CN114355409A - Water surface target motion estimation method - Google Patents

Abstract

The invention relates to a method for estimating the motion of a water surface target, which comprises the following steps: a. preprocessing the motion trace data of the water surface target with low measurement precision; b. the speed and direction of motion of the surface target are estimated. The invention can accurately estimate the speed and the course of the water surface target.

Description

Water surface target motion estimation method

Technical Field

The invention relates to a method for estimating the motion of a water surface target.

Background

Mastering wide-area situation, maintaining benefits and safety at sea is a cornerstone of ocean strategic development. With the improvement of the space-based equipment capability in China, the problems of far seeing and visibility are basically solved, but the differences exist in the aspects of clear seeing, keeping up with the seeing, systematic application and the like. The invention takes the space-based active/passive radar to find unknown ships and auxiliary marine police driving as the application background, and grasps the target intention by estimating the target navigational speed and course.

For high-score four-satellite data, a method for forming a target track by using multi-hypothesis tracking [2] is proposed in a document [1], and then the speed and the course of the target are estimated. The method mainly comprises the following steps: carrying out background suppression on the ocean background remote sensing image by utilizing multi-structure multi-scale morphological filtering; obtaining candidate targets by adopting self-adaptive threshold segmentation and self-organizing clustering; according to the target motion characteristics, multi-target mobile neighborhood judgment is carried out on the candidate targets by using the staring sequence image of the geostationary orbit satellite, and false targets are removed; the ship target is associated and the satellite platform data is fused, so that the information such as the position, the navigational speed, the course, the motion track and the like of each ship can be accurately acquired. The method has the advantages of simple algorithm, high target detection rate, low false alarm rate, good stability and the like. Because the high-resolution four-number satellite adopts an optical imaging system, the motion track of the ship is continuous and smooth on an image plane. In the process of projecting the image plane to the geographic coordinates, although a certain degree of nonlinear system error exists, the spatial position relation of adjacent measuring point tracks is close to the actual situation, and the influence degree on the navigation speed and the heading estimation is small.

The space-based active/passive radar is gradually widely applied due to the long detection distance and the wide coverage range. For space-based passive radar, the incoming wave direction can be further determined by measuring the characteristic parameters of the radiation source signal, and the radiation source positioning and identification are completed, and currently, a single-satellite reconnaissance positioning method and a multi-satellite time difference reconnaissance positioning method [1] are commonly used. The positioning accuracy of the space-based active/passive radar can reach tens of kilometers at most, and the target measurement point trace generates large fluctuation. Due to the disorganized spatial relative positions of adjacent measurement point tracks, the problems that the estimated speed far exceeds the real capability of a ship and the course estimation is reverse frequently occur. The method in the document [3] focuses on links such as background suppression, weak target detection under a complex background, conversion error from image plane coordinates to geographic coordinates and the like, and is not suitable for estimating the speed and the course under low measurement accuracy.

The invention provides a ship target speed and course estimation method under low measurement precision aiming at space-based active/passive radar measurement and aiming at obtaining smooth flight path and more accurate course estimation on the basis of generating a target point path through ocean situation and space-time constraint.

[1] Yaoqiang and the like, ship target tracking based on a high-resolution four-number satellite, and the fourth high-resolution earth observation academic annual meeting in 2017.

[2] Blackman S.multiple perspective tracking for multiple target tracking [ J ]. IEEE Aerospace and Electronic Systems Magazine,2004,19(1): 5-18. Multi-hypothesis tracking algorithm for Multi-target tracking, journal of Aerospace and Electronic Systems of the institute of Electrical and electronics Engineers.

[3] Guo Fucheng et al, the principle of spatial electronic reconnaissance and location.

[4] He friend, et al, radar data processing and applications, electronics industry Press.

Generally speaking, the higher the measurement accuracy and data rate, the more beneficial the accuracy of target tracking and the more accurate the estimation of speed and heading. As can be seen from the high-precision measurement diagram at the adjacent time shown in FIG. 1, the target is at t_k-1The true and measured values of the position at the time are x_k-1And

(solid and triangular, respectively) with a target at t_kThe true value and the measured value of the time are x respectively_kAnd

the larger the measurement error, the larger the range of the measurement distribution shown by the elliptical area, the more the measured values deviate from the true value. However, when the measurement error is larger than the two adjacent pointsHalf of the inter-distance, the spatial relationship of the measured positions no longer matches the actual situation because the measured values deviate far from the true values, resulting in the northeastern speed being estimated as the southwest direction, as shown in fig. 2. Therefore, a series of processes are required to be performed on the trace data in advance to recover the relative position relationship of the trace, so as to avoid the deviation between the estimated course value and the true value exceeding 90 degrees and the speed estimation generating large fluctuation.

For the estimation of the speed and the heading, firstly, stable tracking of a target is established to obtain a smooth track. Generally, a rectangular coordinate system is used to describe the target state equation. Space-based measurements usually describe the position of ship targets in terms of latitude and longitude, so the geographical coordinate system is the most straightforward choice. However, state transition models in geographic coordinates are not sufficient to accurately depict turning or stationary movements of objects.

The possible motion forms of the ship target include static motion, linear motion, reciprocating motion, circular motion and the like, so a filtering method is mostly adopted. Generally, the state transition model adopts a uniform velocity model, a uniform acceleration model or a cooperative turning model [4], and the like. However, in the case of low measurement accuracy, even if the target actually moves at a constant speed, it is easy to be biased to an accelerated motion by the measurement during the filtering process. For this reason, on the one hand, it is necessary to add a state transition model that is closer to the target motion, and on the other hand, it is necessary to reduce process noise so that the set state transition model is more confident in updating the filter gain than the measured value. However, this results in a lag in the response of the filter when the motion of the target deviates from the set state transition model. Therefore, a proper state transition model needs to be selected, so that the motion rule of the ship target is ensured to be met, and the filter is prevented from being biased by low-precision measurement.

Disclosure of Invention

The invention aims to provide a method for estimating the motion of a water surface target.

In order to achieve the above object, the present invention provides a method for estimating the motion of a water surface target, comprising the following steps:

a. preprocessing the motion trace data of the water surface target;

b. fusing weight setting strategies;

c. the speed and direction of motion of the surface target are estimated.

According to an aspect of the present invention, in the step (a), when the distance between two adjacent traces is within a fusion threshold, performing fusion, where the fusion point is a mean value of the connected traces within the fusion threshold;

if the continuous traces are all within the fusion threshold, performing iterative fusion;

and using the distance of the subsatellite point or the measurement error as the basis for calculating the fusion weight.

According to an aspect of the present invention, the step (a) includes the steps of:

a1, sequencing input trace point data according to the measurement time and determining a track starting point;

a2, assuming that the measurement precision of the current sensor is r, setting the fusion threshold as 2r + V_maxΔt；

a3 finding the current time t_k-1Corresponding trace of points x_k-1And the next time t_kCorresponding trace of points x_kCalculating the distance delta d | | | x between two adjacent traces_k-1-x_kL; if no measurement for the next time instant can be found, step a8 is executed;

a4, if the distance delta d between two adjacent points exceeds the fusion threshold 2r + V_maxΔ t, the two traces are not needed to be processed, and the step is skipped to step a 3; otherwise, step a5 is executed;

a5, calculating the position weight of adjacent time, which is normalized w_k-1And w_k. Because the different measurement error generation reasons of different sensors are different, a corresponding fusion weight strategy is used in a matched manner, which is detailed in the following section;

a6, carrying out weighted fusion on two adjacent point positions to obtain a new point position

The corresponding measurement time takes the value as

The measurement time can also be valued as required;

a7, repeating the steps (a2) to (a6) until there is no measuring point at the next moment;

a8, finding no measurement at the next moment, and finishing the preprocessing of the trace point data.

Wherein V is the maximum navigational speed of the ship, and is generally 55.56 kilometers per hour, namely 30 knots; and delta t is the time difference between adjacent point traces in the time sequence.

According to one aspect of the present invention, the position weight is obtained by using the distance between the two points under the satellite in such a manner that the combined weight (w) corresponding to the two points is obtained_k-1And w_k) Is inversely proportional to the distance between the points under the satellite, and the combined weighted value is normalized to obtain the position weights of two point traces

And

the time corresponding to the new fusion trace obtained by weighted re-fusion is an intermediate value of two moments;

the fused trace will replace (cover) the original trace (the trace participating in the fusion);

if the continuous iterative fusion is carried out, the fusion strategy is repeated until the distance between two adjacent points exceeds the fusion threshold;

wherein d is the distance from the current moment measuring point to the subsatellite point.

According to one aspect of the invention, in the step (b), under an interactive multi-model architecture, a state variable and a state transition model set are selected, and the navigation speed and the heading of the water surface target are estimated by using a filtering and smoothing method.

For sensors that cannot compute fusion weights or have no historical statistics, the fused trace points may be taken as the mean of the connected trace points within the fusion distance threshold.

In order to obtain a better point-trace fusion result, different fusion weight setting strategies can be used according to the positioning mechanism, the measurement precision and the like of the sensor, so that the relative position relation between adjacent point traces can be accurately restored, and for example, measurement errors, measurement covariance matrixes, measurement scores, the distance between points under the satellite and the like are used as the basis for calculating the fusion weight. The larger the measurement error and the measurement covariance matrix are, the smaller the measurement score is, and the smaller the weight is in the fusion process; conversely, the larger. The following description focuses on the strategy of setting the fusion weight by using the distance between the substellar points, mainly including the direct method, the theoretical method and the statistical method, as shown in fig. 5.

1) Direct process

The method comprises the following steps: calculating the distance between the subsatellite points of two adjacent point traces;

step two: calculating the current positioning precision estimated value according to the section to which the target distance from the sub-satellite point belongs, and taking the reciprocal of the current positioning precision estimated value as the weight of the current trace point and the weight of the next trace point respectively and recording the reciprocal as w'_k-1And w'_k；

Step three: to w'_k-1And w'_kNormalization processing is carried out to obtain the weight w finally used for fusion_k-1And w_k。

2) Theoretical method

The method comprises the following steps: calculating a positioning error theoretical curve from the zero position of the subsatellite point to the maximum boundary of the precision-preserving area according to a specific sensor;

step two: calculating the distance between the subsatellite points of two adjacent point traces;

step three: calculating the positioning errors of two adjacent point traces based on a positioning error theoretical curve, and respectively taking the reciprocal of the positioning errors as the weight of the current point trace and the weight of the next point trace, and recording as w'_k-1And w'_k；

Step four: to w'_k-1And w'_kNormalization processing is carried out to obtain the weight w finally used for fusion_k-1And w_k。

3) Statistical method

The method comprises the following steps: and carrying out statistical processing on the historical positioning precision condition of the sensor to obtain a positioning error statistical curve related to the distance of the subsatellite point. On the basis, positioning error statistical curves under different target characteristics can be obtained according to the electromagnetic characteristics or the optical characteristics of the target;

step two: calculating the distance between the subsatellite points of two adjacent point traces;

step three: calculating the positioning errors of two adjacent point traces based on a positioning error theoretical curve, and respectively taking the reciprocal of the positioning errors as the weight of the current point trace and the weight of the next point trace, and recording as w'_k-1And w'_k；

Step four: to w'_k-1And w'_kNormalization processing is carried out to obtain the weight w finally used for fusion_k-1And w_k。

According to one aspect of the invention, in the step (b), under an interactive multi-model architecture, a state variable and a state transition model set are selected, and the navigation speed and the heading of the water surface target are estimated by using a filtering and smoothing method.

According to an aspect of the present invention, the step (b) comprises the steps of:

b1, establishing a local northeast coordinate system by taking the track starting point as the origin, and taking the coordinate system as a tracking coordinate system. Then, all the point tracks on the flight path are converted into the local coordinate system from the geographic coordinate system; (ii) a

b2, setting the conversion probability: a probability of a stationary model and a probability of a Cooperative Turning (CT) model;

b3, establishing a three-axis state equation of the static model and the CT model, wherein the CT equation introduces a turning angular velocity state variable, and establishes an observation equation containing three-axis position measurement;

the method comprises the following steps that a state equation of a cooperative turning model introduces a turning angular velocity state variable to establish an observation equation containing three-axis position measurement;

b4, guessing the measurement estimation of the current moment according to the system state estimation and covariance estimation of the previous moment, and then carrying out reinitialization calculation on the model, wherein a new initial value is obtained through a Markov operation matrix among different models; obtaining a new initial value through a Markov operation matrix among different models;

b5, deriving nonlinear state equations and Jacobian matrices, using EKF (Extended Kalman Filter);

b6, giving the weight ratio of the current most suitable tracking model by calculating the similarity of the current model and the current moving object state;

b7, obtaining an estimation result according to the individual calculation result and the model matching weight of each model, and finishing filtering processing;

b8, smoothing all the traces based on the filtering result, wherein the smoothing value is restricted by the measurement covariance matrix of the final state and the process noise covariance matrix; finishing the smoothing treatment;

b9, calculating and outputting the heading based on the speed at each moment; if imaging information close to time is input, the heading is assisted and judged by methods such as identifying the head direction of a target ship and the like, and the navigation speed and the heading are calculated.

b10, finishing the processing of all tracks, and finishing the estimation of the speed and the heading.

According to one aspect of the present invention, in the step (b), the navigational speed v (k) at the k-th time is calculated as:

wherein v is_e(k)、v_n(k) And v_u(k) Respectively are three-axis velocity components in a northeast coordinate system;

defining the heading angle as north-east-off clockwise angle in the range of [0,2 pi), the heading phi (k) at the kth moment is:

according to an aspect of the present invention, the auxiliary heading determining in step (b9) is performed by:

if only one point of the flight path has corresponding imaging information input, identifying the head or tail of the target ship under the condition that the resolution ratio allows, thereby giving the target course;

and if the plurality of tracks in the flight path have corresponding imaging information input, obtaining a course estimation value according to the heading direction and the trail after the static frame image recognition. Generating a track by a multi-frame imaging information positioning result, and outputting a speed and a course estimation through filtering smoothing;

and comprehensively considering the estimated values of the navigational speed and the course, completing multi-party information, and fusing to generate the final estimated values of the navigational speed and the course.

According to one aspect of the invention, the fused estimation of the navigational speed and the heading comprises the following steps:

s1, generating a track according to the imaging information positioning result, and outputting the estimation of the initial speed and the initial course in a filtering and smoothing mode;

s2, identifying the bow of the water surface target in each frame to obtain the course of the water surface target in the static frame;

s3, carrying out weighted fusion on the headings obtained in the step (S1) and the step (S2), wherein the weighted value is an empirical value or is obtained by Kalman filtering/confidence coefficient evaluation fusion;

if the difference between the trace point detection time of the imaging information and the trace point detection time of the sensor is larger (5 minutes), carrying out the subsequent steps, otherwise, directly outputting the estimated values of the course and the navigational speed;

s4, carrying out fusion processing on the flight path generated by low measurement precision and the flight path generated by imaging information to complete the two flight path information;

the fusion processing mode is covariance weighted track fusion or adaptive track fusion;

and S5, filtering and smoothing the merged track to obtain and output estimated values of the course and the speed.

According to the concept of the invention, the invention provides a point trace data iteration fusion preprocessing method, which fuses a weight setting strategy and a ship target navigational speed and course estimation process under low-precision measurement.

According to one scheme of the invention, a proper filter architecture and a proper state model transfer set are selected, so that the requirement on accurate depiction of the movement of the ship target can be met, the problem of deviation caused by using an acceleration model under low-precision measurement is avoided, and the bearing range of large calculation amount caused by the number of models in engineering application can be enlarged.

According to one scheme of the invention, the relative position relation of the track points can be effectively recovered by a point track data fusion preprocessing mode, so that the problem that the deviation between the slow speed or static target course speed estimation and an actual value is large when the positioning accuracy of the sensor is low is solved or weakened, and a foundation is laid for developing stable tracking and obtaining a smooth track.

According to one scheme of the invention, the point trace data fusion preprocessing is to restore the spatial relative position relation of adjacent point traces by setting a fusion threshold and adopting a recursive fusion strategy, so that the deviation between the estimated value and the true value of the course caused by low measurement precision is greatly reduced, and the problems of the deviation exceeding 90 degrees, large fluctuation of speed estimation and the like are solved.

The invention is suitable for the condition that other sensors with low measurement precision estimate the dynamic information (including position, navigational speed, course, motion track, etc.) of the deep target, and can obtain course estimation with higher precision under the condition that other imaging intelligence information confirms.

The practical process of ship target speed and course estimation under low measurement accuracy is provided, and by selecting a proper filter architecture and a proper state model transfer set, the requirement for accurate description of ship target motion can be met, the problem of deviation caused by using an acceleration model under low-accuracy measurement is avoided, and the calculated amount caused by increasing the number of models is increased to increase the bearing range of engineering application.

According to one scheme of the invention, aiming at the data preprocessing with low measurement precision, the navigation speed and the navigation course are estimated by adopting Kalman filtering combined with an interactive multi-model architecture (IMM), and the navigation speed and the navigation course estimation of the water upper target with low measurement precision is optimized.

Drawings

FIG. 1 is a schematic diagram of the relative positions of phases during high-precision measurement;

FIG. 2 is a schematic diagram of the relative positions of phases during low precision measurements;

FIG. 3 is a graph schematically illustrating the result of weighted fusion of neighboring measurement points according to an embodiment of the present invention;

FIG. 4 is a flow diagram that schematically illustrates iterative fusion preprocessing of trace point data, in accordance with an embodiment of the present invention;

FIG. 5 is a flow diagram that schematically illustrates an iterative fusion strategy for trace point data, in accordance with an embodiment of the present invention;

FIG. 6 is a flow chart that schematically illustrates a method of water surface target motion estimation in accordance with an embodiment of the present invention;

FIG. 7 is a schematic representation of a marine low-speed moving object track map before point track fusion in accordance with an embodiment of the present invention;

FIG. 8 is a schematic representation of a point-track fused track map in accordance with an embodiment of the present invention;

FIG. 9 schematically shows a track graph of filtering and smoothing according to an embodiment of the invention.

Detailed Description

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the embodiments will be briefly described below. It is obvious that the drawings in the following description are only some embodiments of the invention, and that for a person skilled in the art, other drawings can be derived from them without inventive effort.

In describing embodiments of the present invention, the terms "longitudinal," "lateral," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," and the like are used in an orientation or positional relationship that is based on the orientation or positional relationship shown in the associated drawings, which is for convenience and simplicity of description only, and does not indicate or imply that the referenced device or element must have a particular orientation, be constructed and operated in a particular orientation, and thus, the above-described terms should not be construed as limiting the present invention.

The present invention is described in detail below with reference to the drawings and the specific embodiments, which are not repeated herein, but the embodiments of the present invention are not limited to the following embodiments.

The method for estimating the motion of the water surface target is suitable for estimating the speed and the course of the water surface target such as a ship and the like under low measurement precision, and the situation that other sensors with low measurement precision estimate the dynamic information (including the position, the speed, the course, the motion track and the like) of the deep-level target. The method comprises the steps of preprocessing the motion trace point data of the water surface target, and estimating the motion speed and direction of the water surface target so as to solve the problems of disordered spatial relative position relation caused by low measurement precision and ship target tracking under low measurement precision.

When the trace point data is preprocessed, an iterative fusion threshold is set by combining the specific conditions of the sensor. When the distance between two adjacent traces is within a fusion threshold (namely, a distance threshold), performing fusion; and if the continuous traces (namely the phenomenon that the continuous traces need to be fused) are all within the fusion threshold, performing iterative fusion. Of course, without a better fusion strategy, the fusion point location may be the mean of the connected point traces within the fusion threshold. In addition, in order to obtain a better fusion result, a fusion strategy can be selected according to a positioning mechanism of the sensor, namely, the distance between the points under the satellite or the measurement error and the like are used as the basis for fusion weight calculation, so that the relative position relation between the point traces is restored. When the speed and the course of the water surface target are estimated, on the basis of point trace data iterative fusion preprocessing, a proper state variable and a state transition model set are selected under an interactive multi-model architecture (IMM) according to the motion characteristics of the ship target, and then the speed and the course of the water surface target are estimated by using a filtering and smoothing method.

Referring to fig. 3 and 4, the trace point data fusion method of the present invention is to sort the input trace point data according to the measurement time and determine the track starting point, and if the measurement accuracy of the current sensor is r, set the fusion threshold to be 2r + V_maxAt. Then, the current time t is determined_k-1Corresponding trace of points x_k-1And the next time t_kCorresponding trace of points x_kCalculating the distance delta d | | | x between two adjacent traces_k-1-x_kL. If the measuring point at the next moment can not be found, the trace point is directly finishedAnd (5) performing a data fusion step. If the distance delta d between two adjacent points exceeds the fusion threshold 2r + V_maxAnd delta t, not processing the two traces, and continuously searching the corresponding traces, otherwise, executing the subsequent steps. Subsequently, the position weights of the adjacent time instants are calculated to be w which is normalized respectively_k-1And w_kAnd a corresponding fusion weight strategy is cooperatively used. And because the different sensor measurement errors are different, the position weight value can be obtained by calculating and normalizing according to the distance between the sub-satellite points, the measurement covariance matrix and the scoring strategy. Then, carrying out weighted fusion on two adjacent point locations to obtain a new point location

The corresponding measurement time takes the value as

And the value of the measurement moment can be taken according to the requirement. Wherein Vmax is the maximum navigational speed of the ship, and is generally 55.56 kilometers per hour, namely 30 knots; and delta t is the time difference between adjacent point traces in the time sequence.

In the invention, when the speed and the course of the water surface target are estimated under low-precision measurement, a local northeast coordinate system is established by taking a track starting point as an origin, and the local northeast coordinate system is taken as a tracking coordinate system. Then, all the point tracks on the flight path are converted from the geographic coordinate system to the local coordinate system. Setting the conversion probability: a probability of a stationary model and a probability of a Cooperative Turning (CT) model; and establishing a three-axis state equation of the static model and the CT model, wherein the CT equation introduces a turning angular velocity state variable, and establishes an observation equation containing three-axis position measurement. Wherein, the state equation (CT equation) of the cooperative turning model introduces the turning angular velocity state variable to establish an observation equation containing three-axis position measurement.

And estimating the current moment measurement estimation according to the previous moment system state estimation and covariance estimation, then performing reinitialization calculation on the model, and obtaining a new initial value through a Markov operation matrix among different models. Deriving a nonlinear equation of state and a Jacobian matrix using EKF; and giving the weight ratio of the current most suitable tracking model by calculating the similarity of the current model and the current moving object state. And obtaining a final estimation result according to the independent calculation result of each model and the model matching weight, and finishing filtering processing. And then smoothing all the traces based on the filtering processing result, wherein the smoothing value is constrained by a measurement covariance matrix of a final state and a process noise covariance matrix, thereby finishing the smoothing processing. Calculating and outputting a course based on the speed at each moment; if imaging information close to time is input, the heading is assisted and judged by methods such as identifying the head direction of a target ship and the like, and the navigation speed and the heading are calculated. And finishing the processing of all tracks, and finishing the estimation of the speed and the heading.

Finally, as shown in fig. 5, after the processing of all the tracks is completed, the estimation of the speed and the heading can be finished.

When the navigation speed and the course are calculated, if no other information is confirmed, the formula for calculating the navigation speed v (k) at the k-th moment is as follows:

wherein v is_e(k)、v_n(k) And v_u(k) Respectively are triaxial velocity components in an ENU coordinate system;

defining the heading angle as north-east-off clockwise angle in the range of [0,2 pi), the heading phi (k) at the kth moment is:

wherein x is_E(k) And x_N(k) A target east and north velocity component, respectively.

The auxiliary course judging mode under the condition that other imaging information is confirmed is that if only a certain point of the whole course has corresponding imaging information input, the head or tail of the target ship is identified under the condition that the resolution ratio allows, so that the target course is given; if multiple points in the whole course track have corresponding imaging information input, the course estimation value can be obtained according to the heading direction and the trail after the static frame image recognition. Generating a track by a multi-frame imaging information positioning result, and outputting a speed and a course estimation through filtering smoothing; and comprehensively considering the estimated values of the navigational speed and the course, completing multi-party information, and fusing to generate the final estimated values of the navigational speed and the course. And (4) estimating the speed and the course by using multi-means fusion. Specifically, a course estimation value is obtained according to the heading direction and the trail after the static frame image recognition. And generating a track by a multi-frame imaging information positioning result, and outputting the speed and the course estimation through filtering and smoothing processing.

The weighted value is an empirical value or obtained by fusion of methods such as Kalman filtering/confidence estimation and the like. If the difference between the trace point detection time of the imaging information and the trace point detection time of the sensor with low measurement precision is larger, the subsequent steps are carried out, otherwise, the course speed estimation value is directly output. And then, carrying out fusion processing on the flight path generated by the low measurement precision and the flight path generated by the imaging information so as to complete the information of the two parties. The fusion processing mode can be covariance weighted track fusion or adaptive track fusion, etc. And finally, filtering and smoothing the fused track according to the mode to obtain and output a final course speed estimation value.

The process according to the invention is described in detail below in a specific embodiment:

in the embodiment, real data of one ship target for about 1 hour is used as input data, the average speed of the input data is 1.3 knots, and the positioning accuracy of the measured data is 5 kilometers, so that a typical ship target low-speed motion or static state scene can be constructed.

Referring to fig. 6 and 7, the real track and the measured data are located in the coordinate system of the northeast of the origin of the track. It can be seen that, because the precision of the measured data is low, the measured flight path has obvious deviation from the real flight path, which is not beneficial to the estimation of the course angle. And the heading angle deviation corresponding to the 4 th, 6 th and 7 th measured values exceeds 90 degrees, as shown in table 1:

TABLE 1 course cruise difference between real point and measured point

In the embodiment, longitude and latitude high coordinates and a timestamp of the target are used as input, and the ENU coordinates are used for describing the track, the speed and the course of the water surface target and are used as output. The setting of the fusion threshold is influenced by the measurement precision r of the sensor, the measurement time difference delta t of two adjacent points and the average speed V of the target in the moment, and the expression is 2r + V delta t. Then, the distance delta d | | | x between two adjacent traces is calculated_k-1-x_kIf delta d is more than 2r + V delta t, the two traces are not processed; otherwise, two points in adjacent time are merged into one point according to the combined weighted value.

The present embodiment adopts the distance between the points under the star as a method for setting the weight value. Specifically, the combining weight value is inversely proportional to the distance between the subsatellite points, and the combining weight value w corresponding to the two points_k-1And w_kPerforming normalization processing to obtain the position weight of

And

and obtaining a time corresponding to the new trace point as an intermediate value of two moments through weighting and fusion, wherein the fused trace point can replace the original trace point. If the continuous iterative fusion is carried out, the fusion strategy is repeated until the distance between two adjacent points exceeds the fusion threshold; wherein d is the distance from the target measurement point to the subsatellite point. As shown in table 2, the course angle and the speed estimation are significantly improved by using the above-mentioned fusion method of the point trace data, and the angle deviation of more than 90 degrees is eliminated:

TABLE 2 course speed difference between real point and pre-processed trace point

For sensors that cannot compute fusion weights or have no historical statistics, the fused trace points may be taken as the mean of the connected trace points within the fusion distance threshold.

In order to obtain a better point-trace fusion result, different fusion weight setting strategies can be used according to the positioning mechanism, the measurement precision and the like of the sensor, so that the relative position relation between adjacent point traces can be accurately restored, and for example, measurement errors, measurement covariance matrixes, measurement scores, the distance between points under the satellite and the like are used as the basis for calculating the fusion weight. The larger the measurement error and the measurement covariance matrix are, the smaller the measurement score is, and the smaller the weight is in the fusion process; conversely, the larger. The main strategies for setting the distance between the substellar points as the fusion weight include a direct method, a theoretical method, and a statistical method, and the present embodiment uses the direct method, as shown in fig. 5, and includes:

the method comprises the following steps: calculating the distance between the subsatellite points of two adjacent point traces;

step two: calculating the current positioning precision estimated value according to the section to which the target distance from the sub-satellite point belongs, and taking the reciprocal of the current positioning precision estimated value as the weight of the current trace point and the weight of the next trace point respectively and recording the reciprocal as w'_k-1And w'_k；

Step three: to w'_k-1And w'_kNormalization processing is carried out to obtain the weight w finally used for fusion_k-1And w_k。

Referring to fig. 8 and 9, due to the existence of the measurement outlier, there is still a certain deviation between the current track and the current heading angle and the true value, so the track after the track pre-processing is subsequently filtered and smoothed, as shown in table 3:

TABLE 3 course airspeed difference between the true point and the smoothed filter

It can be seen that the smoothed flight path is already greatly close to the real flight path, but compared with the pre-processing result of the point path, the errors of the course speed estimated values of the last points are slightly increased, which is caused by the fact that the positioning errors of the last points are larger and no subsequent point correction is smooth. In addition, the whole course navigational speed deviation value after filtering smoothing is more stable.

The above description is only one embodiment of the present invention, and is not intended to limit the present invention, and it is apparent to those skilled in the art that various modifications and variations can be made in the present invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (9)

1. A method for estimating the motion of a water surface target comprises the following steps:

a. preprocessing the motion trace data of the water surface target;

b. the speed and direction of motion of the surface target are estimated.

2. The method according to claim 1, wherein in the step (a), when the distance between two adjacent traces is within a fusion threshold, the fusion is performed, and the fusion point is an average value of the connected point traces within the fusion threshold;

if the continuous traces are all within the fusion threshold, performing iterative fusion;

and using the distance of the subsatellite point or the measurement error as the basis for calculating the fusion weight.

3. The method of claim 2, wherein step (a) comprises the steps of:

a1, sequencing input trace point data according to the measurement time and determining a track starting point;

a2, if the measurement precision of the current sensor is r, setting the fusion threshold as 2r + V_maxΔt；

a3 finding the current time t_k-1Corresponding trace of points x_k-1And the next time t_kCorresponding trace of points x_kCalculating the distance delta d | | | x between two adjacent traces_k-1-x_kL; if the next measurement cannot be foundExecuting the step a 8;

a4, if the distance delta d between two adjacent points exceeds the fusion threshold 2r + V_maxΔ t, the two traces are not needed to be processed, and the step is skipped to step a 3; otherwise, step a5 is executed;

a5, calculating the position weight of adjacent time, which is normalized w_k-1And w_kAnd a corresponding fusion weight strategy is used in a matching way;

a6, carrying out weighted fusion on two adjacent point positions to obtain a new point position

The corresponding measurement time takes the value as

a7, repeating the steps (a2) to (a6) until there is no measuring point at the next moment;

a8, finding out the measurement at the next moment, and finishing the trace point data preprocessing;

wherein Vmax is the maximum ship speed of the ship; and delta t is the time difference between adjacent point traces in the time sequence.

4. The method of claim 3, wherein the position weight is obtained by using the distance between the two points under the satellite, and the combining weight w corresponding to the two points is obtained by_k-1And w_kIs inversely proportional to the distance between the points under the satellite, and the combined weighted value is normalized to obtain the position weights of two point traces

And

the time corresponding to the new fusion trace obtained by weighted re-fusion is an intermediate value of two moments;

the original trace point can be replaced by the fused trace point;

if the continuous iterative fusion is carried out, the fusion strategy is repeated until the distance between two adjacent points exceeds the fusion threshold;

wherein d is the distance from the current moment measuring point to the subsatellite point;

for a sensor which cannot calculate fusion weight or has no historical statistical data, the fusion trace point is the mean value of the connected trace points within the fusion distance threshold;

setting strategies by using different fusion weights according to a positioning mechanism, measurement precision and the like of the sensor, wherein measurement errors, a measurement covariance matrix, measurement scores and the distance of the subsatellite point are used as the basis for calculating the fusion weights;

when the distance between the subsatellite points is used as a strategy for setting the fusion weight, the strategy comprises a direct method, a theoretical method and a statistical method;

the direct method comprises the following steps:

calculating the distance between the subsatellite points of two adjacent point traces;

calculating the current positioning precision estimated value according to the section to which the target distance from the sub-satellite point belongs, and taking the reciprocal of the current positioning precision estimated value as the weight of the current trace point and the weight of the next trace point respectively and recording the reciprocal as w'_k-1And w'_k；

To w'_k-1And w'_kNormalization processing is carried out to obtain the weight w finally used for fusion_k-1And w_k；

The theoretical method comprises the following steps:

calculating a positioning error theoretical curve from the zero position of the subsatellite point to the maximum boundary of the precision-preserving area according to the sensor;

calculating the distance between the subsatellite points of two adjacent point traces;

calculating the positioning errors of two adjacent point traces based on a positioning error theoretical curve, and respectively taking the reciprocal of the positioning errors as the weight of the current point trace and the weight of the next point trace, and recording as w'_k-1And w'_k；

To w'_k-1And w'_kNormalization processing is carried out to obtain the weight w finally used for fusion_k-1And w_k；

The statistical method comprises the following steps:

carrying out statistical processing on the historical positioning accuracy condition of the sensor to obtain a positioning error statistical curve related to the distance between the subsatellite points, and obtaining positioning error statistical curves under different target characteristics according to the electromagnetic characteristics or the optical characteristics of the target;

calculating the distance between the subsatellite points of two adjacent point traces;

calculating the positioning errors of two adjacent point traces based on a positioning error theoretical curve, and respectively taking the reciprocal of the positioning errors as the weight of the current point trace and the weight of the next point trace, and recording as w'_k-1And w'_k；

To w'_k-1And w'_kNormalization processing is carried out to obtain the weight w finally used for fusion_k-1And w_k。

5. The method of claim 1, wherein in step (b), under an interactive multi-model architecture, a set of state variables and state transition models is selected, and a filtering and smoothing method is used to estimate the speed and heading of the water surface target.

6. The method of claim 5, wherein step (b) comprises the steps of:

b1, establishing a local northeast coordinate system as a tracking coordinate system by taking the track starting point as the origin, and converting all point tracks on the track from the geographical coordinate system to the local coordinate system;

b2, setting the conversion probability as the probability of a static model and the probability of a cooperative turning model;

b3, establishing a three-axis state equation of the static model and the CT model, wherein the CT equation introduces a turning angular velocity state variable, and establishes an observation equation containing three-axis position measurement;

b4, conjecturing the measurement estimation of the current moment according to the system state estimation and covariance estimation of the previous moment, then carrying out reinitialization calculation on the model, and obtaining a new initial value through a Markov operation matrix among different models;

b5, deriving a nonlinear equation of state and a Jacobian matrix by using EKF;

b6, giving the weight ratio of the current most suitable tracking model by calculating the similarity of the current model and the current moving object state;

b7, obtaining an estimation result according to the individual calculation result and the model matching weight of each model, and finishing filtering processing;

b8, smoothing all the traces based on the filtering result, wherein the smoothing value is restricted by the measurement covariance matrix of the final state and the process noise covariance matrix; finishing the smoothing treatment;

b9, calculating and outputting the heading based on the speed at each moment; if imaging information with time approaching is input, the course is assisted and judged by methods such as identifying the head direction of the target ship and the like;

b10, finishing the processing of all tracks, and finishing the estimation of the speed and the heading.

7. The method of claim 6, wherein in step (b), the speed at time k, v (k), is calculated as:

wherein v is_e(k)、v_n(k) And v_u(k) Respectively are three-axis velocity components in a northeast coordinate system;

defining the heading angle as north-east-off clockwise angle in the range of [0,2 pi), the heading phi (k) at the kth moment is:

8. the method of claim 6, wherein the assisting in determining the heading in step (b9) is performed by:

if only a certain point of the flight path has corresponding imaging information input, identifying the head or tail of the target ship under the condition that the resolution ratio allows, and giving a target course;

if multiple tracks in the flight path have corresponding imaging information input, obtaining a course estimation value according to the heading direction and the trail after static frame image recognition, generating the flight path according to a multi-frame imaging information positioning result, and outputting the flight speed and the course estimation through filtering smoothing;

and integrating the estimated values of the navigational speed and the course, completing multi-party information, and fusing to generate the final estimated values of the navigational speed and the course.

9. The method of claim 8, wherein the fused estimate of speed and heading comprises:

s1, generating a track according to the imaging information positioning result, and outputting the estimation of the initial speed and the initial course in a filtering and smoothing mode;

s2, identifying the bow of the water surface target in each frame to obtain the course of the water surface target in the static frame;

s3, carrying out weighted fusion on the headings obtained in the step (S1) and the step (S2), wherein the weighted value is an empirical value or is obtained by Kalman filtering/confidence coefficient evaluation fusion;

if the difference between the trace point detection time of the imaging information and the trace point detection time of the sensor is larger, carrying out the subsequent steps, otherwise, directly outputting the estimated values of the course and the navigational speed;

s4, carrying out fusion processing on the flight path generated by low measurement precision and the flight path generated by imaging information to complete the two flight path information;

the fusion processing mode is covariance weighted track fusion or adaptive track fusion;

and S5, filtering and smoothing the merged track to obtain and output estimated values of the course and the speed.

Images