Three-dimensional foresight sonar imaging optimization method based on deconvolution split aperture

Abstract

The invention discloses a three-dimensional foresight sonar imaging optimization method of a deconvolution split aperture, which comprises the following steps: performing conventional beam forming and deconvolution on original echo data of a complete planar array, and then estimating a beam arrival time window of each beam direction according to amplitude energy arrangement of a deconvolution beam forming result; the method comprises the following steps: respectively performing conventional beam forming on original data of left and right sub-arrays split in the horizontal direction and original data of upper and lower sub-arrays split in the vertical direction, and extracting beam phase information; calculating the phase difference of the left and right subarrays and the phase difference of the upper and lower subarrays in the wave beam arrival time window in each wave beam direction to obtain a horizontal splitting phase difference curve and a vertical splitting phase difference curve; and determining the final wave beam arrival time after evaluating the linear trend significance and goodness of fit of the curve, and calculating the scatterer distance in each wave beam direction according to the final wave beam arrival time result. Therefore, the distance measurement precision of the three-dimensional sonar system, the continuity of imaging point cloud and the robustness of strong target imaging can be improved.

Description

Technical Field

[0001] This invention belongs to the field of measurement technology, specifically relating to a three-dimensional forward-looking sonar imaging optimization method based on deconvolution split aperture. Background Technology

[0002] Three-dimensional forward-looking sonar transmits a narrowband acoustic pulse and uses a two-dimensional transducer array to receive the echo scattered signals from the underwater scene for beamforming and image processing, ultimately obtaining high-resolution underwater three-dimensional imaging. Three-dimensional forward-looking sonar operates stably in turbid waters and complex currents, and can be widely used in underwater exploration, salvage operations, structural flaw detection, and other scenarios.

[0003] Sidelobes are additional signals generated outside the main beam of a sonar signal due to factors such as array antenna design and signal processing. These sidelobes often cause false imaging in sonar images, especially in the detection of targets near strong interference sources or at close range. Sidelobes can easily lead to the appearance of false targets, thereby reducing the accuracy and reliability of the system. Traditional sidelobe suppression methods mainly rely on array weighting techniques. However, in complex sonar environments, the suppression effect of sidelobe energy is limited and may affect the performance of the main lobe. Therefore, deconvolved conventional beamforming (dCBF) has been introduced into the field of array signal processing. The algorithm treats the beam azimuth map of the sonar system as a convolution of the target's spatial distribution function and the beam directivity function. By performing deconvolution operations to recover the original signal distribution, it can effectively improve the resolution of the sonar system and suppress sidelobes, exhibiting high robustness.

[0004] The range resolution of a sonar system is a key indicator for evaluating its imaging performance and the effectiveness of sonar image applications. Limited by the massive amount of point cloud data in 3D sonar and the occlusion phenomenon in 3D point cloud display, 3D forward-looking sonar can only select a portion of the point cloud data from each beam direction after beamforming. Common data selection methods for 3D sonar include the echo threshold method and the amplitude maxima method. The echo threshold method requires manually setting a detection amplitude threshold, and then determining the echo arrival time of each beam as the first or all times in the time series of its output exceeding this amplitude threshold. The amplitude threshold is related to factors such as the distance to the imaging target, its material, and noise. Setting the threshold too high will filter the target point cloud, while setting it too low will increase the amount of uploaded data and blur the imaging in the distance dimension. The amplitude maxima method estimates the echo arrival time window at each beam angle and selects the moment of maximum echo energy as the echo arrival time, suitable for ideal models with uniformly distributed scatterers. However, in sonar imaging of small-aperture, continuous targets, the sonar images using the amplitude maxima method have problems such as patchy repetition, discontinuous imaging results, and inaccurate distance measurement because the amplitude of the echo beam from the scattering object varies in strength while the sonar beam footprint is large.

[0005] When the above two data selection methods are applied to three-dimensional forward-looking sonar, there are problems of ambiguity or error in the measurement of the distance dimension, and the beam sidelobe is prone to false imaging when facing close-range targets or highly reflective targets. Summary of the Invention

[0006] In view of the above, the purpose of this invention is to provide a three-dimensional forward-looking sonar imaging optimization method based on deconvolution split aperture. This method combines the conventional deconvolution beamforming algorithm with the design concept of two-dimensional split aperture to improve the ranging accuracy, the continuity of the imaging point cloud, and the robustness of strong target imaging of the three-dimensional sonar system.

[0007] To achieve the above-mentioned objectives, an embodiment provides a three-dimensional forward-looking sonar imaging optimization method based on deconvolution split aperture, comprising the following steps: Perform conventional beamforming on the raw echo data of the complete planar array, and then perform deconvolution on the conventional beamforming results. Based on the amplitude and energy arrangement of the deconvolution beamforming results, predict the beam arrival time window for each beam direction. Conventional beamforming is performed on the original echo data of the horizontally split left and right subarrays and the original data of the vertically split upper and lower subarrays, and beam phase information is extracted. Based on the extracted beam phase information, the phase difference between the left and right subarrays and the phase difference between the upper and lower subarrays within the beam arrival time window in each beam direction are calculated to obtain the horizontal split phase difference curve and the vertical split phase difference curve. After evaluating the significance and goodness of the linear trends of the horizontal and vertical split phase difference curves, the final beam arrival time is determined, and the scatterer distance in each beam direction is calculated based on the final beam arrival time.

[0008] Preferably, conventional beamforming is performed on the raw echo data of the complete array, including: (1) in, The beam azimuth angle, For beam pitch angle, for time Beamforming results in the direction, For the array element in Echo data sampled at any time Frequency coefficients, represents the weighting coefficients of the array elements. The time delay parameters of the receiving array elements in each beam direction under far-field conditions are given. This is the receive array vector.

[0009] Preferably, deconvolution is performed on the conventional beamforming results, including: (2) in, The beam azimuth angle, For beam pitch angle, for time Beamforming results in the direction, and The first +1 times and the iteration The Richard-Lucy deconvolution beamforming results after next-generation calculation The point spread function is related to array parameters and signal frequency. Indicates the relevant operations, This represents the dot product operation.

[0010] Preferably, the array splitting method in the horizontal and vertical directions includes: The array is divided into a left subarray and a right subarray along the horizontal direction, wherein the array centers of the two left and right subarrays are located on the same horizontal line, and the array elements may have overlapping parts; The array is divided into an upper subarray and a lower subarray along the vertical direction, wherein the centers of the two upper and lower subarrays are located on the same vertical line, and the array elements may have overlapping parts; The array splitting method is applicable to various planar arrays including circular arrays, rectangular arrays, and sparse arrays, and allows the array to be split into two or more subarrays in any direction.

[0011] Preferably, the beam phase information of the beamforming results of each subarray is extracted, including: (3) in, , The left and right subarray vectors of the receiving array are split horizontally. and The upper and lower subarray vectors of the receiving array are split vertically. , , , The left, right, upper, and lower pieces are respectively in time Beamforming results in the direction, , , , The left, right, upper, and lower pieces are respectively in time Beam phase in the direction, The operation of taking the phase of a complex number.

[0012] Preferably, the beam arrival time window for each beam direction is estimated based on the amplitude-energy arrangement of the deconvolution beamforming result, including: Based on the deconvolution beamforming results of a complete planar array Amplitude energy distribution and amplitude threshold over time in each beam direction The time window during which the beam first continuously crosses the amplitude threshold is selected as the estimated beam arrival time window. .

[0013] Preferably, the horizontal split phase difference curve and the vertical split phase difference curve are obtained by calculating the phase difference between the left and right subarrays and the phase difference between the upper and lower subarrays within the beam arrival time window in each beam direction based on the extracted beam phase information, including: Calculate the phase difference: (4) (5) in, For the beam arrival time window in each beam direction, , , , The left, right, upper, and lower pieces are respectively in time Beam phase in the direction, To estimate the horizontal subarray within the time window Horizontal split phase difference matrix in the beam direction To estimate the vertical subarray within the time window Vertical split phase difference matrix in the beam direction; Fitted phase difference curve: (6) (7) in, for The linear fitting results, for The linear fitting results, , , , These are the estimated values ​​during the fitting process. To estimate the time matrix of the echo arrival window, .

[0014] Preferably, the final echo beam arrival time is determined after evaluating the significance and goodness of the linear trend of the horizontal split phase difference curve and the vertical split phase difference curve, including: Calculate the determining parameters for linear fitting of the horizontal and vertical split phase differences respectively: (8) (9) in, , These are the coefficients of determination for the linear fitting of the horizontal and vertical split phase differences, respectively, used to evaluate the goodness of linear fit of the model. This represents the summation operation of matrix elements. This represents the operation of averaging matrix elements; By comparing the determination coefficients of the horizontal and vertical phase difference fits, the matrix with the more significant linear trend is selected as the result: (10) in, For phase difference matrices with a more pronounced linear trend, As a threshold for the coefficient of determination, a coefficient of determination lower than [a certain threshold] is set. The phase difference matrix is ​​considered to be non-significantly linear; Final beam arrival time It is obtained directly from the fitted estimate: (11) in This is a calculation that rounds down to the nearest integer.

[0015] Preferably, the distance of the scatterer in each beam direction is calculated based on the final beam arrival time result, including: To improve the robustness of beam arrival time calculation under low signal-to-noise ratio conditions, data with obvious beam arrival time errors are processed. (12) in, The beam arrival time after removing erroneous data, Results of deconvolution beamforming The time corresponding to the point of maximum energy; Based on the imaging principle of active sonar, we can conclude that: (13) in, for Distance of the scatterer in the beam direction This is the speed of sound in water.

[0016] Compared with the prior art, the beneficial effects of the present invention include at least the following: This invention reduces sidelobe errors in strong targets while maintaining robustness in 3D imaging, and is applicable to various planar array configurations such as circular, rectangular, and sparse arrays. It integrates amplitude and phase information from beamforming results to achieve more accurate echo time measurement. Compared to the echo threshold method, this method offers lower data upload bandwidth, higher ranging accuracy, and more robust data selection. Compared to the amplitude maxima method, it offers higher ranging accuracy and continuity of the imaging point cloud. It can significantly improve the resolution of underwater 3D imaging without changing the array aperture, with particularly significant improvements for small-aperture 3D forward-looking sonar. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of the three-dimensional forward-looking sonar imaging optimization method based on deconvolution split aperture provided in the embodiment; Figure 2 This is a schematic diagram of the horizontal and vertical split apertures of the transducer planar array provided in the embodiment; Figure 3 These are the horizontal and vertical split phase difference curves within the estimated echo time window provided in the embodiment; Figure 4 This is an underwater three-dimensional imaging point cloud map using the amplitude maxima method provided in the embodiment; Figure 5 This is an underwater three-dimensional imaging point cloud map using the method of the present invention, provided in an embodiment. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and do not limit the scope of protection of this invention.

[0020] This embodiment provides a three-dimensional forward-looking sonar imaging optimization method based on deconvolution split aperture, applicable to three-dimensional forward-looking sonar systems. This system includes... N A two-dimensional array of arbitrary shapes with array elements is located in the XOY plane of three-dimensional space. For receiving array vector, The relative coordinates of the array elements, and the beam direction is determined by two parameters. and This indicates that the number of beams is For the ( ) , ) beam, its beam direction is ( , The beam signal can be represented as: in, for time( , Beamforming in the direction of ) For the array element in Echo data sampled at any time Frequency coefficients, represents the weighting coefficients of the array elements. Indicates that the array element is for ( , Phase shift in the direction, Represents the imaginary unit, and When the speed of sound in water is: In a specific example, the transducer array in a particular phased array three-dimensional forward-looking sonar system is a... A two-dimensional uniform rectangular array with a signal center frequency of 300kHz and an element spacing of [missing information]. , The signal wavelength is given; the array center is located at the origin, and the calculation is performed within the range of (-50°, 50°). One beam. The original data for the array in this example is underwater topographic data of Mulan Lake in Wuhan.

[0021] The three-dimensional forward-looking sonar imaging optimization method based on deconvolution split aperture provided in the embodiment processes the raw echo data of the above array, including the following steps: S1 performs conventional beamforming on the raw echo data of the complete planar array. After deconvolution of the conventional beamforming results, the arrival time window of each beam direction is estimated based on the amplitude-energy arrangement of the deconvolution beamforming results.

[0022] In the embodiment, according to formula (1), the size is of Perform conventional beamforming to obtain a size of Beamforming results .

[0023] According to formula (2), the results of conventional beamforming of the full array are... Iterative computation of the Richard-Lucy deconvolution is performed. For time-invariant arrays, convolution and correlation operations can be transformed into frequency domain multiplication. The RL algorithm is implemented based on FFT (Fast Fourier Transform), thereby accelerating the iterative computation. in, This indicates element-wise division. For the first After the next iteration Deconvolution beamforming results for The point spread function matrix.

[0024] In the embodiment, based on the amplitude threshold and deconvolution beamforming results Amplitude energy distribution over time in each beam direction The time window during which the beam first continuously crosses the amplitude threshold is selected as the estimated beam arrival time window. With the ( , Taking one beam direction as an example, , .

[0025] S2, perform conventional beamforming on the original echo data of the horizontally split left and right subarrays and the original data of the vertically split upper and lower subarrays, and extract beam phase information. Based on the extracted beam phase information, calculate the phase difference between the left and right subarrays and the phase difference between the upper and lower subarrays within the beam arrival time window for each beam direction to obtain the horizontal split phase difference curve and the vertical split phase difference curve.

[0026] like Figure 2 As shown, The planar receiver array is split horizontally into The rectangular left subarray vector With the right submatrix vector of the rectangle The distance between the centers of the left and right subarrays is ; split vertically into Rectangular subarray vector With rectangular submatrix vector The distance between the centers of the upper and lower subarrays is .

[0027] Similarly, the left, right, upper, and lower subarrays are obtained according to the frequency domain conventional beamforming formula shown in formula (1). Beamforming results , , , .

[0028] In the embodiment, firstly based on the beamforming results , , , The phase difference is calculated using formulas (3), (4), and (5): Among them, the (th) , Taking one beam direction as an example, for The horizontal phase difference matrix, for The vertical phase difference matrix.

[0029] Then, the fit is performed according to formulas (6) and (7) to achieve the desired result. With the ( , Taking one beam direction as an example, , , , To obtain as follows Figure 3The horizontal and vertical split phase difference curves within the estimated echo time window for this beam direction are shown. It can be seen that the horizontal split phase difference curve has a significant linear trend, while the vertical split phase difference curve does not have a significant linear trend.

[0030] S3. After evaluating the significance and goodness of the linear trends of the horizontal split phase difference curve and the vertical split phase difference curve, determine the final beam arrival time. Calculate the scatterer distance in each beam direction based on the final beam arrival time results.

[0031] In the embodiment, the determining parameters for linear fitting of the horizontal and vertical split phase differences are calculated according to formulas (8) and (9). , By comparing the coefficients of determination for fitting horizontal and vertical phase differences, and the threshold of the coefficients of determination... The horizontal split phase difference matrix, which exhibits a more significant linear trend, is selected as the output result, and the zero-crossing time of the fitted curve is calculated. As the beam arrival time .

[0032] In this embodiment, data with obvious errors in beam arrival time are processed using formula (12) to improve the robustness of beam velocity arrival time calculation under low signal-to-noise ratio conditions. Then, the distance to the scatterer in the beam direction is calculated according to formula (13).

[0033] Underwater 3D imaging point cloud map using the amplitude maxima method is shown below. Figure 4 As shown, the method produces patchy point clouds of underwater topography with numerous holes. The bounding boxes indicate misimages caused by side lobes, resulting in significant errors in target distance measurement. The 3D imaging point cloud data volume using the amplitude maxima method is 26288.

[0034] The underwater three-dimensional imaging point cloud image using the method of this invention is shown below. Figure 5 As shown, the point cloud image is continuous and detailed, with higher resolution, smaller distance measurement error, and better reflects the real underwater topographic features. Meanwhile, the method of this invention successfully suppresses... Figure 4 The false imaging caused by side lobes at the box can more accurately and realistically reconstruct underwater scenes under near-target and strong-target conditions. Furthermore, the 3D imaging point cloud data volume using the method of this invention is 26845, which is basically consistent with the amplitude maxima method.

[0035] The method proposed in this invention effectively suppresses sidelobes in sonar systems and significantly improves imaging resolution and accuracy by introducing deconvolution computation and split aperture technology. While this method achieves significant performance improvements, its main cost lies in the substantial increase in computational load, especially during deconvolution computation and subarray beamforming. In practical sonar deployments, the overall system efficiency can be further improved based on parallel computing, heterogeneous architectures of FPGAs and CPUs, and a combination of online and offline computing. Compared to methods that improve imaging quality by increasing array aperture, increasing the number of channels, and expanding device size, deploying the method of this invention is a more acceptable solution.

[0036] The specific embodiments described above illustrate the technical solution and beneficial effects of the present invention in detail. It should be understood that the above description is only the most preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, additions, and equivalent substitutions made within the scope of the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A three-dimensional forward-looking sonar imaging optimization method based on deconvolution split aperture, characterized in that, Includes the following steps: Perform conventional beamforming on the raw echo data of the complete planar array, and then perform deconvolution on the conventional beamforming results. Based on the amplitude and energy arrangement of the deconvolution beamforming results, predict the beam arrival time window for each beam direction. Conventional beamforming is performed on the original echo data of the horizontally split left and right subarrays and the original data of the vertically split upper and lower subarrays, and beam phase information is extracted. Based on the extracted beam phase information, the phase difference between the left and right subarrays and the phase difference between the upper and lower subarrays within the beam arrival time window in each beam direction are calculated to obtain the horizontal split phase difference curve and the vertical split phase difference curve. After evaluating the significance and goodness of the linear trends of the horizontal and vertical split phase difference curves, the final beam arrival time is determined, and the scatterer distance in each beam direction is calculated based on the final beam arrival time.

2. The three-dimensional forward-looking sonar imaging optimization method based on deconvolution split aperture according to claim 1, characterized in that, Perform conventional beamforming on the raw echo data of the complete array, including: in, The beam azimuth angle, For beam pitch angle, for time Beamforming results in the direction, For the array element in Echo data sampled at any time Frequency coefficients, represents the weighting coefficients of the array elements. The time delay parameters of the receiving array elements in each beam direction under far-field conditions are given. This is the receive array vector.

3. The three-dimensional forward-looking sonar imaging optimization method based on deconvolution split aperture according to claim 1, characterized in that, Deconvolution is performed on the results of conventional beamforming, including: in, The beam azimuth angle, For beam pitch angle, for time Beamforming results in the direction, and The first +1 times and the iteration The Richard-Lucy deconvolution beamforming results after next-generation calculation The point spread function is related to array parameters and signal frequency. Indicates the relevant operations, This represents the dot product operation.

4. The three-dimensional forward-looking sonar imaging optimization method based on deconvolution split aperture according to claim 1, characterized in that, Array splitting methods in the horizontal and vertical directions include: The array is divided into a left subarray and a right subarray along the horizontal direction, wherein the array centers of the two left and right subarrays are located on the same horizontal line, and the array elements may have overlapping parts; The array is divided into an upper subarray and a lower subarray along the vertical direction, wherein the centers of the two upper and lower subarrays are located on the same vertical line, and the array elements may have overlapping parts; The array splitting method is applicable to various planar arrays including circular arrays, rectangular arrays, and sparse arrays, and allows the array to be split into two or more subarrays in any direction.

5. The three-dimensional forward-looking sonar imaging optimization method based on deconvolution split aperture according to claim 1, characterized in that, Extract the beam phase information of each subarray beamforming result, including: in, , The left and right subarray vectors of the receiving array are split horizontally. and The upper and lower subarray vectors of the receiving array are split vertically. , , , The left, right, upper, and lower pieces are respectively in time Beamforming results in the direction, , , , The left, right, upper, and lower pieces are respectively in time Beam phase in the direction, The operation of taking the phase of a complex number.

6. The three-dimensional forward-looking sonar imaging optimization method based on deconvolution split aperture according to claim 1, characterized in that, Based on the amplitude-energy arrangement of the deconvolution beamforming results, the beam arrival time windows for each beam direction are estimated, including: Based on the deconvolution beamforming results of a complete planar array Amplitude energy distribution and amplitude threshold over time in each beam direction The time window during which the beam first continuously crosses the amplitude threshold is selected as the estimated beam arrival time window. .

7. The three-dimensional forward-looking sonar imaging optimization method based on deconvolution split aperture according to claim 1, characterized in that, Based on the extracted beam phase information, the phase differences between the left and right subarrays and the phase differences between the upper and lower subarrays within the beam arrival time window in each beam direction are calculated to obtain the horizontal split phase difference curve and the vertical split phase difference curve, including: Calculate the phase difference: in, For the beam arrival time window in each beam direction, , , , The left, right, upper, and lower pieces are respectively in time Beam phase in the direction, To estimate the horizontal subarray within the time window Horizontal split phase difference matrix in the beam direction To estimate the vertical subarray within the time window Vertical split phase difference matrix in the beam direction; Fitted phase difference curve: in, for The linear fitting results, for The linear fitting results, , , , These are the estimated values ​​during the fitting process. To estimate the time matrix of the echo arrival window, .

8. The three-dimensional forward-looking sonar imaging optimization method based on deconvolution split aperture according to claim 7, characterized in that, After evaluating the significance and goodness of the linear trends of the horizontal and vertical split phase difference curves, the final echo beam arrival time is determined, including: Calculate the determining parameters for linear fitting of the horizontal and vertical split phase differences respectively: in, , These are the coefficients of determination for the linear fitting of the horizontal and vertical split phase differences, respectively, used to evaluate the goodness of linear fit of the model. This represents the summation operation of matrix elements. This represents the operation of averaging matrix elements; By comparing the determination coefficients of the horizontal and vertical phase difference fits, the matrix with the more significant linear trend is selected as the result: in, For phase difference matrices with a more pronounced linear trend, As a threshold for the coefficient of determination, a coefficient of determination lower than [a certain threshold] is set. The phase difference matrix is ​​considered to be non-significantly linear; Final beam arrival time It is obtained directly from the fitted estimate: in This is a calculation that rounds down to the nearest integer.

9. The three-dimensional forward-looking sonar imaging optimization method based on deconvolution split aperture according to claim 8, characterized in that, The distances of the scatterers in each beam direction are calculated based on the final beam arrival time results, including: To improve the robustness of beam arrival time calculation under low signal-to-noise ratio conditions, data with obvious beam arrival time errors are processed. in, The beam arrival time after removing erroneous data, Results of deconvolution beamforming The time corresponding to the point of maximum energy; Based on the imaging principle of active sonar, we can conclude that: in, for Distance of the scatterer in the beam direction This is the speed of sound in water.

Images