Method and system for detecting moving targets from multi-frame echograms of a mobile platform

Abstract

The present invention belongs to the technical field of underwater acoustic target tracking and positioning, and relates to a method and system for detecting a mobile target from a multi-frame echo map of a mobile platform, so as to effectively suppress strong pixel points caused by reverberation and clutter in an active sonar echo map on a mobile platform under a complex working environment. The method comprises obtaining t frames of active sonar echo maps, converting the t frames of active sonar echo maps into a geodetic coordinate system, and arranging them in time to obtain a three-dimensional time sequence map represented by a three-dimensional matrix. A time window of size m×n×l is slid frame by frame on the three-dimensional time sequence map to obtain multiple active sonar echo maps with a time window frame number of l frames arranged in time order within the time window, so as to obtain a four-dimensional matrix. The first order k is configured, and the first high-order time gap degree calculation is performed to obtain a high-order three-dimensional feature matrix. The second order k' is configured, and the second high-order time gap degree calculation is performed to obtain a two-dimensional multiple high-order feature matrix, that is, to obtain an underwater mobile target extraction result.

Description

Technical Field

[0001] The invention belongs to the technical field of sonar detection and target recognition, and in particular relates to a method and a system for detecting a moving target from a multi-frame echo map of a mobile platform. Background Art

[0002] In recent years, an important goal of active sonar system design is to detect and track underwater targets (the target may be an unmanned underwater robot or a submarine, etc.). In practical applications, a mobile platform is usually equipped with sonar, which emits sound waves to "irradiate" the underwater target. After the underwater target receives the sound waves, it reflects them to the mobile platform, that is, the mobile platform receives the echo time and echo parameters reflected by the underwater target to determine the parameters of the underwater target, thereby realizing the detection and identification of the underwater target. Considering the complex operating waters of the mobile platform, it is more effective to obtain the target sonar echo map through high-frequency active sonar. However, on the sonar echo map, since the working area of ​​the mobile platform is not fixed, the high-intensity clutter caused by complex geology and bubble fish often submerges the target and causes false alarms. Therefore, it is necessary to design a corresponding method for distinguishing targets from clutter for the mobile platform.

[0003] The literature on extracting moving target features from sonar echo maps provided by the prior art points out that the two-dimensional sonar echo map is expanded into a three-dimensional sonar echo map, and then the high-order feature map is obtained by using the high-order time interval characteristics, so as to realize the dynamic characteristic characterization of the underwater target. However, the underwater environment is slowly time-varying, and the detection area of ​​the mobile platform will change over time, resulting in the failure of the spatiotemporal pixel matching of the original underwater target, where the spatiotemporal pixel refers to the element in the vector extracted by the pixel window. Further analysis of the reason for the failure of spatiotemporal pixel matching is that since the mobile platform is in a state of continuous movement, the corresponding pixels in different frames are not in the same vector, resulting in the failure of spatiotemporal pixel matching. And due to the complexity and variability of the working sea area, the reverberation under different perspectives has different representations, so the method provided by the prior art cannot adapt to the effective suppression of reverberation and clutter pixels by the mobile platform when detecting moving targets. Summary of the invention

[0004] The purpose of the present invention is to address the problem that the above-mentioned prior art is insufficient in suppressing reverberation and clutter pixels when detecting moving targets in active sonar echo maps on mobile platforms, resulting in failure of spatiotemporal pixel matching. A method and system for detecting moving targets from multi-frame echo maps of mobile platforms are proposed, which can effectively suppress strong pixels caused by reverberation and clutter in active sonar echo maps on mobile platforms under complex working environments.

[0005] In a first aspect, the present invention provides a method for detecting a moving target from a multi-frame echo map of a mobile platform, the method comprising the following steps:

[0006] S10. The mobile platform obtains t consecutive frames of active sonar echo images from different positions, and transforms the t consecutive frames of active sonar echo images into a geodetic coordinate system based on the mobile platform's own position coordinates and detection direction, so as to unify the corresponding pixel points in the t consecutive frames of active sonar echo images obtained by the mobile platform at different positions into the geodetic coordinate system, intercept a common detection area, obtain multiple two-dimensional matrices of active sonar echo images in the same detection area, and sort the two-dimensional matrices in chronological order to obtain a three-dimensional time series diagram represented by a three-dimensional matrix;

[0007] S11. The three-dimensional time sequence diagram represented by the three-dimensional matrix is ​​slid frame by frame on the three-dimensional time sequence diagram in a time window of size m×n×l along the time frame direction to obtain a plurality of active sonar echo diagrams with a time window frame number of l in the time window in time order, so as to obtain a four-dimensional matrix; the four-dimensional matrix includes a plurality of three-dimensional matrices, each of which corresponds to one of a plurality of consecutive l-frame active sonar echo diagrams, wherein 2≤l≤t-1, and m×n is a two-dimensional plane of the three-dimensional matrix;

[0008] S12. Configure the value range of the first order k, slide multiple 1×1×1 pixel windows on the three-dimensional matrix in the four-dimensional matrix along the two-dimensional plane direction of the three-dimensional matrix to extract local pixel window vectors, and start from the initial value of the first order k, perform the first high-order time gap calculation in the pixel window, and form a high-order three-dimensional feature matrix with the obtained first high-order time gap result according to the three-dimensional original pixel position of the 1×1×1 pixel window;

[0009] S13. Configure the value range of the second order k′, slide on the high-order three-dimensional feature matrix with 1×1×(t―l+1) pixel windows respectively along the two-dimensional plane direction of the high-order three-dimensional feature matrix to extract the global pixel window vector, and start from the initial value of the second order k′, perform the second high-order time gap calculation in the pixel window of 1×1×(t―l+1), and arrange the second high-order time gap results in all pixel windows according to the two-dimensional original pixel positions of the pixel window of 1×1×(t―l+1) to form a two-dimensional multiple high-order feature matrix, which is the underwater moving target extraction result;

[0010] S14. Execute S11 to S13 to traverse all time window frame numbers l, first order k and second order k′ to obtain multiple moving target extraction results;

[0011] S15. Obtain an optimal moving target extraction result from multiple moving target extraction results.

[0012] As a possible implementation manner, the S12 includes the following steps:

[0013] S120. Slide pixel by pixel in each three-dimensional matrix of the four-dimensional matrix using a plurality of pixel windows of size 1×1×1 to extract local pixel window vectors respectively;

[0014] S121. Set the first order k to 2, and calculate the first high-order time gap; specifically: use the first high-order time gap to calculate the first high-order time gap of the local pixel window vector, and arrange the two-dimensional original pixel positions of the pixel window to obtain the two-dimensional high-order feature map corresponding to each pixel window;

[0015] S122. Arrange the two-dimensional high-order feature maps corresponding to each of the pixel windows calculated in S121 in chronological order to form the high-order three-dimensional feature matrix.

[0016] As a possible implementation manner, the S13 includes the following steps:

[0017] S130. Use a 1×1×(t―l+1) pixel window to slide on the two-dimensional plane of the high-order three-dimensional feature matrix to extract a global pixel window vector;

[0018] S131. Calculate the high-order features of the global pixel window vector using the second high-order time interval and arrange them according to the two-dimensional original pixel positions to obtain the two-dimensional multiple high-order feature matrix, which is the moving target extraction result.

[0019] As a possible implementation method, the first high-order time gap degree is calculated using the following formula:

[0020]

[0021] in, is the first high-order temporal gap degree; the local pixel window vector extracted by the 1×1×l pixel window is expressed as P l (i,j),P l (i,j) time domain expectation m×n is the two-dimensional plane of the multi-frame active sonar echo map; 2≤l≤t-1.

[0022] As a possible implementation method, the second high-order time gap degree is calculated using the following formula:

[0023]

[0024] Among them, the global pixel window vector extracted by the 1×1×(t―l+1) pixel window is expressed as Time Domain Expectation m×n is the two-dimensional plane of the multi-frame active sonar echo map; 2≤l≤t-1.

[0025] As a possible implementation manner, 1≤k≤10, 2≤k′≤10, 2≤l≤t-1.

[0026] In a second aspect, the present invention also provides a system for detecting a mobile target from a multi-frame echo map of a mobile platform, characterized in that the system includes a processor and a communication interface coupled to the processor; the processor is used to run a computer program or instructions to implement the method for detecting a mobile target from a multi-frame echo map of a mobile platform described in the first aspect.

[0027] Compared with the prior art, the present invention has the following technical effects:

[0028] 1. The method provided by the present invention is based on the slow time-varying assumption of the underwater environment. During the target recognition process, unlike the existing target recognition method which uses high-order time gap calculation, this method calculates multiple time gaps. Compared with the existing technology, it can better suppress the slowly changing reverberation and clutter between consecutive frames, thereby realizing the extraction of moving targets.

[0029] 2. The time window frame number in the method provided by the present invention is 1, and the window length can be adjusted according to the different degrees of underwater environmental changes, so that the mobile platform can adapt to the detection of moving targets under different sea conditions.

[0030] 3. The method provided by the present invention introduces high-order time gap calculation. The order selection focus of the first high-order time gap calculation is the inhibitory effect of the intensity fluctuation of corresponding pixel points in adjacent frames. The order selection focus of the second high-order time gap calculation is the correlation extraction of outliers in the global high-order three-dimensional feature matrix. Different order values ​​can be selected for different action areas.

[0031] 4. The method provided by the present invention utilizes the optimal feature image to analyze the target area, and can obtain more reliable detection results. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0033] Figure 1 A flow chart of a method for detecting a moving target from a multi-frame echo map of a mobile platform provided by an embodiment of the present invention;

[0034] Figure 2 The high-order temporal gap feature extraction results provided by the prior art;

[0035] Figure 3 The multiple high-order time gap feature extraction results provided by the embodiment of the present invention;

[0036] Figure 4 The optimal result obtained by using the maximum between-class variance method provided by the embodiment of the present invention;

[0037] Figure 5 is the ROC curve. DETAILED DESCRIPTION

[0038] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0039] It should be noted that when an element is referred to as being "fixed to" or "disposed on" another element, it can be directly on the other element or indirectly on the other element. When an element is referred to as being "connected to" another element, it can be directly connected to the other element or indirectly connected to the other element.

[0040] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined. The meaning of "several" is one or more, unless otherwise clearly and specifically defined.

[0041] In the description of the present invention, it is necessary to understand that the directions or positional relationships indicated by the terms "up", "down", "front", "back", "left", "right", etc. are based on the directions or positional relationships shown in the accompanying drawings, and 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 direction, be constructed and operated in a specific direction, and therefore cannot be understood as a limitation on the present invention.

[0042] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0043] First, see Figure 1The present invention provides a method for detecting a mobile target from a multi-frame echo map of a mobile platform. A sonar is mounted on the mobile platform. The sound waves emitted by the sonar are radiated to the underwater mobile target. The mobile target receives the sound waves and reflects the sound waves back to the mobile platform. Array signal processing is performed on the reflected sound waves to calculate and generate an active sonar echo map. Multiple sound wave emissions and sound wave reflections can calculate and obtain multi-frame active sonar echo maps. The method includes the following steps:

[0044] S10. The mobile platform obtains t consecutive frames of active sonar echo images from different positions, and transforms the t consecutive frames of active sonar echo images into a geodetic coordinate system based on the mobile platform's own position coordinates and detection direction, so as to unify the corresponding pixel points in the t consecutive frames of active sonar echo images obtained by the mobile platform at different positions into the geodetic coordinate system, intercept the common detection area, obtain multiple two-dimensional matrices of active sonar echo images in the same detection area, and sort the two-dimensional matrices in chronological order to obtain a three-dimensional time sequence diagram represented by a three-dimensional matrix. The number of frames of the active sonar echo image is determined according to the speed of the mobile platform and the degree of overlap of the detection area, and is not specifically limited here.

[0045] As an example, the continuous t-frame active sonar echo map is transformed into the geodetic coordinate system based on the position and motion trajectory of the mobile platform itself to obtain a two-dimensional matrix Arrange the two-dimensional matrix according to time to obtain a three-dimensional matrix of active sonar echo images in the same detection area

[0046] In the specific implementation of S10, the following method is used to transform the continuous t-frame active sonar echo map into the geodetic coordinate system:

[0047] For the same detection area, at time T1, the mobile platform is at position α(x1, y1) to obtain the active sonar echo map A. In the active sonar echo map A, any pixel a(r a ,θ a ), and obtain the corresponding pixel b(r) in the active sonar echo map B at the position β(x2,y2) at time T2 b ,θ b ), with the corresponding relationship:

[0048] x a =r a ×cos(θ a )

[0049] y a =r a ×sin(θ a )

[0050] x b =x a+(x2―x1)

[0051] y b =y a +(y2―y1)

[0052]

[0053]

[0054] By using the above coordinate transformation, the corresponding pixel points in the active sonar echo map obtained by the mobile platform at different positions can be unified into a coordinate system (i.e., the earth coordinate system) to obtain multiple two-dimensional matrices: The two-dimensional matrix Arranged by time, we get the expanded three-dimensional matrix V 1～t =[V1…V i …V t ],

[0055] S11. The three-dimensional time series diagram represented by the three-dimensional matrix is ​​slid frame by frame on the three-dimensional time series diagram in the time frame direction with a time window of size m×n×l, and a plurality of active sonar echo images with a time window frame number of l in the time window are obtained in chronological order, so as to obtain a four-dimensional matrix; the four-dimensional matrix includes a plurality of three-dimensional matrices, each of which corresponds to one of a plurality of consecutive l-frame active sonar echo images, wherein 2≤l≤t-1, and m×n is the two-dimensional plane of the three-dimensional matrix.

[0056] When S11 is implemented, the active sonar echo images of consecutive l frames are saved frame by frame as a three-dimensional matrix V in a time window of size m×n×l. 1～t , the three-dimensional matrix V 1～t is expanded to a four-dimensional matrix W, W = [V 1～l V 2～l+1 …V i～i+l-1 …V t―l+1～t ],

[0057] Four-dimensional matrix, that is, W = [V 1～l V 2～l+1 …V i～i+l-1 …V t―l+1～t ], The generation process is as follows:

[0058] Slide frame by frame with a time window of size m×n×l, and V 1～t =[V1…V i …V t ], Expanded to W = [V 1～l …Vi～i+l-1 …V t―l+1～t ], V i～i+l-1 =[V i …V j …V i+l―1 ], 2≤l≤t―1.

[0059] S12. Configure the value range of the first order k, slide multiple 1×1×1 pixel windows on the three-dimensional matrix in the four-dimensional matrix along the two-dimensional plane direction of the three-dimensional matrix to extract the local pixel window vector, and start from the initial value of the first order k, perform the first high-order time gap calculation in the pixel window, and form a high-order three-dimensional feature matrix with the first high-order time gap result according to the three-dimensional original pixel position of the 1×1×1 pixel window. The three-dimensional original pixel position includes the coordinates in the m×n two-dimensional plane, as well as the previous and next frames.

[0060] As an example, starting from k=2, with multiple pixel windows of size 1×1×1, the three-dimensional matrix V in the four-dimensional matrix W is i～i+l-1 , i=1,…,t―l+1 slides up, and performs the first high-order time gap calculation. The k-order time gap values ​​in all pixel windows form a high-order three-dimensional feature matrix

[0061] In specific implementation, S12 includes:

[0062] S120. Slide pixel by pixel in each three-dimensional matrix of the four-dimensional matrix using a plurality of pixel windows of size 1×1×1 (sliding pixel by pixel means sliding from top to bottom and from left to right) to extract local pixel window vectors respectively;

[0063] S121. Set the first order k to 2 and calculate the first high-order time interval degree; specifically: use the first high-order time interval degree to calculate the first high-order time interval degree of the local pixel window vector, and arrange the two-dimensional original pixel positions of the pixel window to obtain the two-dimensional high-order feature map corresponding to each pixel window.

[0064] As an example, the local pixel window vector P is calculated using the first high-order temporal interval l High-order features of (i,j) And arrange according to the original pixel position to obtain a two-dimensional high-order feature map The calculation formula for the first high-order time gap is:

[0065]

[0066] Where P l (i,j) time domain expectation

[0067] S122. Arrange the two-dimensional high-order feature maps corresponding to each pixel window calculated in S121 in chronological order to form a high-order three-dimensional feature matrix.

[0068] As an example, the two-dimensional high-order feature maps in each time window calculated in the previous step are Arrange in order of starting time to form a high-order three-dimensional feature matrix

[0069] S13. Configure the value range of the second order k', slide the high-order three-dimensional feature matrix with 1×1×(t-l+1) pixel windows along the two-dimensional plane direction of the high-order three-dimensional feature matrix to extract the global pixel window vector, and start from the initial value of the second order k', perform the second high-order time gap calculation in the 1×1×(t-l+1) pixel window, and arrange the second high-order time gap results in all pixel windows according to the two-dimensional original pixel position of the 1×1×(t-l+1) pixel window to form a two-dimensional multiple high-order feature matrix, which is the underwater mobile target extraction result. Among them, the two-dimensional original pixel position refers to the coordinate in the m×n two-dimensional plane.

[0070] As an example, starting from k′=2, with a pixel window of size 1×1×(t―l+1), in the three-dimensional feature matrix L k Slide pixel by pixel, and then perform the second high-order time gap calculation. The k′-order time gap values ​​of all pixels form a two-dimensional multiple high-order feature matrix include:

[0071] S130. Use a 1×1×(t―l+1) pixel window to slide on the two-dimensional plane of the high-order three-dimensional feature matrix to extract a global pixel window vector.

[0072] As an example, in the three-dimensional matrix Extract all pixel window vectors using a 1×1×(t―l+1) pixel window on the m×n plane to obtain a global pixel window vector. For example, for each pixel (i, j) on the m×n plane, 1≤i≤m, 1≤j≤n, extract a global pixel window vector with a length of t―l+1 along the third dimension t―l+1.

[0073] S131. Calculate the high-order features of the global pixel window vector using the second high-order temporal gap and arrange them according to the two-dimensional original pixel positions to obtain a two-dimensional multiple high-order feature matrix, which is the moving target extraction result.

[0074] As an example, the second high-order time gap calculation vector The high-order features are arranged according to the two-dimensional original pixel positions to obtain a high-order feature map The second high-order time gap calculation formula is:

[0075]

[0076] Among them, the global pixel window vector extracted by the 1×1×(t―l+1) pixel window is expressed as Time Domain Expectation m×n is a two-dimensional plane matrix of multi-frame active sonar echo images; 2≤l≤t-1.

[0077] S14. Execute S10 to S13 to traverse all time window frame numbers l, first order k and second order k′ to obtain multiple moving target extraction results.

[0078] S15. Obtain an optimal moving target extraction result from the plurality of moving target extraction results.

[0079] As an example, the maximum inter-class variance method is used to perform automatic threshold search on each moving target extraction result, and the difference between the maximum pixel value and the search threshold is used to measure the optimal target extraction result.

[0080] Compared with the prior art, the present invention has the following technical effects:

[0081] 1. The method provided by the present invention is based on the slow time-varying assumption of the underwater environment. During the target recognition process, unlike the existing target recognition method which uses high-order time gap calculation, this method calculates multiple time gaps. Compared with the existing technology, it can better suppress the slowly changing reverberation and clutter between consecutive frames, thereby realizing the extraction of moving targets.

[0082] 2. The time window frame number in the method provided by the present invention is 1, and the window length can be adjusted according to the different degrees of underwater environmental changes, so that the mobile platform can adapt to the detection of moving targets under different sea conditions.

[0083] 3. The method provided by the present invention introduces high-order time gap calculation. The order selection focus of the first high-order time gap calculation is the inhibitory effect of the intensity fluctuation of corresponding pixel points in adjacent frames. The order selection focus of the second high-order time gap calculation is the correlation extraction of outliers in the global high-order three-dimensional feature matrix. Different order values ​​can be selected for different action areas.

[0084] 4. The method provided by the present invention utilizes the optimal feature image to analyze the target area, and can obtain more reliable detection results.

[0085] See also Figures 2 to 4,To prove the effectiveness of the present invention, experiments are conducted on a ,motion image sequence containing 80 frames of active sonar echo images. ,Due to the influence of environmental noise and reverberation, the target is submerged in the clutter. Figure 2 This is the prior art high-order time gap feature extraction result when k=4. It can be seen that, in addition to the results extracted in the form of trajectories, some clutter is not completely suppressed. Figure 3 This is the result of multiple high-order feature extraction with l=3, k=4, k′=2. It can be seen that while the feature results are extracted, clutter is effectively suppressed. In addition, the clutter suppression effect is different in different areas, which shows the influence of clutter characteristics corresponding to different sea areas on the clutter suppression effect. Figure 4 This is the result of multiple high-order feature extraction with l=5, k=4, k′=2. It can be seen that with the increase of feature order, the image background becomes cleaner.

[0086] See also Figure 5 ,Compared with the high-order time gap results, the ROC curve corresponding to the multiple high-order time gap results is closer to the upper left corner and away from the diagonal line, so the AUC value corresponding to the multiple high-order time gap results is higher, indicating that this method can better extract the moving target from the clutter and reverberation background.

[0087] Those skilled in the art can understand from the above description that all or part of the steps in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a program, and the program is stored in a storage medium, including several instructions for making a device (which can be a single-chip microcomputer, chip, etc.) or a processor (processor) execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes but is not limited to various media that can store program codes, such as USB flash drives, mobile hard disks, magnetic storage devices, and optical storage devices.

[0088] It can be seen from the above description that the embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. The above is only a preferred embodiment of the present invention, and the present invention should not be limited to the contents disclosed in the embodiment and the accompanying drawings. Any equivalent or modification that is completed without departing from the spirit disclosed in the present invention falls within the scope of protection of the present invention.

[0089] In the description of the above embodiments, specific features, structures, materials or characteristics may be combined in a suitable manner in any one or more embodiments or examples.

[0090] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for detecting a moving target from a multi-frame echo map of a mobile platform, characterized in that: The method comprises the following steps: S10. The mobile platform acquires t consecutive frames of active sonar echo images from different positions, and transforms the t consecutive frames of active sonar echo images into a geodetic coordinate system based on the position coordinates and detection direction of the mobile platform itself, so as to unify the corresponding pixel points in the t consecutive frames of active sonar echo images acquired by the mobile platform at different positions into the geodetic coordinate system, intercept a common detection area, obtain multiple two-dimensional matrices of active sonar echo images in the same detection area, and sort the two-dimensional matrices in chronological order to obtain a three-dimensional time series diagram represented by a three-dimensional matrix; S11. The three-dimensional time sequence diagram represented by the three-dimensional matrix is ​​slid frame by frame on the three-dimensional time sequence diagram in a time window of size m×n×l along the time frame direction to obtain a plurality of active sonar echo diagrams with a time window frame number of l in the time window in chronological order, so as to obtain a four-dimensional matrix; the four-dimensional matrix comprises a plurality of three-dimensional matrices, each of the three-dimensional matrices corresponds to one of the plurality of consecutive l-frame active sonar echo diagrams, wherein 2≤l≤t-1, and m×n is a two-dimensional plane of the three-dimensional matrix; S12. Configure the value range of the first order k, slide on the three-dimensional matrix in the four-dimensional matrix with multiple 1×1×1 pixel windows respectively along the two-dimensional plane direction of the three-dimensional matrix to extract local pixel window vectors, and start from the initial value of the first order k, perform the first high-order time gap calculation in the 1×1×1 pixel window respectively, and form a high-order three-dimensional feature matrix with the obtained first high-order time gap result according to the three-dimensional original pixel position of the 1×1×1 pixel window; S13. Configure the value range of the second order k′, slide on the high-order three-dimensional feature matrix with 1×1×(t-l+1) pixel windows respectively along the two-dimensional plane direction of the high-order three-dimensional feature matrix to extract the global pixel window vector, and start from the initial value of the second order k′, perform the second high-order time gap calculation in the 1×1×(t-l+1) pixel window respectively, and arrange the second high-order time gap results in all pixel windows according to the two-dimensional original pixel positions of the 1×1×(t-l+1) pixel window to form a two-dimensional multiple high-order feature matrix, which is the underwater moving target extraction result; S14. Execute S11 to S13 to traverse all the time window frame numbers l, the first order k and the second order k′ to obtain a plurality of the mobile target extraction results; S15. Obtain an optimal moving target extraction result from the plurality of moving target extraction results.

2. The method for detecting a moving target from a multi-frame echogram of a mobile platform according to claim 1, characterized in that: The S12 comprises the following steps: S120. Slide pixel by pixel in each three-dimensional matrix of the four-dimensional matrix using a plurality of pixel windows of size 1×1×1 to extract local pixel window vectors respectively; S121. Set the first order k to 2, and calculate the first high-order time gap; specifically: use the first high-order time gap to calculate the first high-order time gap of the local pixel window vector, and arrange the two-dimensional original pixel positions of the pixel window to obtain the two-dimensional high-order feature map corresponding to each pixel window; S122. Arrange the two-dimensional high-order feature maps corresponding to each of the pixel windows calculated in S121 in chronological order to form the high-order three-dimensional feature matrix.

3. The method for detecting a moving target from a multi-frame echo map of a mobile platform according to claim 2, characterized in that: The S13 comprises the following steps: S130. Use a 1×1×(t―l+1) pixel window to slide on the two-dimensional plane of the high-order three-dimensional feature matrix to extract a global pixel window vector; S131. Calculate the high-order features of the global pixel window vector using the second high-order time interval and arrange them according to the two-dimensional original pixel positions to obtain the two-dimensional multiple high-order feature matrix, which is the moving target extraction result.

4. The method for detecting a moving target from a multi-frame echogram of a mobile platform according to claim 2, characterized in that: The first high-order time gap degree is calculated using the following formula: in, is the first high-order temporal gap degree; the local pixel window vector extracted by the 1×1×l pixel window is expressed as P l (i,j),P l (i,j) time domain expectation 1≤i≤m, 1≤j≤n, m×n is the two-dimensional plane of the multi-frame active sonar echo map; 2≤l≤t-1.

5. The method for detecting a moving target from a multi-frame echo map of a mobile platform according to claim 3, characterized in that: The second high-order time gap degree is calculated using the following formula: Among them, the global pixel window vector extracted by the 1×1×(t―l+1) pixel window is expressed as Time Domain Expectation 1≤i≤m, 1≤j≤n, m×n is the two-dimensional plane of the multi-frame active sonar echo map; 2≤l≤t-1.

6. The method for detecting a moving target from a multi-frame echogram of a mobile platform according to claim 1, characterized in that: 1≤k≤10, 2≤k′≤10, 2≤l≤t―1.

7. A system for detecting a moving target from a multi-frame echo map of a mobile platform, characterized in that: The system includes a processor and a communication interface coupled to the processor; the processor is used to run a computer program or instruction to implement the method for detecting a moving target from a multi-frame echo map of a mobile platform as described in any one of claims 1-6.

Images