Sonar long-distance target detection method, device and equipment and readable storage medium

Abstract

The invention provides a sonar long-distance target detection method, device and equipment and a readable storage medium, and the method comprises the steps: intercepting a plurality of detection segments from a matched filtering result of a beam domain signal, the beam domain signal is obtained through the preprocessing of a sonar echo signal, the preprocessing comprises the beam forming, the lengths of all detection segments are equal, and the detection segments are equal in length; the interval between the initial positions of the two detection segments with continuous serial numbers is smaller than or equal to the length of the detection segments; performing kernel density estimation on each detection segment according to a preset kernel function to obtain a probability density function; for each detection segment, the relative entropy of the probability density function of the current detection segment and the relative entropy of the probability density functions of other detection segments are summed to obtain an overall difference value; and judging whether a target exists in the distance range corresponding to the detection segment or not according to the overall difference value of the detection segment. According to the invention, the target echo output signal-to-noise ratio can be improved for a long-distance target, and accurate detection is realized.

Description

Technical Field

[0001] The present application relates to the technical field of target detection, and in particular to a sonar long-distance target detection method, device, equipment and readable storage medium. Background Art

[0002] With the continuous development of fields such as ocean exploration and national defense security, active sonar plays a vital role in target detection. At present, in active sonar target detection technology, the traditional method is mainly based on matched filtering peak amplitude for target detection. In close-range target detection, this traditional method can usually achieve good results. By matching the transmitted signal and the received echo, the target echo signal can be effectively enhanced, and the presence of the target can be determined by comparing the matched filtering peak amplitude with the decision threshold.

[0003] However, when it comes to detecting distant targets, due to long-distance propagation, the sound waves emitted by active sonar will experience greater attenuation during propagation in the water. At the same time, the background noise in the marine environment will accumulate as the propagation distance increases, resulting in a low signal-to-noise ratio for the echo input of distant targets. In this case, traditional methods cannot effectively detect target signals from echoes with low signal-to-noise ratios, and the target may be submerged by noise, misjudged, or simply cannot be detected, which greatly limits the application effect and reliability of active sonar in long-distance target detection and cannot accurately detect distant targets. Summary of the invention

[0004] The present application provides a sonar long-distance target detection method, device, equipment and readable storage medium, which can solve the technical problem that the prior art cannot accurately detect long-distance targets.

[0005] In a first aspect, an embodiment of the present application provides a sonar long-distance target detection method, the sonar long-distance target detection method comprising: Extracting multiple detection segments from a matched filtering result of a beam domain signal, wherein the beam domain signal is obtained by preprocessing the sonar echo signal, the preprocessing includes beam forming, the length of each detection segment is equal, and the interval between the starting positions of two detection segments with consecutive serial numbers is less than or equal to the length of the detection segment; Perform kernel density estimation on each detection segment according to the preset kernel function to obtain a probability density function; For each detection segment, the relative entropy of the probability density function of the current detection segment and the probability density function of other detection segments is summed to obtain the overall difference value. The calculation formula of the overall difference value is:

[0006] Among them, K s(i) is the overall difference value of the i-th detection segment, KL() is the KL divergence calculation formula, PH(i) is the probability density function of the i-th detection segment, and PH(j) is the probability density function of the j-th detection segment; The overall difference value of the detection segment is used to determine whether there is a target within the distance range corresponding to the detection segment.

[0007] Furthermore, in one embodiment, the step of judging whether there is a target within the distance range corresponding to the detection segment according to the overall difference value of the detection segment includes: For each detection segment, the root mean square of the overall difference values ​​of multiple background segments corresponding to the current detection segment is calculated to obtain the background intensity, and whether there is a target within the distance range corresponding to the current detection segment is determined based on the background intensity and the overall difference value, wherein the absolute value of the serial number difference between the current detection segment and each corresponding background segment is greater than or equal to the protection length.

[0008] Furthermore, in one embodiment, the absolute value of the difference between the sequence numbers of the current detection segment and each of its corresponding background segments is less than or equal to the background length.

[0009] Furthermore, in one embodiment, the step of judging whether there is a target within the distance range corresponding to the current detection segment according to the background intensity and the overall difference value includes: If the decibel value of the overall difference value of the current detection segment to the background intensity is greater than or equal to the preset threshold, it is determined that there is a target within the distance range corresponding to the current detection segment.

[0010] Further, in one embodiment, when the horizontal coordinate of the matched filtering result is time and the vertical coordinate is amplitude, the distance range corresponding to the detection segment is obtained by converting the time range of the detection segment in the matched filtering result and the speed of sound in water; When the horizontal coordinate of the matched filtering result is distance and the vertical coordinate is amplitude, the distance range corresponding to the detection segment is the distance range of the detection segment in the matched filtering result.

[0011] Furthermore, in one embodiment, the step of extracting a plurality of detection segments from a matched filtering result of a beam domain signal comprises: A matched filtering result of a beam domain signal is subjected to sliding window interception to obtain multiple detection segments, wherein the window shape intercepted by the sliding window is a rectangular window, and the sliding step length is less than or equal to the rectangular window length.

[0012] Furthermore, in one embodiment, the preset kernel function is one of a uniform kernel function, a triangular kernel function, a biweight kernel function and a triweight kernel function.

[0013] In a second aspect, an embodiment of the present application further provides a sonar long-distance target detection device, the sonar long-distance target detection device comprising: A segmentation interception module is used to intercept multiple detection segments from a matched filtering result of a beam domain signal, wherein the beam domain signal is obtained by preprocessing the sonar echo signal, and the preprocessing includes beam forming, the length of each detection segment is equal, and the interval between the starting positions of two detection segments with consecutive serial numbers is less than or equal to the length of the detection segment; A kernel density estimation module is used to perform kernel density estimation on each detection segment according to a preset kernel function to obtain a probability density function; The difference calculation module is used to sum the relative entropy of the probability density function of the current detection segment and the probability density function of other detection segments for each detection segment to obtain the overall difference value. The calculation formula of the overall difference value is:

[0014] Among them, K s (i) is the overall difference value of the i-th detection segment, KL() is the KL divergence calculation formula, PH(i) is the probability density function of the i-th detection segment, and PH(j) is the probability density function of the j-th detection segment; The judgment module is used to judge whether there is a target within the distance range corresponding to the detection segment according to the overall difference value of the detection segment.

[0015] In a third aspect, an embodiment of the present application also provides a sonar long-distance target detection device, which includes a processor, a memory, and a sonar long-distance target detection program stored in the memory and executable by the processor, wherein when the sonar long-distance target detection program is executed by the processor, the steps of the above-mentioned sonar long-distance target detection method are implemented.

[0016] In a fourth aspect, an embodiment of the present application further provides a readable storage medium, on which a sonar long-distance target detection program is stored, wherein when the sonar long-distance target detection program is executed by a processor, the steps of the above-mentioned sonar long-distance target detection method are implemented.

[0017] In this application, by performing segmented kernel density estimation on the matched filtering results of the active sonar echo, the probability density function of each detection segment is obtained, and the relative entropy is used to characterize the difference in the probability density function between one detection segment and another detection segment. The relative entropy corresponding to one detection segment is summed to obtain the overall difference value to characterize the difference in the probability density function between the echo of the corresponding distance range and the overall echo. According to the overall difference value of the detection segment, it is judged whether there is a target within the distance range corresponding to the detection segment. Since the probability density function of the long-distance background noise under deep-sea conditions has a strong time stationarity, the appearance of the target echo will cause a certain disturbance to the echo probability density, and the overall difference value can amplify this disturbance. Therefore, through this application, the target echo output signal-to-noise ratio can be improved for long-distance targets to achieve accurate detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is a flow chart of a sonar long-distance target detection method in one embodiment of the present application; Figure 2 It is a schematic diagram of the matched filtering result of the measured active sonar background noise; Figure 3 for Figure 2 The schematic diagram of the probability density function of the matched filtering result in the range of 0-50 km is shown; Figure 4 for Figure 2 The schematic diagram of the probability density function of the matched filtering result at 50-100km is shown; Figure 5 for Figure 2 The schematic diagram of the probability density function of the matched filtering result at 100-150km is shown; Figure 6 The figure is a schematic diagram of a three-dimensional model of a typical underwater target; Figure 7 for Figure 6 Schematic diagram of the impulse response function of the underwater target shown; Figure 8 for Figure 6 Schematic diagram of the two-way impulse response function of the fourth shadow channel of the underwater target shown; Fig. 9 for Figure 6 Schematic diagram of the complete echo of the underwater target shown; Fig.10 for Fig. 9 A schematic diagram of the probability density function of the matched filtering result of the complete echo shown; Fig.11 It is a schematic diagram of the matched filtering result of the measured background noise superimposed on the simulated target echo; Fig.12 This method is for Fig.11A schematic diagram of the overall difference values ​​of different detection segments calculated from the matched filtering results shown; Fig.13 This is a comparison chart of the detection results of the traditional method and this method when the echo input signal-to-noise ratio is -20dB and the detection threshold is 10dB; Fig.14 The ROC curve comparison diagram of the traditional method and the method under different echo input signal-to-noise ratios; Fig.15 This is a schematic diagram of the functional modules of a sonar long-distance target detection device in one embodiment of the present application; Fig.16 This is a schematic diagram of the hardware structure of the sonar long-distance target detection device involved in the embodiment of the present application. DETAILED DESCRIPTION

[0019] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0020] In order to make the objectives, technical solutions and advantages of the present application clearer, the implementation methods of the present application will be further described in detail below with reference to the accompanying drawings.

[0021] In a first aspect, an embodiment of the present application provides a sonar long-distance target detection method.

[0022] Figure 1 A schematic flow chart of a sonar long-distance target detection method in an embodiment of the present application is shown.

[0023] Reference Figure 1 In one embodiment, the sonar long-distance target detection method includes the following steps: S11. Extract multiple detection segments from a matched filtering result of a beam domain signal, wherein the beam domain signal is obtained by preprocessing the sonar echo signal, the preprocessing includes beam forming, the length of each detection segment is equal, and the starting position interval of two detection segments with consecutive serial numbers is less than or equal to the length of the detection segment.

[0024] Specifically, the array domain data received by multiple sensors needs to be beamformed to obtain multiple beam domain signals. Different beam domain signals correspond to echoes at different angles. Before beamforming, frequency band shifting, low-pass filtering, and downsampling are usually performed to reduce the amount of calculation. These operations and beamforming are collectively referred to as preprocessing. For each beam domain signal, matched filtering is performed according to the corresponding parameters of the transmitted signal (bandwidth, pulse width, signal form, etc.) to enhance the target characteristics.

[0025] In this embodiment, multiple detection segments of equal length are intercepted from the matched filtering result of a beam domain signal, and the interval between the starting positions of two detection segments with consecutive serial numbers is less than or equal to the length of the detection segment. In this way, each detection segment corresponds to a distance range with equal span, and the overall distance range is continuous. The smaller the length of the detection segment, the higher the distance resolution.

[0026] It should be noted that the ordinate of the matched filtering result is amplitude, and the abscissa may be time or distance according to different operating habits. The conversion relationship between the two is distance = time × speed of sound in water / 2.

[0027] When the horizontal axis of the matched filtering result is time and the vertical axis is amplitude, the distance range corresponding to the detection segment is obtained by converting the time range of the detection segment in the matched filtering result and the speed of sound in water; when the horizontal axis of the matched filtering result is distance and the vertical axis is amplitude, the distance range corresponding to the detection segment is the distance range of the detection segment in the matched filtering result.

[0028] Furthermore, in one embodiment, step S11 specifically includes: A matched filtering result of a beam domain signal is subjected to sliding window interception to obtain multiple detection segments, wherein the window shape intercepted by the sliding window is a rectangular window, and the sliding step length is less than or equal to the rectangular window length.

[0029] In this embodiment, the detection segment is intercepted by a sliding window, the rectangular window length controls the length of the detection segment, and the sliding step length controls the starting position interval of two detection segments with consecutive serial numbers, which is easy to operate.

[0030] S12. Perform kernel density estimation on each detection segment according to a preset kernel function to obtain a probability density function.

[0031] Obtaining the probability density function of the detection segment is a key step in this embodiment. If the probability statistics of the data with a limited number of points are directly performed, the probability density histogram of the data segment will be obtained. The result shows a discrete distribution characteristic and it is difficult to effectively reflect the probability density change trend of the data. Kernel density estimation (KDE) is a non-parametric estimation method in the probability density estimation method. First, the kernel function is fitted to the collected random observation sample values ​​one by one, and then the probability density functions obtained by fitting are superimposed to finally obtain the probability density function of the entire sample set. Therefore, this embodiment uses the kernel density estimation method to calculate the probability density function of the detection segment.

[0032] For example, the kernel density estimate is calculated as:

[0033] Among them, h is the smoothing parameter, called bandwidth, and x i is the i-th sampling point in the data, n is the total number of sampling points, and K() is the kernel function.

[0034] Specifically in this embodiment, the data is a detection segment, h can be 1, and the preset kernel function can be one of a uniform kernel function, a triangular kernel function, a biweight kernel function, and a triweight kernel function.

[0035] S13. For each detection segment, the relative entropy of the probability density function of the current detection segment and the probability density function of other detection segments is summed to obtain an overall difference value. The calculation formula of the overall difference value is:

[0036] Among them, K s (i) is the overall difference value of the i-th detection segment, KL() is the KL divergence calculation formula, PH(i) is the probability density function of the i-th detection segment, and PH(j) is the probability density function of the j-th detection segment.

[0037] In the deep-sea active detection scenario, the background noise is composed of multiple components such as ocean noise and ship noise. It is difficult to obtain the theoretical probability distribution of the echo matching filtering results in the absence of a target (pure background noise), and it is difficult to provide prior information on pure background noise.

[0038] Figure 2 A schematic diagram showing the matched filtering results of the measured active sonar background noise; Figure 3 Shows Figure 2 The schematic diagram of the probability density function of the matched filtering result in the range of 0-50 km is shown; Figure 4 Shows Figure 2 The schematic diagram of the probability density function of the matched filtering result at 50-100km is shown; Figure 5 Shows Figure 2 The diagram shows the probability density function of the matched filtering results at 100-150 km.

[0039] When analyzing active sonar background noise, Gaussian noise is often used as interference background, but Gaussian noise cannot accurately reflect the real background noise. Therefore, the echo data obtained from the actual deep-sea active sonar detection test is used for explanation. This test mainly uses towed sound sources, towed arrays and other sonar equipment to carry out active detection related data collection tests in the range of 5km~260km. The average sea depth of the test sea area is 4300 meters. The transmitted signal is HFM (hyperbolic frequency modulation signal). The received signal is preprocessed and matched filtered to obtain Figure 2 The matched filtering results are shown in Figure 2. From the matched filtering results, three distance ranges of 0-50km, 50-100km, and 100-150km are selected, the rectangular window length is set to 5km, the sliding step length is set to 5km, and the probability density function of each signal segment is calculated using the kernel density estimation method, as shown in Figure 2. Figure 3 , 4 , as shown in Figure 5.

[0040] from Figure 3 , 4 As can be seen from Figure 5, within the range of 0-50km, the probability density function changes extremely drastically, and within the range of 50-100km and 100-150km, the probability density curves are basically consistent, with only a certain deviation at the peak position. The above results show that the probability density function of long-distance background noise under deep-sea conditions is relatively stable on a certain scale, that is, the probability density function changes less over time, while the probability density function of the target echo is significantly different from that of the background noise. Therefore, the appearance of the target echo will cause a certain disturbance to the echo probability density. By detecting and extracting the characteristics of the disturbance, long-distance target detection under low signal-to-noise ratio can be achieved. That is, the difference between the probability density function of the detection segment of the target echo + background noise and the probability density function of the detection segment of pure background noise will be more obvious than the difference in the amplitude of the two detection segments.

[0041] Relative entropy, also known as KL divergence or information divergence, is an asymmetric measure of the difference between two probability distributions. In information theory, relative entropy is equivalent to the information entropy difference of two probability distributions. In this embodiment, relative entropy is used to characterize the difference in probability density function between one detection segment and another detection segment, and the sum of relative entropies of the probability density function of one detection segment and other detection segments is defined as the overall difference value to characterize the difference in probability density function between the echo of the corresponding distance range and the overall echo. Among multiple detection segments, a few detection segments may have targets, and most detection segments are pure background noise. The larger the overall difference value of a detection segment, the more likely it is that a target exists. In the absence of prior information of pure background noise, the overall difference value can amplify the disturbance of the target echo to the echo probability density.

[0042] S14. Determine whether there is a target within the distance range corresponding to the detection segment according to the overall difference value of the detection segment.

[0043] In this embodiment, since the overall difference value can amplify the disturbance of the target echo on the echo probability density, using the overall difference value as the judgment basis can achieve better detection effect in long-distance target detection compared with the traditional method of directly using the matched filter peak amplitude as the judgment basis.

[0044] Therefore, in this embodiment, by performing segmented kernel density estimation on the matched filtering results of the active sonar echo, the probability density function of each detection segment is obtained, and the relative entropy is used to characterize the difference in the probability density function between one detection segment and another detection segment. The relative entropy corresponding to one detection segment is summed to obtain the overall difference value to characterize the difference in the probability density function between the echo of the corresponding distance range and the overall echo. According to the overall difference value of the detection segment, it is judged whether there is a target within the distance range corresponding to the detection segment. Since the probability density function of the long-distance background noise under deep-sea conditions has a strong time stationarity, the appearance of the target echo will cause a certain disturbance to the echo probability density, and the overall difference value can amplify this disturbance. Therefore, through this embodiment, the target echo output signal-to-noise ratio can be improved for long-distance targets to achieve accurate detection.

[0045] It should be noted that according to the background noise characteristics analyzed in the previous article, in order to ensure the detection accuracy, the overall distance range corresponding to multiple detection segments needs to meet the requirements of long distance and a certain span.

[0046] For example, assume that the distance range corresponding to the matched filtering result is 0-200km, among which 0-50km is a short distance and is not applicable to this solution. The span of 50-200km is too large and needs to be further divided into three major sections: 50-100km, 100-150km, and 150-200km. In each section, 10 detection segments with a length of 5km are intercepted, and the overall difference values ​​​​are calculated between the 10 detection segments in each section.

[0047] Furthermore, in one embodiment, step S14 specifically includes: For each detection segment, the root mean square of the overall difference values ​​of multiple background segments corresponding to the current detection segment is calculated to obtain the background intensity, and whether there is a target within the distance range corresponding to the current detection segment is determined based on the background intensity and the overall difference value, wherein the absolute value of the serial number difference between the current detection segment and each corresponding background segment is greater than or equal to the protection length.

[0048] In this embodiment, the background intensity can feedback the overall difference value of multiple background segments corresponding to the current detection segment. The difference between the background intensity and the overall difference value of the target detection segment and other detection segments is greater. Through this embodiment, the difference between the target detection segment and other detection segments can be further amplified, thereby improving the detection accuracy.

[0049] Furthermore, in one embodiment, the absolute value of the difference between the sequence numbers of the current detection segment and each of its corresponding background segments is less than or equal to the background length.

[0050] For example, the protection length is 1, the background length is 4, there are 10 detection segments in total, the background segments of detection segment 1 are detection segments 3, 4, and 5, the background segments of detection segment 2 are 4, 5, and 6, and so on.

[0051] For example, the protection length is 1, the background length is not set, there are 10 detection segments in total, the background segments of detection segment 1 are detection segments 3-10, the background segments of detection segment 2 are 4-10, and so on.

[0052] For example, the background intensity is calculated as:

[0053] Among them, B i is the background intensity of the i-th detection segment, K s (k) is the overall difference value of the kth detection segment, n L is the background length, n p is the protection length, and n is the number of background segments.

[0054] Furthermore, in one embodiment, the step of judging whether there is a target within the distance range corresponding to the current detection segment according to the background intensity and the overall difference value includes: If the decibel value of the overall difference value of the current detection segment to the background intensity is greater than or equal to the preset threshold, it is determined that there is a target within the distance range corresponding to the current detection segment.

[0055] In this embodiment, the relationship between the overall difference value of the current detection segment and the background intensity is described by the decibel value, which conforms to the conventional operating habits.

[0056] For example, the formula for calculating and comparing decibel values ​​is as follows:

[0057] Among them, K s (i) is the overall difference value of the kth detection segment, B i is the background intensity of the ith detection segment, DT is the threshold directly calculated according to the false alarm rate requirement, which is 6 dB, and EM is the engineering margin determined empirically, which is 3 dB.

[0058] The following is combined with specific experimental data to illustrate the improvement of this method in long-distance target detection compared with traditional methods.

[0059] Figure 6 A schematic diagram of a three-dimensional model of a typical underwater target is shown; Figure 7 Shows Figure 6 Schematic diagram of the impulse response function of the underwater target shown; Figure 8 Shows Figure 6 Schematic diagram of the two-way impulse response function of the fourth shadow channel of the underwater target shown; Fig. 9 Shows Figure 6 Schematic diagram of the complete echo of the underwater target shown; Fig.10 for Fig. 9 Schematic diagram of the probability density function of the matched filtering result of the complete echo shown.

[0060] Reference Figures 6 to 10, the finite element method is used to simulate and analyze the acoustic scattering characteristics of typical underwater targets. By three-dimensionally modeling the typical underwater target, the finite element method is used to calculate the target impulse response function, and the ray model is used to calculate the channel impulse response function. The convolution of the two can obtain the complete impulse response of the typical underwater target at different frequency bands and different incident angles. The main simulation conditions are as follows: the hull length is 80m, the cross-sectional diameter is 8m, the bridge height is 5m, the incident angle is 45° (with the bow direction as 0°), the signal frequency band is 100Hz to 2kHz, the sampling rate is 5kHz, the transmission signal is assumed to be an HFM signal, the pulse width is 4s, the frequency band range is 375Hz to 625Hz, and the target is located at 200km (the fourth shadow area). The bellhop acoustic tool is used to calculate the channel impulse response, the sound velocity profile is MUNK type, and the depth of the target and the transmitting sound source are both 100m.

[0061] Fig.11 A schematic diagram showing the matched filtering result of the measured background noise superimposed on the simulated target echo; Fig.12 This method shows that Fig.11 Schematic diagram of the overall difference values ​​of different detection segments calculated from the matched filtering results shown.

[0062] The echo input signal-to-noise ratio is -19dB, and the measured background noise is superimposed on the simulated target echo. The input / output signal-to-noise ratio calculation formula is:

[0063]

[0064] Among them, a mf is the peak amplitude of the signal, b mf is the background intensity, a ij is the signal amplitude at (i, j), (i, j) is the area within the background range Q and outside the protection area P, the background range Q refers to the echo signal with a distance range of 10 km centered on the peak amplitude position, the protection area P refers to the signal within 1 km from the peak position, and N is the total number of calculation points.

[0065] For distant targets with weak echo, Fig.11 It can be seen that the difference between the matched filter peaks of the target area and other areas is not obvious. Fig.12 It can be seen that there is a significant difference in the overall difference value (KL divergence integral) between the detection segment (range unit) where the target is located and other detection segments, proving that the overall difference value can indeed amplify the disturbance of the target echo on the echo probability density.

[0066] Fig.13The figure shows the comparison of the detection results of the traditional method and the present method when the echo input signal-to-noise ratio is -20dB and the detection threshold is 10dB.

[0067] Reference Fig.13 When the echo input signal-to-noise ratio is -20dB and the detection threshold is 10dB, the detection result of the traditional method is that there is no target, while the detection result of this method is that there is a target within the distance range of 200-205km.

[0068] Fig.14 The figure shows the ROC curve comparison between the traditional method and the present method under different echo input signal-to-noise ratios.

[0069] The receiver operating characteristic curve (ROC curve) can quantitatively measure the target detection capabilities of different methods. By changing the detection threshold, performing M = 10,000 Monte Carlo tests under Gaussian noise background, and counting the number of target detections N d And the number of false alarms N fa , the false alarm rate and detection rate can be expressed as:

[0070]

[0071] Based on the above formula, the ROC curves of the traditional method and the proposed method under different echo input signal-to-noise ratios are calculated respectively.

[0072] Reference Fig.14 , when the echo input signal-to-noise ratio (SRNin) is -19dB, -20dB and -21dB, the ROC curve of this method is above the ROC curve of the traditional method, indicating that this method has stronger detection ability. Fig.14 In (a) and (b), when the echo input signal-to-noise ratio decreases, the ROC curves of the two methods move downward, indicating that the detection performance of both methods decreases. Fig.14 In (a) and (c), when the echo input SNR increases, the ROC curves of the two methods move upward, indicating that the detection performance of both methods increases at this time, and the echo input SNR has a limiting effect on the detection performance of the two methods. When the echo input SNR is -19dB, the detection performance of the two methods is good, indicating that the lower the echo input SNR, the more obvious the advantage of this method.

[0073] Since the ROC curve changes with the echo input signal-to-noise ratio, in order to intuitively measure the comprehensive detection capabilities of the two methods, the echo input signal-to-noise ratio at which the false alarm rate is no more than 0.1% and the detection probability is no less than 90% is taken as the minimum detectable input signal-to-noise ratio SNR of each method. in-min , the SNR of the two methods is calculated based on the ROC curve under the echo input signal-to-noise ratio in-min , the SNR of the traditional methodin-min is -18.7dB, the SNR of this method in-min It is -22.3dB. It can be concluded that under Gaussian background, compared with the traditional method, this method can reduce the minimum detectable input signal-to-noise ratio threshold by 3.6dB.

[0074] In a second aspect, an embodiment of the present application also provides a sonar long-distance target detection device.

[0075] Fig.15 A schematic diagram of the functional modules of a sonar long-distance target detection device in one embodiment of the present application is shown.

[0076] Reference Fig.15 In one embodiment, the sonar long-distance target detection device includes: A segmentation interception module 10 is used to intercept multiple detection segments from a matched filtering result of a beam domain signal, wherein the beam domain signal is obtained by preprocessing the sonar echo signal, and the preprocessing includes beam forming, the length of each detection segment is equal, and the interval between the starting positions of two detection segments with consecutive serial numbers is less than or equal to the length of the detection segment; A kernel density estimation module 20, used for performing kernel density estimation on each detection segment according to a preset kernel function to obtain a probability density function; The difference calculation module 30 is used to sum the relative entropy of the probability density function of the current detection segment and the probability density function of other detection segments for each detection segment to obtain an overall difference value. The calculation formula of the overall difference value is:

[0077] Among them, K s (i) is the overall difference value of the i-th detection segment, KL() is the KL divergence calculation formula, PH(i) is the probability density function of the i-th detection segment, and PH(j) is the probability density function of the j-th detection segment; The judgment module 40 is used to judge whether there is a target within the distance range corresponding to the detection segment according to the overall difference value of the detection segment.

[0078] Furthermore, in one embodiment, the determination module 40 is used to: For each detection segment, the root mean square of the overall difference values ​​of multiple background segments corresponding to the current detection segment is calculated to obtain the background intensity, and whether there is a target within the distance range corresponding to the current detection segment is determined based on the background intensity and the overall difference value, wherein the absolute value of the serial number difference between the current detection segment and each corresponding background segment is greater than or equal to the protection length.

[0079] Furthermore, in one embodiment, the absolute value of the difference between the sequence numbers of the current detection segment and each of its corresponding background segments is less than or equal to the background length.

[0080] Furthermore, in one embodiment, the determination module 40 is used to: If the decibel value of the overall difference value of the current detection segment to the background intensity is greater than or equal to the preset threshold, it is determined that there is a target within the distance range corresponding to the current detection segment.

[0081] Further, in one embodiment, when the horizontal coordinate of the matched filtering result is time and the vertical coordinate is amplitude, the distance range corresponding to the detection segment is obtained by converting the time range of the detection segment in the matched filtering result and the speed of sound in water; When the horizontal coordinate of the matched filtering result is distance and the vertical coordinate is amplitude, the distance range corresponding to the detection segment is the distance range of the detection segment in the matched filtering result.

[0082] Furthermore, in one embodiment, the segmentation interception module 10 is used to: A matched filtering result of a beam domain signal is subjected to sliding window interception to obtain multiple detection segments, wherein the window shape intercepted by the sliding window is a rectangular window, and the sliding step length is less than or equal to the rectangular window length.

[0083] Furthermore, in one embodiment, the preset kernel function is one of a uniform kernel function, a triangular kernel function, a biweight kernel function and a triweight kernel function.

[0084] Among them, the functional implementation of each module in the above-mentioned sonar long-distance target detection device corresponds to the various steps in the above-mentioned sonar long-distance target detection method embodiment, and its functions and implementation processes will not be repeated here one by one.

[0085] In a third aspect, an embodiment of the present application provides a sonar long-distance target detection device, which may be a device with data processing capabilities, such as a personal computer (PC), a laptop computer, or a server.

[0086] Fig.16 A schematic diagram of the hardware structure of a sonar long-distance target detection device involved in an embodiment of the present application is shown.

[0087] Reference Fig.16 In an embodiment of the present application, the sonar long-range target detection device may include a processor, a memory, a communication interface, and a communication bus.

[0088] The communication bus may be of any type and is used to interconnect the processor, the memory, and the communication interface.

[0089] Communication interfaces include input / output (I / O) interfaces, physical interfaces, and logical interfaces, which are used to interconnect devices within the sonar long-distance target detection device, and interfaces used to interconnect the sonar long-distance target detection device with other devices (such as other computing devices or user devices). Physical interfaces can be Ethernet interfaces, optical fiber interfaces, ATM interfaces, etc.; user devices can be display screens (Display), keyboards (Keyboard), etc.

[0090] The memory can be various types of storage media, such as random access memory (RAM), read-only memory (ROM), non-volatile RAM (NVRAM), flash memory, optical storage, hard disk, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), etc.

[0091] The processor may be a general-purpose processor, which may call the sonar long-distance target detection program stored in the memory and execute the sonar long-distance target detection method provided in the embodiment of the present application. For example, the general-purpose processor may be a central processing unit (CPU). The method executed when the sonar long-distance target detection program is called may refer to the various embodiments of the sonar long-distance target detection method of the present application, which will not be described in detail here.

[0092] Those skilled in the art will understand that Fig.16 The hardware structure shown in the figure does not constitute a limitation on the present application, and may include more or less components than shown in the figure, or combine certain components, or arrange the components differently.

[0093] In a fourth aspect, an embodiment of the present application also provides a readable storage medium.

[0094] The readable storage medium of the present application stores a sonar long-distance target detection program, wherein when the sonar long-distance target detection program is executed by a processor, the steps of the sonar long-distance target detection method as described above are implemented.

[0095] Among them, the method implemented when the sonar long-distance target detection program is executed can refer to the various embodiments of the sonar long-distance target detection method of the present application, and will not be repeated here.

[0096] It should be noted that the serial numbers of the above-mentioned embodiments of the present application are only for description and do not represent the advantages or disadvantages of the embodiments.

[0097] The terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally includes other steps or units inherent to these processes, methods, products or devices. The terms "first", "second" and "third" are used to distinguish different objects, etc., and do not represent a sequence, nor do they limit "first", "second" and "third" to different types.

[0098] In the description of the embodiments of the present application, "exemplary", "for example" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary", "for example" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary", "for example" or "for example" is intended to present related concepts in a specific way.

[0099] In the description of the embodiments of the present application, unless otherwise specified, “ / ” means or, for example, A / B can mean A or B; the “and / or” in the text is merely a description of the association relationship of associated objects, indicating that three relationships may exist, for example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, in the description of the embodiments of the present application, “multiple” refers to two or more than two.

[0100] In some processes described in the embodiments of the present application, multiple operations or steps that appear in a specific order are included, but it should be understood that these operations or steps may not be executed in the order in which they appear in the embodiments of the present application or in parallel, and the sequence number of the operation is only used to distinguish the different operations, and the sequence number itself does not represent any execution order. In addition, these processes may include more or fewer operations, and these operations or steps may be executed in sequence or in parallel, and these operations or steps may be combined.

[0101] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, disk, CD) as described above, and includes a number of instructions for a terminal device to execute the methods described in each embodiment of the present application.

[0102] The above are only preferred embodiments of the present application, and are not intended to limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A sonar long-distance target detection method, characterized in that: The sonar long-distance target detection method comprises: Extracting multiple detection segments from a matched filtering result of a beam domain signal, wherein the beam domain signal is obtained by preprocessing the sonar echo signal, the preprocessing includes beam forming, the length of each detection segment is equal, and the interval between the starting positions of two detection segments with consecutive serial numbers is less than or equal to the length of the detection segment; Perform kernel density estimation on each detection segment according to the preset kernel function to obtain a probability density function; For each detection segment, the relative entropy of the probability density function of the current detection segment and the probability density function of other detection segments is summed to obtain the overall difference value. The calculation formula of the overall difference value is: Among them, K s (i) is the overall difference value of the i-th detection segment, KL() is the KL divergence calculation formula, PH(i) is the probability density function of the i-th detection segment, and PH(j) is the probability density function of the j-th detection segment; The overall difference value of the detection segment is used to determine whether there is a target within the distance range corresponding to the detection segment.

2. The sonar long-distance target detection method according to claim 1, characterized in that: The step of judging whether there is a target within the distance range corresponding to the detection segment according to the overall difference value of the detection segment comprises: For each detection segment, the root mean square of the overall difference values ​​of multiple background segments corresponding to the current detection segment is calculated to obtain the background intensity, and whether there is a target within the distance range corresponding to the current detection segment is determined based on the background intensity and the overall difference value, wherein the absolute value of the serial number difference between the current detection segment and each corresponding background segment is greater than or equal to the protection length.

3. The sonar long-distance target detection method according to claim 2, characterized in that: The absolute value of the difference between the sequence numbers of the current detection segment and each of its corresponding background segments is less than or equal to the background length.

4. The sonar long-distance target detection method according to claim 2, characterized in that: The step of judging whether there is a target within the distance range corresponding to the current detection segment according to the background intensity and the overall difference value comprises: If the decibel value of the overall difference value of the current detection segment to the background intensity is greater than or equal to the preset threshold, it is determined that there is a target within the distance range corresponding to the current detection segment.

5. The sonar long-distance target detection method according to claim 1, characterized in that: When the horizontal axis of the matched filtering result is time and the vertical axis is amplitude, the distance range corresponding to the detection segment is obtained by converting the time range of the detection segment in the matched filtering result and the speed of sound in water; When the horizontal coordinate of the matched filtering result is distance and the vertical coordinate is amplitude, the distance range corresponding to the detection segment is the distance range of the detection segment in the matched filtering result.

6. The sonar long-distance target detection method according to claim 1, characterized in that: The step of extracting a plurality of detection segments from a matched filtering result of a beam domain signal comprises: A matched filtering result of a beam domain signal is subjected to sliding window interception to obtain multiple detection segments, wherein the window shape intercepted by the sliding window is a rectangular window, and the sliding step length is less than or equal to the rectangular window length.

7. The sonar long-distance target detection method according to claim 1, characterized in that: The preset kernel function is one of the uniform kernel function, triangular kernel function, biweight kernel function and triweight kernel function.

8. A sonar long-distance target detection device, characterized in that: The sonar long-distance target detection device comprises: A segmentation interception module is used to intercept multiple detection segments from a matched filtering result of a beam domain signal, wherein the beam domain signal is obtained by preprocessing the sonar echo signal, and the preprocessing includes beam forming, the length of each detection segment is equal, and the interval between the starting positions of two detection segments with consecutive serial numbers is less than or equal to the length of the detection segment; A kernel density estimation module is used to perform kernel density estimation on each detection segment according to a preset kernel function to obtain a probability density function; The difference calculation module is used to sum the relative entropy of the probability density function of the current detection segment and the probability density function of other detection segments for each detection segment to obtain the overall difference value. The calculation formula of the overall difference value is: Among them, K s (i) is the overall difference value of the i-th detection segment, KL() is the KL divergence calculation formula, PH(i) is the probability density function of the i-th detection segment, and PH(j) is the probability density function of the j-th detection segment; The judgment module is used to judge whether there is a target within the distance range corresponding to the detection segment according to the overall difference value of the detection segment.

9. A sonar long-distance target detection device, characterized in that: The sonar long-range target detection device includes a processor, a memory, and a sonar long-range target detection program stored in the memory and executable by the processor, wherein when the sonar long-range target detection program is executed by the processor, the steps of the sonar long-range target detection method as described in any one of claims 1 to 7 are implemented.

10. A readable storage medium, characterized in that: The readable storage medium stores a sonar long-distance target detection program, wherein when the sonar long-distance target detection program is executed by a processor, the steps of the sonar long-distance target detection method according to any one of claims 1 to 7 are implemented.

Images