A three-dimensional sonar target detection method

Abstract

The application discloses a three-dimensional sonar target detection method, and relates to the technical field of underwater acoustic signal processing. The method is characterized in that: firstly, the central beam and the non-central beam of overall imaging are divided; a detection method of time-dimension CFAR and beam-domain template matching is adopted for the central beam; a beam-domain CFAR detection method is adopted for the non-central beam; finally, the detection results of the central beam and the non-central beam are fused to obtain the target detection results of the whole scene. The technical scheme in the application processes the central beam and the non-central beam separately, can fully consider the interference characteristics of beams in different regions, realizes effective background suppression and false target filtering, and improves the accuracy of target detection.

Description

Technical Field

[0001] This invention relates to the technical field of underwater acoustic signal processing, and more specifically, to a three-dimensional sonar target detection method. Background Technology

[0002] The multi-channel echo signals received by 3D imaging sonar, after being processed by beamforming technology, yield time-series echoes varying over time at different beam azimuths. This allows for the estimation of target intensity and the generation of the original 3D sonar image. However, the original 3D image contains a large number of useless noise and interference points, severely impacting the display quality of the 3D imaging. Therefore, to highlight effective targets and remove noise and interference points, target detection processing is required on the original beam domain echo time series of the sonar.

[0003] Based on the characteristics of 3D imaging sonar, its target detection methods can be categorized into three types: 1) Fixed beam domain, performing CFAR (Constant False Alarm Rate) detection along the time dimension; 2) Fixed time dimension, performing CFAR detection along the beam domain; 3) Joint CFAR detection in both the beam domain and time dimension. These three methods achieve target detection from different dimensions of the original 3D image. None of these methods distinguish between the center beam and non-center beams; all beam signals are processed identically. However, due to circuit interference, the signal amplitude at the center beam of a sonar image is usually higher than that of other beams, and it often contains crosstalk spikes. This complex interference at the center beam causes the target detection performance of the aforementioned methods to degrade near the center beam, resulting in a large number of false target points remaining at the center beam position in the detected image. These false target points impair the clarity of the target outline, increase the difficulty of post-processing, and make the overall image appear blurry and cluttered, affecting the observer's intuitive judgment of the scene structure and reducing the accuracy of detection. Therefore, to address the problem of numerous false target points still existing at the center after filtering in 3D sonar imaging, it is urgent to propose a 3D sonar target detection algorithm that can independently process both the center and non-center beams to reduce the impact of circuit interference and improve detection accuracy. Summary of the Invention

[0004] The purpose of this invention is to propose a three-dimensional sonar target detection method based on independent processing of center and non-center beams, so as to solve the problem of a large number of false target points at the center of three-dimensional sonar imaging.

[0005] The technical solution of this invention is: to provide a three-dimensional sonar target detection method, the method comprising:

[0006] Step 1: Use three-dimensional sonar to acquire multi-channel echo signals from the target, and perform beamforming processing and mode extraction on the multi-channel echo signals to obtain the signal model after mode extraction.

[0007] Step 2: Based on the preset number of horizontal and vertical beams in the central region, divide the signal model after model taking into central and non-central regions.

[0008] Step 3: Perform target detection processing on the beam signal in the central region. Specifically, construct a beam amplitude template for the central region, use the beam amplitude template to perform beam domain template matching processing on the beam in the central region, and perform time-dimensional CFAR detection on the beam in the central region. Finally, fuse the template matching result with the time-dimensional CFAR detection result to obtain the target detection result in the central region.

[0009] Step 4: Perform beam domain CFAR detection processing on the beam signal in the non-center area to obtain the target detection result in the non-center area.

[0010] Step 5: Fuse the target detection results of the central region and non-central region to obtain the target detection results of the whole scene.

[0011] Furthermore, in step 1, the signal model after modulus extraction is represented as:

[0012]

[0013] In the formula, m is the horizontal number of the array elements, n is the vertical number of the array elements, M is the number of horizontal array elements of the uniform planar array, N is the number of vertical array elements of the uniform planar array, and d x d y These represent the element spacing in the horizontal and vertical directions, respectively. The azimuth of the target. The pitch angle of the target, m b The beam number in the horizontal direction, n b Beam numbering in the vertical direction, 0≤m b ≤M b -1, 0≤n b ≤N b -1, M b N represents the number of horizontal beams. b s represents the number of vertical beams. m,n (i r ) represents the i-th element signal of array element (m,n). r Each sample point value, α m,n Let f be the window function coefficient of the (m,n) array element. c denoted as the center frequency of the sonar signal, and c as the speed of sound in water.

[0014] Furthermore, step 2 specifically includes:

[0015] The number of horizontal and vertical beams in the central area are set to M respectively. bc N bc According to M bc and N bc , in the horizontal direction at [(M b -M bc ) / 2,(M b +M bc The range is [(N) / 2] and the vertical direction is within [(N) / 2]. b -N bc ) / 2,(N b +N bc The beam within the range of ) / 2] is divided into the central region beam, and the remaining beam is divided into the non-central region beam. The horizontal beam number of the central region is denoted as m. bc The vertical beam number of the central region is denoted as n. bc The horizontal beam number of the non-central area is denoted as The vertical beam number in the non-central region is denoted as Obtain the central region beam B(i) r ,m bc ,n bc ) and non-central area beam The partitioning result is expressed as follows:

[0016]

[0017] In the formula, R mc For m bc The set of allowed values, R nc For n bc The set of allowed values. for The set of allowed values. for The set of allowed values.

[0018] Furthermore, in step 2, set R mc ∈[(M b -M bc ) / 2,(M b +M bc ) / 2],

[0019] R nc ∈[(N b -N bc ) / 2,(N b +N bc ) / 2],

[0020] Furthermore, in step 3, constructing the central region beam amplitude template specifically includes:

[0021] First, simulate an ideal beam amplitude pattern when the target is centered. Then, use the center (M) b / 2,N b The beam amplitude at / 2) is normalized for all beam amplitudes in the ideal beam amplitude diagram, and then the beam amplitude of the central region is extracted to form the central region beam amplitude template, which is represented as:

[0022]

[0023] In the formula, B bt (i r ,m b ,n b B represents the ideal beam pattern in the simulation. bt (i r M b / 2,N b / 2) The simulated ideal beam pattern is located at the center (M) b / 2,N b Beam amplitude at / 2).

[0024] Furthermore, in step 3, beam domain template matching is performed on the central region beam using the central region beam amplitude template, specifically including:

[0025] Set the tolerance factor α for center beam amplitude comparison. bc Perform beam domain template matching to divide the central region beam B(i) r ,m bc ,n bc Amplitude normalization is performed on its center beam to obtain... Set the detection result of the center beam to 1, and compare it with the detection result of other center region beams. and like The detection result is 1 if the result is not found, and 0 otherwise. This yields the first detection mask for beam domain template matching, represented as:

[0026]

[0027] In the formula, (m bc ,n bc )=(M bc / 2,N bc / 2) represents the center beam, (m bc ,n bc )≠(M bc / 2,N bc / 2) represents other beams besides the center beam;

[0028] Set the horizontal neighborhood window length M' of the beam bc and vertical neighbor window length N' bc At the same time, set a threshold value for the number of target points in the neighborhood. Perform a second match based on the first detection mask D. c1 (i r ,m bc ,n bc ), Statistical center (M) bc / 2,N bc The number of target points with a result of 1 within the neighborhood of / 2) is considered if the number of target points is greater than the threshold value. The detection result of the center beam is set to 1, and the detection result of the other beam is set to 0, thus obtaining the second detection mask for beam domain template matching, represented as:

[0029] K np (i r )=sum{D c1 (i r ,m' bc ,n' bc )}

[0030]

[0031] In the formula, K np (i r D represents the number of target points with a detection result of 1 counted within the central neighborhood of the first detection mask. c1 (i r ,m' bc ,n' bc ) represents the point within the central neighborhood of the first detection mask, and m'bc represents the beam number in the horizontal direction within the neighborhood.

[0032] m' bc ∈[(M bc -M' bc ) / 2,(M bc +M' bc ) / 2],n' bc n' is the beam number in the vertical direction within the neighborhood. bc ∈[(N bc -N' bc ) / 2,(N bc +N' bc ) / 2];

[0033] The results of the two matching operations are merged to obtain the third detection mask for beam domain template matching, represented as:

[0034]

[0035] In the formula, D bc (i r ,m bc ,n bc ) is the third detection mask for beam domain template matching.

[0036] Furthermore, in step 3, time-dimensional CFAR detection is performed on the central region beam, specifically including:

[0037] Set the threshold factor α for time-dimensional CFAR detection. t For the central region beam, compared to B(i) r ,m bc ,n bc ) and α t Tt(i r ,m bc ,n bc ), Tt(i r ,m bc ,n bc ) is the B(i) estimated by the time-dimensional CFAR detection. r ,m bc ,n bc ) in i r The background noise value at point B(i) r ,m bc ,n bc )>α t Tt(i r ,m bc ,n bc If the result is 1, the detection result is 0; otherwise, the detection result is 1. This yields the fourth detection mask for the time-dimensional CFAR detection, represented as:

[0038]

[0039] In the formula, B(i) r ,m bc ,n bc () is the central region beam.

[0040] Furthermore, step 3 also includes:

[0041] For the central region beam, if the corresponding detection results in the third and fourth detection masks are both 1, then the fused detection result is 1; otherwise, the fused detection result is 0. The fused detection mask is then obtained, represented as follows:

[0042]

[0043] In the formula, D bt (i r,m bc ,n bc (This refers to the mask used for detection after merging.)

[0044] The amplitude value of the central region beam is set according to the fused detection mask. If the detection result corresponding to the fused detection mask is 1, the amplitude value of the central region beam is the original value; otherwise, the amplitude value of the central region beam is 0. The target detection result of the central region beam is obtained, represented as:

[0045]

[0046] In the formula, B d (i r ,m bc ,n bc () represents the target detection results of the central region beam.

[0047] Furthermore, step 4 specifically includes:

[0048] Set the threshold factor α for beam-domain CFAR detection. b For non-central region beams, in contrast and like If the detection result is 1, the corresponding beam amplitude value is retained as its original value; otherwise, if the detection result is 0, the corresponding beam amplitude value is set to 0. For beam domain CFAR detection estimation exist The background noise value at a given location is used to obtain the target detection result of the beam in the non-center region, which is expressed as:

[0049]

[0050] In the formula, It is a non-central region beam.

[0051] Furthermore, the target detection results for the entire scene in step 5 are represented as follows:

[0052]

[0053] In the formula, B d (i r ,m b ,n b () represents the target detection results for the entire scene.

[0054] The beneficial effects of this invention are:

[0055] The technical solution of this invention proposes a framework for independent detection of the center and non-center regions in 3D sonar imaging. For the beam signal in the center region, CFAR detection is first performed in the time dimension. Taking advantage of the high signal-to-noise ratio of effective targets in this region, smooth background interference is removed, and strong target points are retained. Then, beam domain template construction and matching steps are introduced in parallel during the detection process. Taking advantage of the similarity between the normalized amplitude of false target points and the beam domain template, strong target points caused by instantaneous strong circuit interference are selectively filtered out through template matching. Finally, the processing results of time-dimensional CFAR detection and beam domain template matching are fused to obtain a sonar image of the center region with a clear background and no circuit interference. For the beam signal in the non-center region, background interference is directly filtered out by the beam domain CFAR detection method to obtain the sonar image of the non-center region. Finally, the target detection results of the center region and the non-center region are fused to obtain the target detection result of the whole scene.

[0056] The technical solution of this invention can process the central and non-central region beams separately, making full use of the interference characteristics of different region beams to effectively suppress background and filter out false targets, thus improving the accuracy of target detection. For the central region beam with high signal amplitude, time-dimensional CFAR plus template matching can accurately identify and filter out false target points caused by circuit interference, improving the quality of sonar images. Compared with existing methods, the technical solution of this invention can better cope with complex electromagnetic interference and background noise, maintaining high detection accuracy and stability under various interference environments, and ultimately obtaining a three-dimensional sonar target detection image with significant false point suppression, high fidelity of real targets, and high overall imaging clarity. Attached Figure Description

[0057] The advantages of the above and / or additional aspects of the present invention will become apparent and readily understood in the description of the embodiments taken in conjunction with the following drawings, wherein:

[0058] Figure 1 This is a schematic flowchart of a three-dimensional sonar target detection method according to an embodiment of the present invention;

[0059] Figure 2 This is a full-scene detection result image obtained using a full-beam domain CFAR detection method according to an embodiment of the present invention;

[0060] Figure 3(a) is a graph of the time-dimensional CFAR detection results in the central region according to an embodiment of the present invention;

[0061] Figure 3(b) is a diagram of the beam domain template matching detection results in the central region according to an embodiment of the present invention;

[0062] Figure 3(c) is a detection result diagram of the fusion of time-dimensional CFAR detection in the central region and beam domain template matching according to an embodiment of the present invention;

[0063] Figure 4 This is a full-scene detection result image obtained using the method of the present invention according to an embodiment of the present invention. Detailed Implementation

[0064] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments of the present invention and the features thereof can be combined with each other.

[0065] In the following description, many specific details are set forth in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0066] Example 1:

[0067] like Figure 1 As shown, this embodiment provides a three-dimensional sonar target detection method, which includes:

[0068] Step 1: Use three-dimensional sonar to acquire multi-channel echo signals from the target, and perform beamforming processing and model extraction on the multi-channel echo signals to obtain the modeled signal.

[0069] The signal model after modulus extraction is represented as follows:

[0070]

[0071] In the formula, m is the horizontal number of the array elements, n is the vertical number of the array elements, M is the number of horizontal array elements of the uniform planar array, N is the number of vertical array elements of the uniform planar array, and d x d y These represent the element spacing in the horizontal and vertical directions, respectively. The azimuth (left and right deflection) of the target. The pitch angle (up and down) of the target, m b The beam number in the horizontal direction, n b Beam numbering in the vertical direction, 0≤m b ≤M b -1, 0≤n b ≤N b -1, M b N represents the number of horizontal beams. b s represents the number of vertical beams. m,n (i r) represents the i-th element signal of array element (m,n). r There are 1 sampling point values, 0≤i r ≤I r -1, I r α represents the total number of time sampling points. m,n Let f be the window function coefficient of the (m,n) array element. c Let be the center frequency of the sonar signal, c be the speed of sound in water, and ||·|| be the modulus of the signal.

[0072] It should be noted that beamforming is a common method for processing multi-channel echo signals, and will not be elaborated upon here. After the multi-channel echo signals acquired by the three-dimensional sonar are processed by beamforming technology, the echo time series of different beam orientations are obtained. These sequence signals are in complex form. Modulus taking is the process of converting the complex form signals into real amplitude sequences, which facilitates subsequent detection, imaging, and display.

[0073] Step 2: Based on the preset number of horizontal and vertical beams in the central region, divide the signal model after model taking into central and non-central regions.

[0074] The number of horizontal and vertical beams in the central area are set to M respectively. bc N bc According to M bc and N bc , in the horizontal direction at [(M b -M bc ) / 2,(M b +M bc The range is [(N) / 2] and the vertical direction is within [(N) / 2]. b -N bc ) / 2,(N b +N bc The beam within the range of ) / 2] is divided into the central region beam, and the remaining beam is divided into the non-central region beam. The horizontal beam number of the central region is denoted as m. bc The vertical beam number of the central region is denoted as n. bc The horizontal beam number of the non-central area is denoted as The vertical beam number in the non-central region is denoted as Obtain the central region beam B(i) r ,m bc ,n bc ) and non-central area beam The partitioning result is expressed as follows:

[0075]

[0076] In the formula, R mc For m bcThe set of allowed values, R nc For n bc The set of allowed values, R mc ∈[(M b -M bc ) / 2,(M b +M bc ) / 2],R nc ∈[(N b -N bc ) / 2,(N b +N bc ) / 2], for The set of allowed values. for The set of allowed values.

[0077] It should be noted that during the acquisition and processing of multi-channel echo signals, 3D sonar is affected by circuit interference (circuit interference typically refers to interference caused by electrical noise, signal crosstalk, or other electromagnetic phenomena in electronic devices or systems). This causes the beam amplitude in the central region of the sonar image to be higher than that in the non-central region. If full-scene beam domain CFAR detection is directly used, a large number of false target points in the central region will be preserved, while the beam in the non-central region will remain unaffected. Figure 2 As shown, in the full-beam domain CFAR detection results, there are a large number of false target points in the central region, which increases the target detection error. Therefore, in order to eliminate false target points in the central region and minimize the computational load, this invention processes the target detection of the central region and non-central region beams separately.

[0078] Step 3: Perform target detection processing on the beam signal in the central region. Specifically, construct a beam amplitude template for the central region, use this template to perform beam domain template matching processing on the central region beam, and simultaneously perform time-dimensional CFAR detection on the central region beam. Finally, fuse the template matching result with the time-dimensional CFAR detection result to obtain the target detection result for the central region. This includes the following steps:

[0079] Step 3.1: First, simulate an ideal beam amplitude diagram when the target is centered, then use the center (M) b / 2,N b The beam amplitude at / 2) is normalized for all beam amplitudes in the ideal beam amplitude diagram, and then the beam amplitude of the central region is extracted to form the central region beam amplitude template, which is represented as:

[0080]

[0081] In the formula, B bt (i r ,m b ,n b B represents the ideal beam pattern in the simulation. bt (i r M b / 2,N b / 2) The simulated ideal beam pattern is located at the center (M) b / 2,N b Beam amplitude at / 2).

[0082] In this embodiment, normalizing the beam amplitude refers to adjusting the values ​​of the ideal beam amplitude map to the range of [0,1], truncating the ideal beam amplitude map, and retaining only the central region of the map. The resulting central region template, which is the normalized and truncated beam amplitude map, can be used as a reference for subsequent comparison of interference beam amplitudes.

[0083] Step 3.2 involves performing beam domain template matching on the central region beam using the central region beam amplitude template. This includes the following steps:

[0084] Set the tolerance factor α for center beam amplitude comparison. bc Beam domain template matching is performed to divide the central region beam B(i) from the actual signal. r ,m bc ,n bc The amplitude of the beam is normalized towards its center (i.e., using the normalization method in step 3.1) to obtain... Set the detection result of the center beam to 1. For other center region beams outside the center beam, compare... and like The detection result is 1 if the result is not found, and 0 otherwise. This yields the first detection mask for beam domain template matching, represented as:

[0085]

[0086] In the formula, (m bc ,n bc )=(M bc / 2,N bc / 2) represents the center beam, (m bc ,n bc )≠(M bc / 2,N bc / 2) represents other beams besides the center beam, D c1 (i r ,m bc ,nbc ) is the first detection mask for beam domain template matching; tolerance factor α bc The empirical value is 0.1.

[0087] Set the horizontal neighborhood window length M' of the beam bc and vertical neighbor window length N' bc At the same time, set a threshold value for the number of target points in the neighborhood. Perform a second match based on the first detection mask D c1 (i r ,m bc ,n bc ), Statistical center (M) bc / 2,N bc The number of target points with a result of 1 within the neighborhood of / 2) is considered if the number of target points is greater than the threshold value. The detection result of the center beam is set to 1, and the detection result of the other beam is set to 0, thus obtaining the second detection mask for beam domain template matching, represented as:

[0088] K np (i r )=sum{D c1 (i r ,m' bc ,n' bc )}

[0089]

[0090] In the formula, K np (i r D represents the number of target points with a detection result of 1 counted within the central neighborhood of the first detection mask. c1 (i r ,m' bc ,n' bc ) represents the point within the central neighborhood of the first detection mask, and m'bc represents the beam number in the horizontal direction within the neighborhood.

[0091] m' bc ∈[(M bc -M' bc ) / 2,(M bc +M' bc ) / 2],n' bc n' is the beam number in the vertical direction within the neighborhood. bc ∈[(N bc -N' bc ) / 2,(N bc +N' bc ) / 2], sum{·} represents the target point count.

[0092] The results of the two matching operations are merged to obtain the third detection mask for beam domain template matching, represented as:

[0093]

[0094] In the formula, D bc (i r ,m bc ,n bc ) is the third detection mask for beam domain template matching.

[0095] Step 3.3: Perform time-dimensional CFAR detection on the central region beam. Specifically, set the threshold factor α for time-dimensional CFAR detection. t For the central region beam, compared to B(i) r ,m bc ,n bc ) and α t Tt(i r ,m bc ,n bc ), Tt(i r ,m bc ,n bc ) is the B(i) estimated by the time-dimensional CFAR detection. r ,m bc ,n bc ) in i r The background noise value at point B(i) r ,m bc ,n bc )>α t Tt(i r ,m bc ,n bc If the result is 1, the detection result is 0; otherwise, the detection result is 1. This yields the fourth detection mask for the time-dimensional CFAR detection, represented as:

[0096]

[0097] In the formula, B(i) r ,m bc ,n bc () is the central region beam, with a threshold factor α t The empirical value is 4.

[0098] In this embodiment, the large threshold factor α t There will be a lower false alarm rate, but at the same time the false negative rate of real targets will increase. Conversely, a small threshold factor α t It has a high false alarm rate, but a low false negative rate for real targets. Here, the threshold factor α is... t When the value of is 4, the false alarm rate and the false negative rate will show a balanced relationship, which can achieve the best detection effect.

[0099] Step 3.4 involves fusing the third detection mask from beam domain template matching with the fourth detection mask from time-dimension CFAR detection to obtain the target detection result for the central region beam. This includes the following steps:

[0100] For the central region beam, if the corresponding detection results in the third and fourth detection masks are both 1, then the fused detection result is 1; otherwise, the fused detection result is 0. The fused detection mask is then obtained, represented as follows:

[0101]

[0102] In the formula, D bt (i r ,m bc ,n bc (This refers to the mask used for detection after merging.)

[0103] The amplitude value of the central region beam is set according to the fused detection mask. If the detection result corresponding to the fused detection mask is 1, the amplitude value of the central region beam is the original value; otherwise, the amplitude value of the central region beam is 0. The target detection result of the central region beam is obtained, represented as:

[0104]

[0105] In the formula, B d (i r ,m bc ,n bc () represents the target detection results of the central region beam.

[0106] Step 4: Perform beam domain CFAR detection processing on the beam signal in the non-center region to obtain the target detection result in the non-center region; specifically, it includes the following steps:

[0107] Set the threshold factor α for beam-domain CFAR detection. b For non-central region beams, in contrast and like If the detection result is 1, the corresponding beam amplitude value is retained as its original value; otherwise, if the detection result is 0, the corresponding beam amplitude value is set to 0. For beam domain CFAR detection estimation exist The background noise value at a given location is used to obtain the target detection result of the beam in the non-center region, which is expressed as:

[0108]

[0109] In the formula, For non-central region beams, threshold factor α b The empirical value is 4.

[0110] Step 5: Fuse the target detection results of the central region and non-central region to obtain the target detection results for the entire scene, as follows:

[0111]

[0112] In the formula, B d (i r ,m b ,n b () represents the target detection results for the entire scene.

[0113] In this embodiment, a framework for independent detection of the center and non-center regions in 3D sonar imaging is proposed. The process of constructing a template for the center region and matching the signal beam of the center region using the constructed template is added. Ultimately, the goal of eliminating a large number of false target points caused by circuit interference is achieved, realizing target detection in the entire 3D imaging scene.

[0114] Specifically, due to circuit interference, the beam amplitude in the central region is higher than that in the non-central region. Directly using full-scene beam domain CFAR detection will result in a large number of false target points being retained in the central region.

[0115] This application considers the high signal-to-noise ratio characteristics of effective targets in the time dimension and performs time-dimensional CFAR detection on the beam signal in the central region to remove smooth background interference and retain prominent candidate target points. However, due to the strong instantaneous interference from the circuit, the retained prominent candidate target points will still contain a large number of false target points. These false target points cannot be removed by time-dimensional CFAR detection. However, research has found that the normalized amplitude of false target points caused by the instantaneous strong interference from the circuit is similar to that of the constructed template. Utilizing this characteristic, the instantaneous strong interference from the circuit can be removed by center beam template matching. To eliminate false target points, a primary detection mask is used to compare the normalized amplitude of the signal with the constructed template. If the amplitude difference is less than a set tolerance factor, the signal is considered to lack support from real target points and is identified as a false target point caused by circuit interference, thus eliminating it. A secondary detection mask is then added to the center beam. Utilizing the point density characteristics and the fact that real target points cannot exist in isolation, the effective target points in the neighborhood of the target point at the center are counted. If the number is less than a set threshold, it is also identified as a false target point caused by circuit interference and eliminated. Therefore, while center beam template matching can eliminate instantaneous strong circuit interference, it is less effective at removing time-varying noise signals, leaving a significant amount of background interference in the sonar image. However, by fusing time-dimensional CFAR detection with center beam domain template matching, the advantages of both methods can be fully utilized to eliminate a large number of false target points in the central region caused by circuit interference, ultimately resulting in a clean sonar image free from circuit interference.

[0116] Example 2:

[0117] The proposed three-dimensional sonar target detection method was validated through sea trials. The sea data source was long-range target detection imaging from a three-dimensional sonar device. The original full-beam CFAR detection imaging results are as follows: Figure 2 As shown in Figure 3, the results of temporal CFAR detection within the central area are shown in Figure 3(a), the results of beam domain template matching are shown in Figure 3(b), and the results of fusion detection of temporal CFAR and beam domain template matching are shown in Figure 3(c). The final imaging result after processing by the method of this application is shown in Figure 3(c). Figure 4 As shown.

[0118] The original beam domain CFAR detection full-scene imaging results are as follows: Figure 2As shown, the result forms a false "pipe" in the central region, which is caused by circuit interference. The central portion is cropped, and time-dimensional CFAR detection and beam domain template matching are performed, as shown in Figures 3(a) and 3(b), respectively. The results show that both methods can remove some of the false target points in the "pipe". The two methods are then fused, and the final detection result for the central region is shown in Figure 3(c). A large number of false points in the center are removed, and the selected target is continuous and unbroken in the center, indicating that the real target points are preserved. The detection results from the center and non-center are then fused to obtain the full-scene imaging result processed by the proposed method, as shown in Figure 3(c). Figure 4 As shown. Comparison Figure 2 and Figure 4 The detection results show that the method of this application can effectively eliminate a large number of false target points located in the middle of the sonar image, and realize effective target point cloud detection for three-dimensional imaging sonar.

[0119] The steps in this application can be rearranged, combined, or deleted according to actual needs.

[0120] The units in the device of this application can be merged, divided, and deleted according to actual needs.

[0121] Although this application has been disclosed in detail with reference to the accompanying drawings, it should be understood that these descriptions are merely exemplary and not intended to limit the application of this application. The scope of protection of this application is defined by the appended claims and may include various variations, modifications, and equivalents of the invention without departing from the scope and spirit of this application.

Claims

1. A three-dimensional sonar target detection method, characterized in that, The three-dimensional sonar target detection method includes: Step 1: Use three-dimensional sonar to acquire multi-channel echo signals from the target, and perform beamforming processing and mode extraction on the multi-channel echo signals to obtain the signal model after mode extraction. Step 2: Based on the preset number of horizontal and vertical beams in the central region, divide the signal model after model taking into central and non-central regions. Step 3: Perform target detection processing on the beam signal in the central region. Specifically, construct a beam amplitude template for the central region, use the beam amplitude template to perform beam domain template matching processing on the beam in the central region, and perform time-dimensional CFAR detection on the beam in the central region. Finally, fuse the template matching result with the time-dimensional CFAR detection result to obtain the target detection result in the central region. The construction of the central region beam amplitude template includes: first simulating an ideal beam amplitude pattern when the target is located in the center, and then using the center (M) b / 2,N b The beam amplitude at / 2) is normalized for all beam amplitudes in the ideal beam amplitude diagram, and then the beam amplitude of the central region is extracted to form the central region beam amplitude template, M. b N represents the number of horizontal beams. b This refers to the number of vertical beams; Step 4: Perform beam domain CFAR detection processing on the beam signal in the non-center area to obtain the target detection result in the non-center area. Step 5: Fuse the target detection results of the central region and non-central region to obtain the target detection results of the whole scene.

2. The three-dimensional sonar target detection method as described in claim 1, characterized in that, In step 1, the signal model after modulus extraction is represented as follows: ； In the formula, m is the horizontal number of the array elements, n is the vertical number of the array elements, M is the number of horizontal array elements of the uniform planar array, N is the number of vertical array elements of the uniform planar array, and d x d y These represent the element spacing in the horizontal and vertical directions, respectively. The azimuth of the target. The pitch angle of the target, m b The beam number in the horizontal direction, n b Beam numbering in the vertical direction, 0≤m b ≤M b -1, 0≤n b ≤N b -1,s m,n (i r ) represents the i-th element signal of array element (m,n). r Each sample point value, Let f be the window function coefficient of the (m,n) array element. c denoted as the center frequency of the sonar signal, and c as the speed of sound in water.

3. The three-dimensional sonar target detection method as described in claim 2, characterized in that, Step 2 specifically includes: The number of horizontal and vertical beams in the central area are set to M respectively. bc N bc According to M bc and N bc , in the horizontal direction at [(M b -M bc ) / 2,(M b +M bc The range is [(N) / 2] and the vertical direction is within [(N) / 2]. b -N bc ) / 2,(N b +N bc The beam within the range of ) / 2] is divided into the central region beam, and the remaining beam is divided into the non-central region beam. The horizontal beam number of the central region is denoted as m. bc The vertical beam number of the central region is denoted as n. bc The horizontal beam number of the non-central area is denoted as The vertical beam number in the non-central region is denoted as The central region beam B(i) is obtained. r ,m bc ,n bc ) and non-central region beam B(i r , bc , bc The partitioning result is expressed as: ； In the formula, R mc For m bc The set of allowed values, R nc For n bc The set of allowed values. for The set of allowed values. for The set of allowed values.

4. The three-dimensional sonar target detection method as described in claim 3, characterized in that, In step 2, set R mc ∈[(M b -M bc ) / 2,(M b +M bc ) / 2],R nc ∈[(N b -N bc ) / 2,(N b +N bc ) / 2], ∈[0,(M b -M bc ) / 2)∪((M b +M bc ) / 2,M b -1], ∈[0,(N b -N bc ) / 2)∪((N b +N bc ) / 2,N b -1].

5. The three-dimensional sonar target detection method as described in claim 3, characterized in that, In step 3, the beam amplitude template of the central region is represented as follows: ； In the formula, B bt (i r ,m b ,n b B represents the ideal beam pattern in the simulation. bt (i r M b / 2,N b / 2) The simulated ideal beam pattern is located at the center (M) b / 2,N b Beam amplitude at / 2).

6. The three-dimensional sonar target detection method as described in claim 5, characterized in that, In step 3, beam domain template matching is performed on the central region beam using the central region beam amplitude template, specifically including: Set the tolerance factor for center beam amplitude comparison. bc Perform beam domain template matching to divide the central region beam B(i) r ,m bc ,n bc Amplitude normalization is performed on its center beam to obtain... (i r ,m bc ,n bc Set the detection result of the center beam to 1, and compare it with the detection result of other center region beams. (i r ,m bc ,n bc )and bt (i r ,m bc ,n bc )+ bc ,like (i r ,m bc ,n bc )> bt (i r ,m bc ,n bc )+ bc If the result is 1, the detection result is 0; otherwise, the detection result is 1. This yields the first detection mask for beam domain template matching, represented as: ； In the formula, (m bc ,n bc )=(M bc / 2,N bc / 2) represents the center beam, (m bc ,n bc )≠(M bc / 2,N bc / 2) represents other beams besides the center beam; Set the horizontal neighborhood window length M' of the beam bc and vertical neighbor window length N' bc At the same time, set a threshold value for the number of target points in the neighborhood. np Perform a second match based on the first detection mask D. c1 (i r ,m bc ,n bc ), Statistical center (M) bc / 2,N bc The number of target points with a result of 1 within the neighborhood of / 2) is considered if the number of target points is greater than the threshold value. np If the detection result of the center beam is positive, then the detection result is set to 1; otherwise, the detection result is set to 0. This yields the second detection mask for beam domain template matching, represented as: ； ； In the formula, K np (i r D represents the number of target points with a detection result of 1 counted within the central neighborhood of the first detection mask. c1 (i r ,m' bc ,n' bc ) represents the point within the central neighborhood of the first detection mask, m' bc The horizontal beam number in the neighborhood, m' bc ∈[(M bc -M' bc ) / 2,(M bc +M' bc ) / 2],n' bc n' is the beam number in the vertical direction within the neighborhood. bc ∈[(N bc -N' bc ) / 2,(N bc +N' bc ) / 2]; The results of the two matching operations are merged to obtain the third detection mask for beam domain template matching, represented as: ； In the formula, D bc (i r ,m bc ,n bc ) is the third detection mask for beam domain template matching.

7. The three-dimensional sonar target detection method as described in claim 6, characterized in that, Step 3 involves performing time-dimensional CFAR detection on the central region beam, specifically including: Set the threshold factor for time-dimensional CFAR detection. t For the central region beam, compared to B(i) r ,m bc ,n bc )and t Tt(i r ,m bc ,n bc ), Tt(i r ,m bc ,n bc ) is the B(i) estimated by the time-dimensional CFAR detection. r ,m bc ,n bc ) in i r The background noise value at point B(i) r ,m bc ,n bc )> t Tt(i r ,m bc ,n bc If the result is 1, the detection result is 0; otherwise, the detection result is 1. This yields the fourth detection mask for the time-dimensional CFAR detection, represented as: ； In the formula, B(i) r ,m bc ,n bc () is the central region beam.

8. The three-dimensional sonar target detection method as described in claim 7, characterized in that, Step 3 also includes: For the central region beam, if the corresponding detection results in the third and fourth detection masks are both 1, then the fused detection result is 1; otherwise, the fused detection result is 0. The fused detection mask is then obtained, represented as follows: ； In the formula, D bt (i r ,m bc ,n bc (This refers to the mask used for detection after merging.) The amplitude value of the central region beam is set according to the fused detection mask. If the detection result corresponding to the fused detection mask is 1, the amplitude value of the central region beam is the original value; otherwise, the amplitude value of the central region beam is 0. The target detection result of the central region beam is obtained, represented as: ； In the formula, B d (i r ,m bc ,n bc () represents the target detection results of the central region beam.

9. The three-dimensional sonar target detection method as described in claim 8, characterized in that, Step 4 specifically includes: Set the threshold factor for beam-domain CFAR detection. b For non-central region beams, compared to B(i) r , bc , bc )and b T b (i r , bc , bc If B(i) r , bc , bc )> b T b (i r , bc , bc If the detection result is 1, the corresponding beam amplitude value is retained as its original value; otherwise, the detection result is 0, and the corresponding beam amplitude value is set to 0. Where, T b (i r , bc , bc ) is the estimated B(i) for beam-domain CFAR detection. r , bc , bc )exist( bc , bc The background noise value at point () is used to obtain the target detection result of the non-center region beam, which is expressed as: ； In the formula, B(i) r , bc , bc () is a non-central region beam.

10. The three-dimensional sonar target detection method as described in claim 9, characterized in that, The target detection results for the entire scene in step 5 are represented as follows: ； In the formula, B d (i r ,m b ,n b () represents the target detection results for the entire scene.

Images