to the threshold value, the iteration termination condition is determined to be met; otherwise, if the ratio is greater than the threshold value, the iteration termination condition is determined to be not met.

8. The method according to claim 5, characterized in that, Based on the signal covariance matrix in the real number field The upper bound of noise power is configured for noise power estimation, including: in, N This indicates the number of elements in a high-order acoustic field sensor array; N c This indicates the number of modal signal channels output by the high-order sound field sensor.

9. The method according to claim 5, characterized in that, The spatial orientation spectrum is updated based on the covariance matrix in the real field and the variable exponential factor corresponding to the current iteration period, including: in, α Indicates a variable exponential factor; Represents an array manifold matrix of the real number field.

10. The method according to claim 5, characterized in that, The covariance matrix in the real domain is updated based on the noise power estimate, spatial orientation spectrum, and array manifold matrix in the real domain, including: in, Represents the identity matrix; Represents an array manifold matrix of the real number field; Represents the spatial orientation spectrum; This indicates a noise power estimate.

Images