An underwater multi-target tracking method

Abstract

The present application relates to a kind of underwater multi-target tracking methods, comprising the following steps: establishing three-dimensional space coordinate system, obtaining the coordinate of sensor, monitoring water volume, water surface in the coordinate Z2 of Z axis, the coordinate Z3 of water bottom in the Z axis, clutter density λ;Sensor periodically obtains measurement information, the measurement information includes azimuth, elevation and the time delay between direct wave, the set of measurement information in the i frame is recorded as Z (i), the number of measurement information that sensor obtains single target is L;Randomly generate K in monitoring water volume Virtual target c K =[c1,c2,...,c K ] In prior formula, the possibility that a measurement information comes from certain propagation path is expressed;According to log likelihood ratio, establish underwater target model;According to underwater target model, the actual target number and the optimal coordinate of each actual target are obtained. Efficiently positioning and tracking underwater multi-target can be carried out.

Description

Technical Field

[0001] This invention relates to the field of underwater detection technology, specifically to an underwater multi-target tracking method. Background Technology

[0002] With the development of technology, underwater activities using manned or unmanned underwater vehicles are becoming more and more frequent, making the monitoring of underwater vehicles particularly important.

[0003] When tracking and locating targets such as unmanned underwater vehicles underwater, there are often cases of missed detections and false detections due to underwater noise interference and multipath effects.

[0004] Therefore, accurate tracking and positioning of multiple underwater targets in noisy and multipath environments has become an urgent technical problem to be solved. Summary of the Invention

[0005] In view of this, the present invention provides an underwater multi-target tracking method that can efficiently locate and track multiple underwater targets.

[0006] The technical solution adopted in this invention is as follows:

[0007] An underwater multi-target tracking method includes the following steps:

[0008] S100. Establish a three-dimensional spatial coordinate system, obtain the sensor coordinates (x0, y0, z0), monitor the water volume V, the water surface coordinate Z2 on the Z-axis, the bottom coordinate Z3 on the Z-axis, and the clutter density λ.

[0009] S200. The sensor periodically acquires measurement information, including azimuth angle θ and elevation angle. The time delay τ between the sensor and the direct wave, the set of measurement information in the i-th frame is denoted as Z(i), the number of measurement information acquired by the sensor for a single target is L, i is any positive integer, and the j-th measurement information in the i-th frame is denoted as z. j (i), the total number of measurement information in the i-th frame is m. i ;

[0010] S300. Randomly generate K virtual targets c within the monitored water volume. K =[c1,c2,...,c K ], and according to the prior formula, represent the probability that a measurement information comes from a certain propagation path, the formula is expressed as:

[0011]

[0012]

[0013] in, For target ck The detection probability π generated by path l for measuring information. 00 To measure the probability that the information originates from clutter; π kl The measurement information for path l originates from target c. k The probability of;

[0014] S400. Based on the log-likelihood ratio, an underwater target model is established, expressed as follows:

[0015]

[0016] Where, p l (z j (i)|c k )=N[z j (i); h l (c k ,c s [i)] represents the target c k Measurement z is generated via path l. j The likelihood function of (i); c s The coordinates of the sensor are (x0, y0, z0), c k Represents the coordinates (x, y) of the kth virtual target k y k , z k ), h l (·) represents the model of the measurement information corresponding to the l-th path;

[0017] S500: Based on the underwater target model, obtain the actual number of targets and the optimal coordinates of each actual target.

[0018] Preferably, the measurement information of a single target is L=3, which are the direct wave path, the water surface reflected wave path, and the bottom reflected wave path, respectively;

[0019] in,

[0020] The measurement information corresponding to the direct wave path is expressed as follows:

[0021] The measurement information corresponding to the water surface reflected wave path is expressed as follows:

[0022] The measurement information corresponding to the underwater reflected wave path is expressed as follows:

[0023] Preferably,

[0024]

[0025]

[0026] τ1(ck ) = 0

[0027]

[0028]

[0029]

[0030]

[0031]

[0032]

[0033] Where c is the speed of sound underwater.

[0034] Preferably, in step S100, the azimuth variance μ, elevation variance ω, and time delay variance υ are also obtained.

[0035] The sensor is a vector hydrophone.

[0036] The azimuth variance and the time delay variance are obtained directly from the sensor;

[0037] The pitch angle variance ω of the direct wave path 直达 satisfy

[0038] ω 直达 ~GMM=0.6*N(0,1^2)+0.4*N(0,4^2),

[0039] The pitch angle variance ω of the water surface reflected wave path and the water bottom reflected wave path 多径 satisfy

[0040] ω 多径 ~GMM=0.4*N(0,1^2)+0.6*N(0,4^2);

[0041] The measurement information corresponding to the direct wave path is expressed as follows:

[0042]

[0043] The measurement information corresponding to the water surface reflected wave path is expressed as follows:

[0044]

[0045] The measurement information corresponding to the underwater reflected wave path is expressed as follows:

[0046]

[0047] Preferably, in step S200, the sensor continuously acquires Nw frames of data;

[0048] In step S400, an underwater target model is established based on the log-likelihood ratio, expressed as:

[0049]

[0050] The beneficial effects of this invention are:

[0051] This invention acquires actual measurement information through sensors, then randomly generates K virtual targets, each of which can generate L virtual measurement information. These virtual measurement information are compared with the actual measurement information to obtain the detection probability and clutter probability of each virtual measurement information. Based on this, an underwater target model is established, thereby obtaining the number of underwater targets and the optimal coordinates of each target, realizing the localization and tracking of multiple underwater targets in a noisy environment. Attached Figure Description

[0052] The above and other objects, features and advantages of the present invention will become clearer from the following description of embodiments of the invention with reference to the accompanying drawings, in which:

[0053] Figure 1 This is a schematic diagram showing how the actual target reaches the sensor through three propagation paths;

[0054] Figure 2 This is a schematic diagram of three measurement information acquired by the sensor within one frame;

[0055] Figure 3 It is a state diagram obtained from an underwater target model. Detailed Implementation

[0056] The present invention is described below based on embodiments, but the present invention is not limited to these embodiments. In the following detailed description of the present invention, some specific details are described in detail, but well-known methods, processes, procedures, and elements are not described in detail in order to avoid obscuring the essence of the present invention.

[0057] Furthermore, those skilled in the art should understand that the accompanying drawings provided herein are for illustrative purposes only and are not necessarily drawn to scale.

[0058] Unless the context explicitly requires it, the words "comprising," "including," and similar terms throughout the specification and claims should be interpreted as encompassing rather than being exclusive or exhaustive; that is, meaning "including but not limited to."

[0059] In the description of this invention, it should be understood that the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance. Furthermore, in the description of this invention, unless otherwise stated, "a plurality of" means two or more.

[0060] See Figures 1-3 This invention provides an underwater multi-target tracking method for locating and tracking underwater vehicles. The method includes the following steps:

[0061] S100. Establish a three-dimensional spatial coordinate system, obtain the sensor coordinates (x0, y0, z0), monitor the water volume V, the water surface coordinate Z2 on the Z-axis, the bottom coordinate Z3 on the Z-axis, and the clutter density λ.

[0062] S200. The sensor periodically acquires measurement information, including azimuth angle θ and elevation angle. The time delay τ between the sensor and the direct wave, the set of measurement information in the i-th frame is denoted as Z(i), the number of measurement information acquired by the sensor for a single target is L, i is any positive integer, and the j-th measurement information in the i-th frame is denoted as z. j (i), the total number of measurement information in the i-th frame is m. i ;

[0063] S300. Randomly generate K virtual targets c within the monitored water volume. K =[c1,c2,...,c K ], and according to the prior formula, represent the probability that a measurement information comes from a certain propagation path, the formula is expressed as:

[0064]

[0065]

[0066] in, For target c k The detection probability π generated by path l for measuring information. 00 To measure the probability that the information originates from clutter; π kl The measurement information for path l originates from target c. k The probability of;

[0067] S400. Based on the log-likelihood ratio, an underwater target model is established, expressed as follows:

[0068]

[0069] Where, p l (z j (i)|c k )=N[z j(i); h l (c k ,c s [i)] represents the target c k Measurement z is generated via path l. j The likelihood function of (i); c s The coordinates of the sensor are (x0, y0, z0), c k Represents the coordinates (x, y) of the kth virtual target k y k , z k ), hl(·) represents the measurement model corresponding to the l-th path;

[0070] S500: Based on the underwater target model, obtain the actual number of targets and the optimal coordinates of each actual target.

[0071] For step S100, the sensor is first set to a fixed position underwater, and a three-dimensional spatial coordinate system is established. The water body monitored by the sensor and the sensor itself are both located in the three-dimensional spatial coordinate system. That is, the position of the sensor can be represented by coordinates, and any point in the monitored water body can also be represented by coordinates.

[0072] "Monitored water body" refers to the effective range that the sensor can collect data from. The specific value of the monitored water body volume V depends on the location of the sensor in the water body, the depth of the water body, and the performance of the sensor itself.

[0073] In a three-dimensional coordinate system, the X-axis and Y-axis are perpendicular to each other and both are horizontal, while the Z-axis is vertical.

[0074] Assuming that the water surface and bottom are both planes and parallel to the horizontal plane, the height of the water surface and the height of the bottom can both be represented on the Z-axis. Therefore, the coordinates of the water surface on the Z-axis are Z2 and the coordinates of the bottom on the Z-axis are Z3. Z2 and Z3 are specific values ​​that can be obtained from monitoring the water body.

[0075] Clutter density λ can be obtained by pre-collecting data in the monitored water body, and it is a fixed value.

[0076] In step S200, the sensor periodically acquires measurement information. For example, the sensor acquires measurement information once per second. To distinguish it from the measurement information generated by the virtual target later, the measurement information acquired by the sensor is called the actual measurement information.

[0077] See Figure 1The sound waves emitted by the target (actual target) propagate through water (which can be seawater) to the sensor, where they are detected, thus obtaining the aforementioned measurement information. When the sound waves propagate in water, there are three modes: the first mode (corresponding to the direct wave path) is that the sound waves travel directly from the target to the sensor in an almost straight line; the second mode (corresponding to the water surface reflection wave path) is that the sound waves are reflected by the water surface (e.g., the sea surface) and then reach the sensor; and the third mode (corresponding to the underwater reflection wave path) is that the sound waves are reflected by the underwater surface (e.g., the seabed) and then reach the sensor.

[0078] Because the direct wave path is the shortest, the energy intensity detected by the sensor is the greatest, and the time point detected by the sensor is the earliest. The water surface reflected wave path and the water bottom reflected wave path are longer. Compared with the direct wave path, the time point detected by the sensor is later than the time point when the direct wave path is detected, and the energy intensity detected is lower than the energy intensity when the direct wave path is detected. Thus, based on the above characteristics, the actual measurement information of the direct wave path can be distinguished.

[0079] Regarding the actual measured information of the water surface reflected wave path and the actual measured information of the bottom reflected wave path, at the sensor's location, the water surface reflected wave path is tilted downwards, and the bottom reflected wave path is tilted upwards. Therefore, the pitch angle in the actual measured information can be used as a basis for further analysis. Distinguish between the actual measurement information of the water surface reflected wave path and the actual measurement information of the water bottom reflected wave path.

[0080] like Figure 2 As shown, when a target is present underwater, the sensor obtains a frame of actual measurement information containing three consecutive waves. The first wave has the earliest time point and the highest energy intensity; it represents the actual measurement information of the direct wave path. The subsequent two measurements represent the actual measurement information of the water surface reflected wave path and the water bottom reflected wave path, respectively. These two can be distinguished by the pitch angle in the actual measurement information. To achieve this.

[0081] Each actual measurement includes azimuth angle θ and elevation angle. And the time delay τ between the direct wave and the azimuth and elevation angles can be directly measured. As mentioned earlier, the actual measurement information of the direct wave path, the actual measurement information of the surface reflected wave path, and the actual measurement information of the bottom reflected wave path can be distinguished. Therefore, for the actual measurement information of the direct wave path, the time delay between it and the direct wave is 0. For the actual measurement of the surface reflected wave path and the actual measurement of the bottom reflected wave path, the time delays of these two are the time difference between the time point of the corresponding actual measurement information and the time point of the actual measurement information of the direct wave path.

[0082] Taking the acquisition of one frame of actual measurement information by the sensor as an example, in this frame, the total number of measurement information acquired is 100, and at this time m i Each actual measurement is assigned a specific number, such as the 1st actual measurement, the 2nd actual measurement, and so on up to the 100th actual measurement (the number corresponds to j). As mentioned earlier, in each frame, theoretically, the sensor can acquire three actual measurement messages emitted by each actual target (i.e., L=3). Some of these 100 actual measurement messages correspond to the actual target, while the rest are generated by clutter. However, current technology cannot effectively distinguish between these two, resulting in the inability to effectively locate and track the actual target (which may also be multiple).

[0083] To solve this problem, in step S300 of the present invention, K virtual targets are randomly generated within the monitored water body. Therefore, the coordinates of each virtual target in the three-dimensional spatial coordinate system are also clear. For each virtual target, the sensor can theoretically obtain three virtual measurement information (corresponding to three propagation paths) corresponding to the virtual target. The aforementioned actual measurement information is compared with the virtual measurement information to determine whether there are three actual measurement information that are close to the three virtual measurement information of a certain virtual target. If so, it means that there is a corresponding actual target for the three actual measurement information, and the actual target is located near the coordinates of the virtual target (specifically, the virtual target that is close to the three virtual measurement information).

[0084] The above process can be represented by the following formula:

[0085]

[0086]

[0087] in, For target c k The detection probability π generated by path l for measuring information. 00 To measure the probability that the information originates from clutter; π kl The measurement information for path l originates from target c. k The probability of.

[0088] In step S400, an underwater target model is established.

[0089]

[0090] Where, p l (z j (i)|c k )=N[z j (i); h l (ck ,c s [i)] represents the target c k Measurement z is generated via path l. j The likelihood function of (i); c s The coordinates of the sensor are (x0, y0, z0), c k Represents the coordinates (x, y) of the kth virtual target k y k , z k ), h l (·) represents the model of the measurement information corresponding to the l-th path (that is, the virtual measurement information corresponding to each virtual target).

[0091] In this way, an underwater target model corresponding to each frame can be established based on the actual measurement information obtained by the sensor in each frame.

[0092] In step S500, the number of actual targets and the optimal coordinates of each actual target are obtained based on the underwater target model.

[0093] like Figure 3 As shown in the diagram, there are three obvious protrusions, which indicates that there are three actual targets. The coordinates corresponding to the positions of these three protrusions are the preferred coordinates of these three actual targets, thus realizing the positioning and tracking of underwater targets.

[0094] Of course, the above positioning only describes the measurement information of a single frame acquired by the sensor. In other words, the positioning of the target is only the preferred coordinates of the actual target at that frame moment. Therefore, by processing the measurement information of each frame acquired by the sensor according to steps S100-S500, the continuous preferred coordinates of the actual target can be obtained, thereby enabling real-time tracking of the actual target and drawing the actual target's movement trajectory, thus predicting the target's direction of travel and position.

[0095] The measurement information for a single target is L=3, which are the direct wave path, the water surface reflected wave path, and the bottom reflected wave path, respectively.

[0096] For each virtual target,

[0097] The measurement information corresponding to the direct wave path is expressed as follows:

[0098] The measurement information corresponding to the water surface reflected wave path is expressed as follows:

[0099] The measurement information corresponding to the underwater reflected wave path is expressed as follows:

[0100] In this way, corresponding virtual measurement information can be generated for each virtual target.

[0101] Specifically,

[0102]

[0103]

[0104] τ1(c k ) = 0

[0105]

[0106]

[0107]

[0108]

[0109]

[0110]

[0111] Where c is the speed of sound underwater, for example, in seawater at 25℃, c = 1531 m / s.

[0112] Because the sensor's coordinates c s (x0, y0, z0), the coordinates c of the virtual target k (x k y k , z k All of these are known, so we can obtain the virtual measurement information generated by each virtual target.

[0113] In step S100, the azimuth variance μ, elevation variance ω, and time delay variance υ are also obtained.

[0114] The sensor is a vector hydrophone.

[0115] The azimuth variance and the time delay variance are obtained directly from the sensor;

[0116] The pitch angle variance follows a Gaussian mixture distribution, therefore:

[0117] The pitch angle variance ω of the direct wave path 直达 satisfy

[0118] ω 直达 ~GMM=0.6*N(0,1^2)+0.4*N(0,4^2),

[0119] The pitch angle variance ω of the water surface reflected wave path and the water bottom reflected wave path多径 satisfy

[0120] ω 多径 ~GMM=0.4*N(0,1^2)+0.6*N(0,4^2);

[0121] The measurement information corresponding to the direct wave path is expressed as follows:

[0122]

[0123] The measurement information corresponding to the water surface reflected wave path is expressed as follows:

[0124]

[0125] The measurement information corresponding to the underwater reflected wave path is expressed as follows:

[0126]

[0127] Taking into account the azimuth variance μ, elevation variance ω, and time delay variance υ can enhance the accuracy of actual target positioning.

[0128] In step S200, the sensor continuously acquires Nw frames of data;

[0129] In step S400, an underwater target model is established based on the log-likelihood ratio, expressed as:

[0130]

[0131] Nw is preferably 3-5. In practical applications, the period for the sensor to acquire actual measurement information is controlled within 2 seconds, for example, once per second. By constructing the target model with continuous Nw frame data, the model becomes more accurate, and the obtained preferred coordinates are closer to the actual coordinates.

[0132] It should be understood that the above embodiments are merely exemplary and not restrictive. Various obvious or equivalent modifications or substitutions that can be made by those skilled in the art regarding the above details without departing from the basic principles of the present invention will be included within the scope of the claims of the present invention.

Claims

1. A method of underwater multi-target tracking, characterized in that, The method comprises the steps of: S100, establishing a three-dimensional space coordinate system, obtaining coordinates (x0, y0, z0) of the sensor, monitoring a water volume V, a coordinate Z2 of a water surface on a Z axis, a coordinate Z3 of a water bottom on the Z axis, and a clutter density λ; S200, the sensor periodically acquires measurement information, the measurement information including an azimuth angle θ, a pitch angle and a time delay τ between a direct wave, a set of measurement information in an i-th frame is denoted as Z(i), a number of measurement information of a single target acquired by the sensor is L, i is an arbitrary positive integer, a j-th measurement information in the i-th frame is denoted as z j (i), a total number of measurement information in the i-th frame is m i ; S300、In the monitoring water volume, K virtual targets c are randomly generated K = [c1, c2,..., c K ], and the likelihood that a measurement information comes from a certain propagation path is expressed according to a prior formula, which is expressed as: where, is the target c k is the detection probability of the measurement information generated by path l, π 00 is the probability that the measurement information originates from clutter; π kl is the probability that the measurement information generated by path l originates from target c k ; S400, establishing an underwater target model according to a log-likelihood ratio, and the underwater target model is expressed as: Where, p l (z j (i)|c k )=N[z j (i); h l (c k ,c s [i)] represents the target c k Measurement z is generated via path l. j The likelihood function of (i); c s The coordinates of the sensor are (x0, y0, z0), c k Represents the coordinates (x, y) of the kth virtual target k y k , z k ), h l (·) represents the model of the measurement information corresponding to the l-th path; S500, obtaining an actual target quantity and optimal coordinates of each actual target according to the underwater target model.

2. The underwater multi-target tracking method according to claim 1, characterized in that, The measurement information L of a single target is 3, which is a direct wave path, a water surface reflection wave path and a water bottom reflection wave path respectively; Wherein, The measurement information corresponding to the direct wave path is expressed as The measurement information corresponding to the water surface reflection wave path is expressed as The measured information corresponding to the water bottom reflected wave path is expressed as 3. The underwater multi-target tracking method according to claim 2, characterized in that, τ1(c k ) = 0 Wherein, c is an underwater sound speed.

4. The underwater multi-target tracking method according to claim 3, characterized in that, In step S100, an azimuth angle variance μ, a pitch angle variance ω and a time delay variance υ are also obtained, The sensor is a vector hydrophone, The azimuth angle variance and the time delay variance are directly obtained from the sensor; the elevation angle variance ω of the direct path 直达 satisfies ω 直达 ~GMM = 0.6 * N(0, 1^2) + 0.4 * N(0, 4^2), The elevation angle variance ω of the water surface reflected wave path and the water bottom reflected wave path 多径 satisfies ω 多径 ~GMM = 0.4 * N(0, 1^2) + 0.6 * N(0, 4^2); The measurement information corresponding to the direct wave path is expressed as The measurement information corresponding to the water surface reflection wave path is expressed as The measurement information corresponding to the water bottom reflection wave path is expressed as 5. The underwater multi-target tracking method according to any one of claims 1 to 4, characterized in that, In step S200, the sensor continuously obtains Nw frames of data; In step S400, an underwater target model is established according to a log-likelihood ratio, and the underwater target model is expressed as:

Images