Weak target detection method, system and computing device for sonar images

Abstract

The present invention discloses a weak target detection method, system and computing device for sonar images, and relates to the technical field of sonar image recognition, wherein the weak target detection method comprises: performing Gamma modeling on the sonar image after smoothing; designing a constant false alarm detection window for detecting the sonar image after modeling; binarizing the detected sonar image to estimate the background; clustering the binary image; and after extracting objects that meet the brightness characteristics in the sonar image, judging whether the object is a real moving target or a pseudo target. Through the above steps, in a complex and changeable environment, the weak target detection method can also provide robust detection, and accurately and efficiently detect weak targets in the sonar image.

Description

Technical Field

[0001] The present invention relates to the technical field of sonar image recognition, and in particular to a sonar image weak target detection method, system and computing device. Background Art

[0002] Target detection technology is an important part of sonar image signal processing. However, in shallow offshore environments, active sonar target detection will be disturbed by the reverberation generated by a large number of scatterers on the sea surface, seabed and ocean. The complex changes of these reverberations directly affect the target detection performance of the active sonar system. In addition, the fluctuation of target echo intensity is also a major difficulty in detection. The main difficulty of the current detection algorithm lies in autonomous detection, especially the adaptability of the detection algorithm to complex background interference. Accurate dynamic background estimation, interference suppression, false alarm control, and highlighting weak targets are the key points of target detection. When the target echo signal-to-noise ratio is low, the classic local threshold or dual-parameter constant false alarm detection technology is prone to lose the target, affecting the stability of real-time target detection and subsequent recognition processing. Summary of the invention

[0003] An object of the present invention is to provide a weak target detection method, system and computing device for sonar images, wherein in the presence of reverberation interference, the weak target detection method of the present invention can accurately detect weak targets in sonar images to determine whether they are real moving targets or false targets.

[0004] In order to achieve the above object, the technical solution provided by the present invention is: a weak target detection method for sonar images, wherein the weak target detection method comprises the following steps:

[0005] S1, Gamma modeling of the smoothed sonar image;

[0006] S2, design a constant false alarm detection window for detecting the sonar image after modeling;

[0007] S3, binarizing the detected sonar image to estimate the background;

[0008] S4, clustering the binary image;

[0009] S5, after extracting the object that meets the brightness feature in the sonar image, it is determined whether the object is a real moving target or a false target.

[0010] As a further improvement of the technical solution of the present invention, in step S1, the parameter Indicates that there is a target echo reflection point at the position to be detected. Indicates that there is only noise at the position to be detected, where the parameter and parameters satisfy:

[0011] , where the parameters is the smoothed signal strength, parameter is the decision threshold to be determined, the smoothed signal strength In the assumption The probability density function under ,parameter is the specific value of each signal, parameter is the number of smoothed data, parameter is the scale parameter, signal strength In the assumption The cumulative distribution probability under , The closed form expression for .

[0012] As a further improvement of the technical solution of the present invention, in step S2, the constant false alarm detection window includes a unit to be detected, protection units located on opposite sides of the unit to be detected, and reference units located on the upper and lower sides of the unit to be detected and the two protection units, wherein the mean parameter of the reference unit on the upper side is , the mean parameter of the reference unit on the lower side is ,parameter The mean of ,parameter The mean of ,parameter is the number of pixels in the reference unit on the upper side, and the parameter is the number of pixels in the reference unit on the lower side, and the parameter is the reference unit, parameter is the total number of pixels in the reference unit, which satisfies the relationship , wherein in step S3, the background clutter estimation value is , the decision threshold satisfy , where the parameters is the threshold factor, parameter It is calculated by the corresponding false alarm probability expression, and the calculation formula satisfies:

[0013] .

[0014] As a further improvement of the technical solution of the present invention, in step S4, the DBSCAN algorithm is used to cluster the binary image.

[0015] As a further improvement of the technical solution of the present invention, the step S5 further comprises:

[0016] S5.1, calculating the geometric size of the extracted highlighted object, and marking the object whose geometric size is larger than a set threshold as a suspected target;

[0017] S5.2, mark and number the suspected targets;

[0018] S5.3, calculating the center of mass position of the suspected target and storing it in an array to record the movement trajectory of the suspected target in the time series;

[0019] S5.4, analyze the change in the center of mass position of the suspected target in multiple adjacent frames of sonar images to determine the motion state of the suspected target, wherein when the change in the center of mass position of the suspected target exceeds a set motion threshold, the object is determined to be a real moving target; when the change in the center of mass position of the suspected target does not exceed the set motion threshold, the object is determined to be a false target.

[0020] According to another aspect of the present invention, the present invention further provides a weak target detection system for sonar images, comprising:

[0021] A modeling unit, used for performing Gamma modeling on the sonar image after smoothing;

[0022] A design unit, used for designing a constant false alarm detection window for detecting the sonar image after modeling;

[0023] A binarization unit is used to perform binarization processing on the detected sonar image to estimate the background;

[0024] A clustering unit, used for clustering binary images;

[0025] The judgment unit is used to judge whether the object is a real moving target or a false target after extracting the object that meets the brightness characteristics in the sonar image.

[0026] As a further improvement of the technical solution of the present invention, the modeling unit includes a smoothing module and a modeling module, the smoothing module is used to smooth the sonar signal data, and the modeling module is used to perform Gamma modeling, wherein the parameter Indicates that there is a target echo reflection point at the position to be detected. Indicates that there is only noise at the position to be detected, where the parameter and parameters satisfy:

[0027] , where the parameters is the smoothed signal strength, parameter is the decision threshold to be determined, the smoothed signal strength In the assumption The probability density function under ,parameter is the specific value of each signal, parameter is the number of smoothed data, parameter is the scale parameter, signal strength In the assumption The cumulative distribution probability under , The closed form expression for .

[0028] As a further improvement of the technical solution of the present invention, the constant false alarm detection window includes a unit to be detected, a protection unit located on opposite sides of the unit to be detected, and a reference unit located on the upper and lower sides of the unit to be detected and the two protection units, wherein the mean parameter of the reference unit on the upper side is , the mean parameter of the reference unit on the lower side is ,parameter The mean of ,parameter The mean of ,parameter is the number of pixels in the reference unit on the upper side, and the parameter is the number of pixels in the reference unit on the lower side, and the parameter is the reference unit, parameter is the total number of pixels in the reference unit, which satisfies the relationship , where the background clutter estimate is , the decision threshold satisfy , where the parameters is the threshold factor, parameter It is calculated by the corresponding false alarm probability expression, and the calculation formula satisfies:

[0029] .

[0030] As a further improvement of the technical solution of the present invention, the judgment unit includes a first calculation module, a marking module, a second calculation module and a judgment module, wherein the first calculation module is used to calculate the geometric dimensions of the extracted highlighted objects, and mark objects whose geometric dimensions are greater than a set threshold as suspected targets; the marking module is used to perform regional marking and numbering on suspected targets; the second calculation module is used to calculate the center of mass position of the suspected target, and store it in an array to record the motion trajectory of the suspected target in a time series; the judgment module is used to analyze the change in the center of mass position of the suspected target in adjacent multiple frames of sonar images, and judge the motion state of the suspected target, wherein when the change in the center of mass position of the suspected target exceeds the set motion threshold, the object is judged to be a real moving target, and when the change in the center of mass position of the suspected target does not exceed the set motion threshold, the object is judged to be a false target.

[0031] According to another aspect of the present invention, the present invention further provides a computing device, comprising a processor and a memory coupled to the processor, wherein the memory stores a code, and the processor is configured to execute the code. When the code is executed, the computing device executes a weak target detection method, wherein the weak target detection method comprises the following steps:

[0032] S1, Gamma modeling of the smoothed sonar image;

[0033] S2, design a constant false alarm detection window for detecting the sonar image after modeling;

[0034] S3, binarizing the detected sonar image to estimate the background;

[0035] S4, clustering the binary image;

[0036] S5, after extracting the object that meets the brightness feature in the sonar image, it is determined whether the object is a real moving target or a false target.

[0037] Compared with the prior art, the present invention has at least the following beneficial effects:

[0038] 1. In the presence of reverberation interference, the weak target detection method of the present invention can accurately detect weak targets in sonar images to determine whether the weak targets are real moving targets or false targets;

[0039] 2. In complex environments, the weak target detection method of the present invention can provide robust detection and realize detection by an active sonar system.

[0040] Other features and beneficial effects of the present invention will be further disclosed and explained in the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a block diagram of a weak target detection method for sonar images according to a preferred embodiment of the present invention.

[0042] Figure 2 It is a detection flow chart of the weak target detection method according to the above preferred embodiment of the present invention.

[0043] Figure 3 It is a schematic diagram of the sonar data distribution fitting result of the weak target detection method according to the above preferred embodiment of the present invention.

[0044] Figure 4 It is a schematic diagram of a constant false alarm detection window designed according to the weak target detection method of the above preferred embodiment of the present invention.

[0045] Figure 5 It is a detection effect diagram of the weak target detection method according to the above preferred embodiment of the present invention.

[0046] Figure 6 This is the detection effect diagram of the traditional dual-parameter detection method.

[0047] Figure 7 It is a block diagram of a weak target detection system for sonar images according to a preferred embodiment of the present invention.

[0048] Figure 8 It is a block diagram of a computing device according to a preferred embodiment of the present invention.

[0049] In the figure:

[0050] 10. CFAR detection window; 11. Unit to be detected; 12. Reference unit; 13. Protection unit;

[0051] 20. Weak target detection system; 21. Modeling unit; 211. Smoothing module; 212. Modeling module; 22. Design unit; 23. Binarization unit; 24. Clustering unit; 25. Judgment unit; 251. First calculation module; 252. Labeling module; 253. Second calculation module; 254. Judgment module;

[0052] 30. Computing device; 31. Processor; 32. Memory; 33. Transceiver. DETAILED DESCRIPTION

[0053] Before describing in detail any embodiment of the present invention, it should be understood that the present invention is not limited in its application to the construction and arrangement details of the components described below or illustrated in the following figures. The present invention can have other embodiments and can be practiced or carried out in various ways. In addition, it should be understood that the words and terms used here are for descriptive purposes and should not be considered restrictive. The use of "including" or "having" and its variations herein is intended to cover the items and their equivalents and additional items displayed below. Unless otherwise specified or limited, the terms "install", "connect", "support" and "couple" and their variations are widely used and cover direct installation and indirect installation, connection, support and connection. In addition, "connect" and "couple" are not limited to physical or mechanical connections or connections.

[0054] Furthermore, on the first aspect, in the disclosure of the present invention, the terms "longitudinal", "lateral", "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the accompanying drawings, which are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore the above terms cannot be understood as limitations on the present invention; on the second aspect, the term "one" should be understood as "at least one" or "one or more", that is, in one embodiment, the number of an element may be one, while in another embodiment, the number of the element may be multiple, and the term "one" cannot be understood as a limitation on the quantity.

[0055] Attached Figure 1 A weak target detection method for a sonar image according to a preferred embodiment of the present invention is shown, and the weak target detection method includes the following steps: S1, performing Gamma modeling on the sonar image after smoothing; S2, designing a constant false alarm detection window 10 for detecting the sonar image after modeling; S3, binarizing the detected sonar image to estimate the background; S4, clustering the binary image; S5, after extracting objects that meet the brightness characteristics in the sonar image, judging whether the object is a real moving target or a false target. Through such steps, in the presence of reverberation interference, the weak target detection method of the present invention can accurately detect weak targets in sonar images to judge whether the weak target is a real moving target or a false target, and the weak target detection method can provide robust detection to achieve detection of active sonar systems.

[0056] More specifically, step S1 includes a smoothing process and a Gamma modeling process. In other words, in step S1, the weak target detection method of the present invention first performs smoothing on the sonar signal data to remove speckle noise to a certain extent, thereby improving the target detection, classification and filtering performance of the sonar. The echo signal received by the sonar is an independent and identically distributed single variable. , The smoothed signal is , where the parameters is the echo signal, parameter is the amount of smoothed data.

[0057] Reference Figure 3 The columnar distribution in the figure is the real signal data of the sonar, and the other three curves represent the normal distribution, Rayleigh distribution and Gamma distribution respectively. By fitting the data of each channel after smoothing, it can be seen that compared with the normal distribution and Rayleigh distribution, the Gamma distribution is more in line with the structure of the sonar data at this time. Therefore, the weak target detection method of the present invention uses the Gamma distribution model in the step S1 to perform Gamma modeling on the smoothed sonar image.

[0058] In step S1, the process of using the Gamma distribution model to perform Gamma modeling on the smoothed sonar image is as follows: the target detection process is described as a binary hypothesis testing process, assuming that the parameter Indicates that there is a target echo reflection point at the position to be detected. Indicates that there is only noise at the position to be detected, where the parameter and parameters satisfy:

[0059] ,parameter is the smoothed signal strength, parameter is the decision threshold to be determined, where when the smoothed signal strength is greater than the decision threshold (i.e., > ), indicating that there is a target at the position to be detected. Accordingly, when the smoothed signal strength is not greater than the decision threshold (i.e., ≤ ), indicating that there is no target at the location to be detected. Through the above distribution fitting, the smoothed signal strength can be determined Obeying Gamma distribution, the smoothed signal strength In the assumption The probability density function under ,parameter is the specific value of each signal, parameter is the number of smoothed data, parameter is the scale parameter, signal strength In the assumption The cumulative distribution probability under , The closed form expression for .

[0060] In step S2, refer to the attached Figure 4 The weak target detection method of the present invention uses a two-dimensional smooth window when designing the constant false alarm detection window 10. The constant false alarm detection window 10 includes a unit to be detected 11, two reference units 12 and two protection units 13. The reference unit 12 is respectively provided on the upper side and the lower side of the unit to be detected 11. The reference unit 12 is used to count background clutter. Since the target appears as an arc in the fan diagram of the sonar, it is considered to provide a protection unit 13 on the opposite sides of the unit to be detected 11 when designing the constant false alarm detection window 10, so as to protect the unit to be detected 11, prevent the unit to be detected 11 from falling into the reference unit 12, and avoid the unit to be detected 11 from participating in the calculation as background clutter and affecting the subsequent detection results. Preferably, the reference unit 12 on the upper side is located on the upper side of the two protection units 13 and the unit to be detected 11, and the reference unit 12 on the lower side is located on the lower side of the two protection units 13 and the unit to be detected 11.

[0061] The mean parameter of the reference unit 12 on the upper side is , the mean parameter of the reference unit 12 on the lower side is ,parameter The mean of ,parameter The mean of ,parameter is the number of pixels in the reference unit 12 on the upper side, and the parameter is the number of pixels in the reference unit 12 at the bottom, and the parameter is the reference unit 12, parameter is the total number of pixels in the reference unit 12, which satisfies the relationship .

[0062] In step S3, the background clutter estimation value is , the decision threshold satisfy , where the parameters is the threshold factor, parameter It is calculated by the corresponding false alarm probability expression, and the calculation formula satisfies:

[0063] .

[0064] After constant false alarm detection, the sonar image is processed into a binary image, wherein the highlight area in the binary image is marked as the target, and its value is 1, and other areas are regarded as the background, and its value is 0. The contrast between the target and the background can be significantly improved by the method of binary image, which is convenient for subsequent regional extraction and analysis. At the same time, the distribution information of the target area in the binary image reflects the potential object characteristics, but due to the possible noise interference or false detection points in the detection process, it is also necessary to cluster the binary image in the future to ensure that the target area extracted later is more accurate and complete. Optionally, in other examples of the weak target detection method of the present invention, the weak target detection method can also perform morphological operations or filtering on the binary image to further optimize the binary image to ensure that the target area extracted later is more accurate and complete.

[0065] In step S4, the DBSCAN clustering method is used to cluster the binary image generated after constant false alarm detection. In this way, during the clustering step, the weak target detection method can find clusters of any shape in the sonar data, and effectively identify noise points. By analyzing the density distribution of data points, the areas with higher density are divided into the same cluster, and the points with lower density are identified as noise or isolated points. During the clustering process, the DBSCAN clustering method does not need to pre-specify the number of clusters. Therefore, the weak target detection method of the present invention can process complex data sets with nonlinear distribution, and is particularly suitable for extracting high-density target areas in noisy data.

[0066] Reference Figure 5After constant false alarm detection and target area clustering, the weak target detection method extracts objects that meet the brightness characteristics in the sonar image. These objects may include moving targets that need to be identified (for example, submarines, fishing boats, etc.), stationary background objects (for example, seabed rocks or fixed structures), and pseudo targets introduced by the sonar system or environmental noise. Therefore, in order to further screen out real moving targets, it is necessary to combine the information of multiple frames of sonar images for accurate analysis. In order to achieve the above purpose, the step S5 may further include: S5.1, calculating the geometric size of the extracted highlighted object, marking the object whose geometric size is greater than the set threshold as a suspected target; S5.2, marking and numbering the suspected target; S5.3, calculating the centroid position of the suspected target, and storing it in an array to record the motion trajectory of the suspected target in the time series; S5.4, analyzing the change in the centroid position of the suspected target in the adjacent multi-frame sonar image, and judging the motion state of the suspected target, wherein when the change in the centroid position of the suspected target exceeds the set motion threshold, the object is judged to be a real moving target, and when the change in the centroid position of the suspected target does not exceed the set motion threshold, the object is judged to be a false target, so that the weak target detection method of the present invention realizes the detection of weak targets in sonar images. Through the above steps, the weak target detection method combines spatial and temporal information, which can not only improve the detection accuracy of moving targets, but also effectively reduce the false detection caused by noise or stationary objects, and ensure the stability and reliability of the active sonar system.

[0067] Attached Figure 5 Sonar images and Figure 6 The sonar data corresponding to the sonar images shown are the same. Figure 5 is a detection effect diagram of the weak target detection method of the present invention, Figure 6 It is a detection effect diagram of the traditional dual-parameter detection method. When the traditional dual-parameter detection method is used for detection, both real moving targets and false targets are detected as real moving targets, resulting in false detections and affecting the detection accuracy of the moving target side. The weak target detection method of the present invention can effectively remove false targets, effectively reduce false detections, and ensure the stability and reliability of the active sonar system.

[0068] Reference Figure 2, which shows the specific process of the weak target detection method for detecting weak targets in sonar images. First, read the sonar signal data; second, smooth the sonar signal data; third, perform Gamma modeling on the smoothed sonar signal data; fourth, detect the modeled sonar image in the constant false alarm detection window 10 (i.e., CFAR detection); fourth, cluster the binary image generated after constant false alarm detection using the DBSCAN clustering method; fifth, after calculating the geometric size of the extracted highlighted object and marking the object as a suspected target based on the geometric size, mark and number the suspected target in the region; sixth, calculate the centroid position of the suspected target and store it in an array to record the motion trajectory of the suspected target in the time series; seventh, calculate the coordinate displacement of the centroid position in adjacent frames, if the change in the centroid position of the suspected target exceeds the set motion threshold, the object is detected as a real moving target, if the change in the centroid position of the suspected target does not exceed the set motion threshold, the object is detected as a false target.

[0069] Attached Figure 7 It is a weak target detection system 20 of sonar images according to another preferred embodiment of the present invention, which includes a modeling unit 21, a design unit 22, a binarization unit 23, a clustering unit 24 and a judgment unit 25. The modeling unit 21 is used to perform Gamma modeling on the smoothed sonar image, the design unit 22 is used to design the constant false alarm detection window 10 for detecting the modeled sonar image, the binarization unit 23 is used to binarize the detected sonar image to estimate the background, the clustering unit 24 is used to cluster the binary image, and the judgment unit 25 is used to judge whether the object is a real moving target or a false target after extracting the object that meets the brightness feature in the sonar image. Through such a process, in the presence of reverberation interference, the weak target detection system 20 of the present invention can accurately detect the weak target in the sonar image to judge whether the weak target is a real moving target or a false target, and the weak target detection system 20 can provide robust detection to achieve detection of the active sonar system.

[0070] In a specific example of the weak target detection system 20 of the present invention, the modeling unit 21 further includes a smoothing module 211 and a modeling module 212, wherein the smoothing module 211 is used to smooth the sonar signal data, remove speckle noise to a certain extent, and improve the target detection, classification and filtering performance of the sonar, wherein the modeling module 212 is used to perform Gamma modeling, and the parameter Indicates that there is a target echo reflection point at the position to be detected. Indicates that there is only noise at the position to be detected, and the parameter and parameters satisfy:

[0071] ,parameter is the smoothed signal strength, parameter is the decision threshold to be determined, the smoothed signal strength In the assumption The probability density function under ,parameter is the specific value of each signal, parameter is the number of smoothed data, parameter is the scale parameter, parameter The cumulative distribution probability is , The closed form expression for .

[0072] In a specific example of the weak target detection system 20 of the present invention, the constant false alarm detection window 10 designed by the design unit 22 includes one unit to be detected 11, two reference units 12 and two protection units 13. One reference unit 12 is respectively provided on the upper side and the lower side of the unit to be detected 11. The reference unit 12 is used to count background clutter. Since the target appears as an arc in the sonar fan diagram, when designing the constant false alarm detection window 10, it is considered to provide one protection unit 13 on the opposite sides of the unit to be detected 11 to protect the unit to be detected 11 and prevent the unit to be detected 11 from falling into the reference unit 12, so as to avoid the unit to be detected 11 from participating in the calculation as background clutter and affecting the subsequent detection results. Preferably, the reference unit 12 on the upper side is located on the upper side of the two protection units 13 and the unit to be detected 11, and the reference unit 12 on the lower side is located on the lower side of the two protection units 13 and the unit to be detected 11. The mean parameter of the reference unit 12 on the upper side is , the mean parameter of the reference unit 12 on the lower side is ,parameter The mean of ,parameter The mean of ,parameter is the number of pixels in the reference unit 12 on the upper side, and the parameter is the number of pixels in the reference unit 12 at the bottom, and the parameter is the reference unit 12, parameter is the total number of pixels in the reference unit 12, which satisfies the relationship .

[0073] The background clutter is estimated to be , the decision threshold satisfy , where the parameters is the threshold factor, parameter It is calculated by the corresponding false alarm probability expression, and the calculation formula satisfies:

[0074] .

[0075] After the constant false alarm detection, the binarization unit 23 is used to process the sonar image into a binary image, wherein the highlight area in the binary image is marked as the target, and its value is 1, and the other areas are regarded as the background, and its value is 0. The contrast between the target and the background can be significantly improved by the binary image method, which is convenient for subsequent area extraction and analysis. At the same time, the distribution information of the target area in the binary image reflects the potential object features, but since there may be noise interference or false detection points in the detection process, it is necessary to cluster the binary image in the future to ensure that the target area extracted later is more accurate and complete.

[0076] In a specific example of the weak target detection system 20 of the present invention, the clustering unit 24 uses the DBSCAN clustering method to cluster the binary image generated after constant false alarm detection. In this way, the clustering unit 24 can find clusters of arbitrary shapes in the sonar data and effectively identify noise points. By analyzing the density distribution of data points, the areas with higher density are divided into the same cluster, and the points with lower density are identified as noise or isolated points. During the clustering process, the DBSCAN clustering method does not require the pre-specified number of clusters. Therefore, the weak target detection system 20 of the present invention can process complex data sets with nonlinear distribution, and is particularly suitable for extracting high-density target areas in noisy data.

[0077] In a specific example of the weak target detection system 20 of the present invention, the judgment unit 25 includes a first calculation module 251, a marking module 252, a second calculation module 253 and a judgment module 254, wherein the first calculation module 251 is used to calculate the geometric size of the extracted highlighted object, and mark the object whose geometric size is greater than the set threshold as a suspected target, wherein the marking module 252 is used to perform regional marking and numbering on the suspected target, wherein the second calculation module 253 is used to calculate the center of mass position of the suspected target and store it in an array to record the motion trajectory of the suspected target in a time series, wherein the judgment module 254 is used to analyze the change in the center of mass position of the suspected target in adjacent multiple frames of sonar images, and judge the motion state of the suspected target, wherein when the change in the center of mass position of the suspected target exceeds the set motion threshold, the object is judged to be a real moving target, and when the change in the center of mass position of the suspected target does not exceed the set motion threshold, the object is judged to be a false target. Through the above process, the weak target detection system 20 combines spatial and temporal information, which can not only improve the detection accuracy of moving targets, but also effectively reduce false detections caused by noise or stationary objects, ensuring the stability and reliability of the active sonar system.

[0078] Reference Figure 8 According to another aspect of the present invention, the present invention further provides a computing device 30, wherein the computing device 30 includes a processor 31 and a memory 32 coupled to the processor 31, the memory 32 stores instructions, and the processor 31 executes the instructions, so that the computing device 30 performs the following steps: performing Gamma modeling on the sonar image after smoothing; designing the constant false alarm detection window 10 for detecting the sonar image after modeling; binarizing the detected sonar image to estimate the background; clustering the binary image; after extracting the object that meets the brightness characteristics in the sonar image, judging whether the object is a real moving target or a false target. In this way, in the presence of reverberation interference, the computing device 30 of the present invention can accurately detect weak targets in the sonar image to judge whether the weak target is a real moving target or a false target, and the computing device 30 can provide robust detection to achieve detection of the active sonar system.

[0079] In a specific example of the computing device 30 of the present invention, the computing device 30 further includes a transceiver 33, and the transceiver 33 communicates with the processor 31 and the memory 32 through an internal connection path.

[0080] It should be understood by those skilled in the art that the embodiments of the present invention described above and shown in the accompanying drawings are only examples and do not limit the present invention. The purpose of the present invention has been fully and effectively achieved. The functional and structural principles of the present invention have been demonstrated and explained in the embodiments, and the embodiments of the present invention may be deformed or modified in any way without departing from the principles.

Claims

1. A weak target detection method for sonar images, characterized in that: The weak target detection method comprises the following steps: S1, Gamma modeling of the smoothed sonar image; S2, design a constant false alarm detection window for detecting the sonar image after modeling; S3, binarizing the detected sonar image to estimate the background; S4, clustering the binary image; S5, after extracting the object that meets the brightness feature in the sonar image, determining whether the object is a real moving target or a false target; In step S1, the parameters Indicates that there is a target echo reflection point at the position to be detected. Indicates that there is only noise at the position to be detected, where the parameter and parameters satisfy: , where the parameters is the smoothed signal strength, parameter is the decision threshold to be determined, the smoothed signal strength In the assumption The probability density function under ,parameter is the specific value of each signal, parameter is the number of smoothed data, parameter is the scale parameter, signal strength In the assumption The cumulative distribution probability under , The closed form expression for ; In step S2, the constant false alarm detection window includes a unit to be detected, protection units located on opposite sides of the unit to be detected, and reference units located on the upper and lower sides of the unit to be detected and the two protection units, wherein the mean parameter of the reference unit on the upper side is , the mean parameter of the reference unit on the lower side is ,parameter The mean of ,parameter The mean of ,parameter is the number of pixels in the reference unit on the upper side, and the parameter is the number of pixels in the reference unit on the lower side, and the parameter is the reference unit, parameter is the total number of pixels in the reference unit, which satisfies the relationship , wherein in step S3, the background clutter estimation value is , the decision threshold satisfy , where the parameters is the threshold factor, parameter It is calculated by the corresponding false alarm probability expression, and the calculation formula satisfies: 。 2. The weak target detection method of sonar images according to claim 1, characterized in that: In step S4, the DBSCAN algorithm is used to cluster the binary image.

3. The weak target detection method of sonar images according to claim 1, characterized in that: The step S5 further comprises: S5.1, calculating the geometric size of the extracted highlighted object, and marking the object whose geometric size is larger than a set threshold as a suspected target; S5.2, mark and number the suspected targets; S5.3, calculating the center of mass position of the suspected target and storing it in an array to record the movement trajectory of the suspected target in the time series; S5.4, analyze the change in the center of mass position of the suspected target in multiple adjacent frames of sonar images to determine the motion state of the suspected target, wherein when the change in the center of mass position of the suspected target exceeds a set motion threshold, the object is determined to be a real moving target; when the change in the center of mass position of the suspected target does not exceed the set motion threshold, the object is determined to be a false target.

4. The weak target detection method of sonar images according to claim 2, characterized in that: The step S5 further comprises: S5.1, calculating the geometric size of the extracted highlighted object, and marking the object whose geometric size is larger than a set threshold as a suspected target; S5.2, mark and number the suspected targets; S5.3, calculating the center of mass position of the suspected target and storing it in an array to record the movement trajectory of the suspected target in the time series; S5.4, analyze the change in the center of mass position of the suspected target in multiple adjacent frames of sonar images to determine the motion state of the suspected target, wherein when the change in the center of mass position of the suspected target exceeds a set motion threshold, the object is determined to be a real moving target; when the change in the center of mass position of the suspected target does not exceed the set motion threshold, the object is determined to be a false target.

5. A weak target detection system for sonar images, characterized in that: include: A modeling unit, used for performing Gamma modeling on the sonar image after smoothing; A design unit, used for designing a constant false alarm detection window for detecting the sonar image after modeling; A binarization unit is used to perform binarization processing on the detected sonar image to estimate the background; A clustering unit, used for clustering binary images; A judgment unit, used to judge whether the object is a real moving target or a false target after extracting the object that meets the brightness characteristics in the sonar image; The modeling unit includes a smoothing module and a modeling module, wherein the smoothing module is used to smooth the sonar signal data, and the modeling module is used to perform Gamma modeling, wherein the parameter Indicates that there is a target echo reflection point at the position to be detected. Indicates that there is only noise at the position to be detected, where the parameter and parameters satisfy: , where the parameters is the smoothed signal strength, parameter is the decision threshold to be determined, the smoothed signal strength In the assumption The probability density function under ,parameter is the specific value of each signal, parameter is the number of smoothed data, parameter is the scale parameter, signal strength In the assumption The cumulative distribution probability under , The closed form expression for ; The constant false alarm detection window includes a unit to be detected, protection units located on opposite sides of the unit to be detected, and reference units located on the upper and lower sides of the unit to be detected and the two protection units, wherein the mean parameter of the reference unit on the upper side is , the mean parameter of the reference unit on the lower side is ,parameter The mean of ,parameter The mean of ,parameter is the number of pixels in the reference unit on the upper side, and the parameter is the number of pixels of the reference unit on the lower side, parameter is the reference unit, parameter is the total number of pixels in the reference unit, which satisfies the relationship , where the background clutter estimate is , the decision threshold satisfy , where the parameters is the threshold factor, parameter It is calculated by the corresponding false alarm probability expression, and the calculation formula satisfies: 。 6. The weak target detection system of sonar images according to claim 5, characterized in that: The clustering unit clusters the binary images using the DBSCAN algorithm.

7. The weak target detection system of sonar images according to claim 5, characterized in that: The judgment unit includes a first calculation module, a marking module, a second calculation module and a judgment module, wherein the first calculation module is used to calculate the geometric size of the extracted highlighted object and mark the object whose geometric size is greater than a set threshold as a suspected target; the marking module is used to mark and number the suspected target in an area; The second calculation module is used to calculate the center of mass position of the suspected target and store it in an array to record the motion trajectory of the suspected target in a time series; the judgment module is used to analyze the change in the center of mass position of the suspected target in adjacent multiple frames of sonar images to judge the motion state of the suspected target, wherein when the change in the center of mass position of the suspected target exceeds a set motion threshold, the object is judged to be a real moving target, and when the change in the center of mass position of the suspected target does not exceed the set motion threshold, the object is judged to be a false target.

8. The weak target detection system of sonar images according to claim 6, characterized in that: The judgment unit includes a first calculation module, a marking module, a second calculation module and a judgment module, wherein the first calculation module is used to calculate the geometric size of the extracted highlighted object and mark the object whose geometric size is greater than a set threshold as a suspected target; the marking module is used to mark and number the suspected target in an area; The second calculation module is used to calculate the center of mass position of the suspected target and store it in an array to record the motion trajectory of the suspected target in a time series; the judgment module is used to analyze the change in the center of mass position of the suspected target in adjacent multiple frames of sonar images to judge the motion state of the suspected target, wherein when the change in the center of mass position of the suspected target exceeds a set motion threshold, the object is judged to be a real moving target, and when the change in the center of mass position of the suspected target does not exceed the set motion threshold, the object is judged to be a false target.

9. A computing device, characterized in that The computing device comprises a processor and a memory coupled to the processor, wherein the memory stores codes, and the processor is configured to execute the codes. When the codes are executed, the computing device executes the weak target detection method according to any one of claims 1 to 4.

Images