Angle measurement data association method under unknown position condition based on minimum angular distance sum

Abstract

The invention discloses a minimum angular distance sum-based angle measurement data association method under an unknown position condition. The method comprises the following steps of: inputting angle measurements of all optical sensors; based on the angle measurement of each optical sensor, calculating an angular distance between the angle measurement of any two optical sensors; taking the minimum angular distance sum corresponding to the incidence relation as a target, and constructing an incidence relation target function; and solving the objective function to obtain an optimal association result. According to the invention, the calculation process does not involve the position information of the optical sensor, so that the method gets rid of the dependence of a classical method on the position information of the optical sensor, and is suitable for many-to-many combat scenes based on a low-cost mobile platform, such as unmanned aerial vehicle swarm confrontation. And the problem that the performance of a data association algorithm is reduced due to platform navigation positioning function failure or insufficient precision in an electromagnetic interference environment is avoided. The result of data association can provide information support for target allocation and collaborative strike.

Description

Technical Field

[0001] The present invention relates to the field of sensor information processing, and in particular to an angle measurement data association method based on a minimum angular distance and an unknown position condition. Background Art

[0002] Using multi-platform collaborative defense against drone swarms, warhead swarms and other target groups can effectively improve interception efficiency. Data association technology can determine the correspondence between target data of different platforms, laying the foundation for target allocation and coordinated strikes. Considering factors such as accuracy, volume, weight, cost, and power consumption, many platforms will choose optical sensors (infrared or visible light) for target detection. Due to the small size and long distance of the target, they often have only a few pixels (i.e., point targets) on the imaging plane of the optical sensor. In addition, swarms or multi-warhead targets often have the same appearance, which makes it difficult to associate them through features such as color, shape, and texture. Angle measurement has become an important way to associate target data. Researchers usually judge the correlation between angle measurements based on the distance between lines of sight. The so-called line of sight refers to a ray determined by the target angle measurement direction starting from the optical sensor position. Under ideal conditions without measurement errors, the lines of sight corresponding to the same target intersect at the target location, and the distance between them is zero. Therefore, the closer the distance between the lines of sight, the greater the possibility that they correspond to the same target. Calculating the target line of sight requires the precise knowledge of the sensor position coordinates. The greater the spatial density of the target, the higher the requirement for the position accuracy of the sensor itself. For mobile platforms, satellite navigation or inertial navigation systems are usually used to determine their own position coordinates. Satellite signals may be interfered with or blocked, and the inertial navigation positioning error accumulates over time, often making it difficult to meet the accuracy requirements required for line of sight intersection. If the correlation between target data can still be determined under the condition that the sensor position is unknown, it can not only avoid problems such as interference with the navigation and positioning system or the introduction of measurement errors, but also save the high cost of high-precision positioning systems.

[0003] For the multi-sensor multi-target data association problem under the condition of unknown sensor positions, researchers usually rely on the mutual topological relationship between target data for data association, and propose a data association algorithm based on the reference pattern (REP). The so-called reference pattern refers to the topological structure implied between a given target position and its adjacent target positions. The problem with this type of method is that they all assume that the sensor can obtain complete target position information (such as xyz coordinates in a Cartesian coordinate system, or distance and angle information in a polar coordinate system), which is essentially solving the correlation problem between point sets. However, the present invention seeks to solve the correlation problem between angles (lines of sight), which is to infer the correlation relationship between three-dimensional targets through two-dimensional measurements, and to infer high-dimensional states using low-dimensional information, which is more difficult. Summary of the invention

[0004] In view of the shortcomings of the prior art, the present invention provides an angle measurement data association method under unknown position conditions based on the minimum angular distance and.

[0005] To achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0006] The present invention provides an angle measurement data association method under unknown position conditions based on the minimum angular distance and the position, comprising:

[0007] Input angle measurements of all optical sensors;

[0008] Calculating the angular distance between the angle measurements of any two optical sensors based on the angle measurements of each optical sensor;

[0009] The association objective function is constructed with the goal of finding the minimum angular distance corresponding to the association relationship;

[0010] Solve the objective function to obtain the best correlation result.

[0011] Further, for the jth and kth optical sensors, j=1,...,N S ,k=1,...,N S , and j≠k, the angular distance between any two angle measurements of the jth optical sensor and the kth optical sensor is calculated by the following steps:

[0012] Get the pth angle measurement of the jth optical sensor and the qth angle measurement of the kth optical sensor Where p = 1, ..., M j ,q=1,...,M k ;M j represents the number of angle measurements in the jth optical sensor angle measurement set, M k represents the number of angle measurements in the k-th optical sensor angle measurement set;

[0013] Get The corresponding angle measurement direction vector Their corresponding equivalent target sight lines are

[0014] based on calculate The angular distance between

[0015]

[0016] Among them, dot represents the vector dot product, |·| represents the vector modulus; Represents equivalent target sight The angle formed.

[0017] Furthermore, suppose there are N S Optical sensors, where N S =2, with the goal of finding the minimum angular distance corresponding to the association relationship, the association relationship objective function is constructed as follows:

[0018]

[0019]

[0020] Among them, a* represents the best correlation result; is a binary indicator function. When (i1,i2) forms a tuple, otherwise

[0021] Furthermore, the solution method of the association relationship objective function includes the Hungarian algorithm and the auction algorithm.

[0022] Furthermore, suppose there are N S Optical sensors, where N S =3, with the goal of finding the minimum angular distance corresponding to the association relationship, the association relationship objective function is constructed, including:

[0023] Based on the angular distance between the angular measurements of any two optical sensors, a correlation cost matrix between the measurements of any two optical sensors is constructed;

[0024] Calculate the m-best correlation results of the correlation cost matrix between any two optical sensor measurements respectively;

[0025] Based on the m-best correlation results, a set of multi-tuples with a high probability of correlation between angle measurements of any two optical sensors is obtained;

[0026] For each tuple in the tuple set, calculate the associated cost corresponding to each tuple;

[0027] The association cost matrix between multi-optical sensor measurements is constructed based on the association cost corresponding to each multi-tuple.

[0028] Furthermore, the construction of a correlation cost matrix between any two optical sensor measurements based on the angular distance between the angular measurements of any two optical sensors includes:

[0029] Construct the correlation cost matrix CM between any two sensor measurements (1,2) , C.M. (1,3) and CM (2,3) ,as follows:

[0030]

[0031] Further, the respectively calculating m-best association results of the association cost matrix between any two optical sensor measurements includes:

[0032] Using Murty's method, CM is calculated separately. (1,2) , C.M. (1,3) and CM (2,3) The m-best association results (j,k)∈{(1,2),(1,3),(2,3)}; is the i-th correlation result between the measurements of sensors j and k.

[0033] Furthermore, the method of obtaining a set of multi-tuples having a greater possibility of correlation between angle measurements of any two optical sensors based on the m-best correlation results includes:

[0034] For each Perform the following operations

[0035]

[0036] Get the union of the multi-tuples in the m-best association results measured in optical sensors j and k (j,k) .

[0037] Furthermore, for each tuple in the tuple set, calculating the associated cost corresponding to each tuple includes:

[0038] Traverse all (i1, i2, i3) and calculate their corresponding association costs by the following formula

[0039]

[0040] in, Represents the cost of the tuple (i1,i2,i3).

[0041] Furthermore, the correlation cost matrix between the multi-optical sensor measurements constructed based on the correlation cost corresponding to each multi-tuple is as follows:

[0042]

[0043] in, is an indicator function. When (i1,i2,i3) form a tuple, otherwise

[0044] Compared with the prior art, the beneficial technical effects of the present invention are:

[0045] The method for associating angle measurement data of unknown position sensors based on the minimum angular distance and provided by the present invention does not involve the position information of the optical sensor itself in the calculation process, thus getting rid of the dependence of the classical method on the position information of the optical sensor itself. The method described in the present invention is suitable for many-to-many combat scenarios based on low-cost mobile platforms such as drone swarm confrontation, avoiding the problem of data association algorithm performance degradation caused by the failure of platform navigation and positioning functions or insufficient accuracy in electromagnetic interference environments. The data association results provided by the present invention can provide information support for target allocation and coordinated strikes. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying creative work.

[0047] Figure 1 A schematic flow chart of a method for associating angle measurement data of an unknown position sensor based on a minimum angular distance and provided in an embodiment;

[0048] Figure 2 Schematic diagram of the effect of the difference in optical sensor observation angles on the data association algorithm based on the reference mode, where (a) is a schematic diagram when the optical sensors' observation angles are almost the same, and (b) is a schematic diagram when the optical sensors' observation angles are greatly different;

[0049] Figure 3 A schematic diagram of an equivalent data association scenario that does not rely on the position information of the optical sensor provided by an embodiment, wherein Figure 3 (a) is a classic data association scenario. Figure 3 (b) is the equivalent data association scenario;

[0050] Figure 4 A schematic diagram of angular distance definition provided in an embodiment;

[0051] Figure 5 A schematic diagram of target angle measurement calculation provided by an embodiment;

[0052] Figure 6 A two-optical sensor scenario experiment design diagram provided in an embodiment, wherein Figure 6 (a) is the side view of posture combination 1. Figure 6 (b) is the front view of posture combination 1. Figure 6 (c) is the front view of posture combination 2. Figure 6 (d) is the front view of posture combination 3. Figure 6 (e) is a front view of posture combination 4;

[0053] Figure 7 A diagram of an experimental scene of two optical sensors provided in an embodiment, wherein Figure 7 (a) is a schematic diagram of sensor posture combination 1. Figure 7 (b) is a schematic diagram of sensor posture combination 2. Figure 7 (c) is a schematic diagram of sensor posture combination 3. Figure 7 (d) is a schematic diagram of sensor posture combination 4;

[0054] Figure 8 for Figure 7 The actual picture of the hanging rack is shown in the figure marked with 1;

[0055] Fig. 9 for Figure 7 The actual picture of the luminous beads is shown in the figure marked 2;

[0056] Fig.10 for Figure 7 The actual picture of an industrial camera equipped with a gesture sensor is shown in the figure marked 3, where Fig.10 (a) is the overall schematic diagram of an industrial camera equipped with a gesture sensor. Fig.10 (b) is a real picture of an industrial camera. Fig.10 (c) is a physical picture of the attitude sensor;

[0057] Fig.11 A target photo taken by a camera provided in an embodiment;

[0058] Fig.12 The figure is a graph of the experimental results of two optical sensors provided in an embodiment, wherein Fig.12 (a) is an example of the association result of pose combination 1. Fig.12 (b) is an example of the association result of pose combination 2. Fig.12 (c) is an example of the association result of pose combination 3. Fig.12 (d) is an example of the association result of pose combination 4;

[0059] Fig.13 A three-optical sensor scenario experiment design diagram provided in an embodiment;

[0060] Fig.14 A diagram of a three-optical sensor experimental scene provided in an embodiment;

[0061] Fig.15 A diagram showing the experimental results of three optical sensors provided in an embodiment;

[0062] Figure annotation:

[0063] 1. Hanging rack; 2. Luminous beads; 3. Industrial camera equipped with gesture sensor. DETAILED DESCRIPTION

[0064] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0065] The reference pattern method in the prior art uses the relative position relationship of the targets on the optical sensor imaging plane as a reference pattern to calculate the correlation relationship between the targets. However, this method is only applicable to scenes with almost the same orientation of the optical sensors. As the difference in sensor orientation increases, the relative position relationship of the targets on the imaging plane will undergo perspective transformation, and the original position relationship will change, causing the above method to fail. Figure 2 ,exist Figure 2 In (a), the observation angles of the two optical sensors are almost the same, and the reference patterns of target A on the imaging planes of the two sensors are also almost the same. In this case, the reference pattern method is effective. Figure 2 In (b), the observation angles of the two sensors are quite different, and the reference patterns of target A on the imaging planes of the two sensors are quite different, so the reference pattern method fails.

[0066] Reference Figure 1 In one embodiment, a method for associating angle measurement data of unknown position sensors based on minimum angular distance and is provided, comprising:

[0067] Input angle measurements of all optical sensors;

[0068] Calculating the angular distance between the angle measurements of any two optical sensors based on the angle measurements of each optical sensor;

[0069] The association objective function is constructed with the goal of finding the minimum angular distance corresponding to the association relationship;

[0070] Solve the objective function to obtain the best correlation result.

[0071] Reference Figure 5 , obtaining the angle measurement includes the following steps:

[0072] Assume that the target is at the optical sensor S s The coordinates on the imaging plane are First, calculate the deflection angle of the target in the optical sensor coordinates Pitch Angle

[0073]

[0074] Among them, cu and cv represent the coordinates of the camera's principal point; fu and fv represent the dimensions of the focal length converted into pixel coordinates in the u and v directions;

[0075] Use the attitude sensor to calculate the direction vector of the optical axis of the optical sensor in the world coordinate system; assume that the attitude sensor reading is φ x ,φ y ,φ z , rotate the y-axis direction vector in the world coordinate system around the z-axis, x-axis, and y-axis by φ x ,φ y ,φ z The direction vector of the optical axis of the optical sensor in the world coordinate system can be obtained; the direction vector of the main axis of the optical sensor is rotated around the z′ axis first Then rotate around the x' axis The direction vector of the optical axis of the optical sensor in the world coordinate system can be obtained

[0076] Angle measurement of the target, that is, the deflection angle of the target in the world coordinate system Pitch Angle Calculated by the following formula

[0077]

[0078] From the above content, it can be seen that calculating the target angle measurement through an optical sensor requires knowing the coordinates of the target on the imaging plane, the internal parameters of the sensor, and the posture of the sensor, but does not require knowing the position information of the sensor itself.

[0079] The set of angle measurements from all optical sensors is denoted as The angle measurement set of the sth optical sensor is recorded as

[0080]

[0081] Among them, N S Represents the number of optical sensors; Represents the set Z (s) The i-th s Angle measurement; M s Represents Z (s) The number of mid-angle measurements;

[0082]

[0083] in, Respectively represent the deflection angle and pitch angle of the target in the world coordinate system; w A Represents the angle measurement error.

[0084] The correlation result a divides the angle measurement set of all optical sensors into n T subsets, using multiple groups Record No. i T The number of the angle measurement in the subset,

[0085]

[0086] Among them, ξ records i T and its corresponding tuple The mapping relationship between them.

[0087] The association result is recorded as a

[0088]

[0089] Among them, n T Represents the hypothetical target quantity.

[0090] Since each target can only form one angle measurement on an optical sensor, and each tuple can only correspond to one target, the indexes of the angle measurements should satisfy a one-to-one mapping constraint. If an optical sensor misses the target corresponding to a tuple, a dummy measurement number 0 is added to the corresponding index position of the tuple.

[0091] If a certain association result a and its contained tuples satisfy the above constraints, then a is called a feasible solution. The set of all feasible solutions is denoted as The correctness of a is usually evaluated by the association result cost function Cost(a). Generally speaking, the greater the possibility that the association result is correct, the smaller the corresponding cost value. Based on the above, solving the data association problem is equivalent to finding a feasible solution a that minimizes the cost of the association result. The solution process can be expressed as

[0092]

[0093] Among them, a * Indicates the best correlation result.

[0094] For the jth and kth optical sensors, j=1,...,N S ,k=1,...,N S , and j≠k, the angular distance between any two angle measurements of the jth optical sensor and the kth optical sensor is calculated by the following steps:

[0095] Get the pth angle measurement of the jth optical sensor and the qth angle measurement of the kth optical sensor Where p = 1, ..., Mj ,q=1,...,M k ;M j represents the number of angle measurements in the jth optical sensor angle measurement set, M k represents the number of angle measurements in the k-th optical sensor angle measurement set;

[0096] Get The corresponding angle measurement direction vector Their corresponding equivalent target sight lines are

[0097] The so-called equivalent target sight line comes from an equivalent data association scenario provided by the present invention that does not rely on the position information of the optical sensor. Classic data association scenarios such as Figure 3 As shown in (a), the equivalent scenario is Figure 3 (b) The classic scenario can be understood as the following process: the target is stationary, and the sensor generates a measurement set Z at position S1 (1) , and then keep the posture unchanged, along the vector Direction translation to S2 position generates measurement set Z (2) The equivalent scenario can be understood as the following process: Assume that the sensor generates a measurement set Z at position S1 (1) , and then the sensor is stationary and the target moves along the vector Direction moving distance Go to the new position and generate the measurement set Z (2) More generally, in an equivalent scenario, the sensor position is recorded as S; the starting position of the target is recorded as T1 (1) , The set of targets located at these positions is called “equivalent target set 1”, corresponding to the angle measurements generated by sensor S1; the targets are classified according to vector The positions after the move are recorded as T1 (2) , The set of targets located at these positions is called "equivalent target set 2", corresponding to the angle measurements generated by sensor S2. Obviously, solving the data association problem in scenario 1 is equivalent to solving the data association problem in scenario 2.

[0098] based on calculate The angular distance between

[0099]

[0100] Among them, dot represents the vector dot product, |·| represents the vector modulus; Represents equivalent target sight The angle formed.

[0101] In one embodiment, referring to Figure 4 , assuming The corresponding angle measurement direction vectors are Their corresponding equivalent target sight lines are ST1 (1) 、ST1 (2) ,So and The angular distance between them is:

[0102]

[0103] The angular distance and SumAD(a) corresponding to the correlation result a are defined as

[0104]

[0105] The angular distance sum SumAD(a) corresponding to the association result a refers to the sum of the angular distances corresponding to all association relationships in the association result.

[0106] In one embodiment, assume that there are N S Optical sensors, where N S =2, with the goal of finding the minimum angular distance corresponding to the association relationship, the association relationship objective function is constructed as follows:

[0107]

[0108] Among them, a * represents the best correlation result; is a binary indicator function. When (i1,i2) forms a tuple, otherwise

[0109] Formula (12) (13) is a standard two-dimensional assignment problem, and the cost between any two data can be expressed by the angular distance. The solution method of the association relationship objective function includes the Hungarian algorithm and the auction algorithm.

[0110] In one embodiment, assume that there are N S Optical sensors, where N S =3, with the goal of finding the minimum angular distance corresponding to the association relationship, the association relationship objective function is constructed, including:

[0111] Based on the angular distance between the angular measurements of any two optical sensors, a correlation cost matrix between the measurements of any two optical sensors is constructed;

[0112] Calculate the m-best correlation results of the correlation cost matrix between any two optical sensor measurements respectively;

[0113] Based on the m-best correlation results, a set of multi-tuples with a high probability of correlation between angle measurements of any two optical sensors is obtained;

[0114] For each tuple in the tuple set, calculate the associated cost corresponding to each tuple;

[0115] The association cost matrix between multi-optical sensor measurements is constructed based on the association cost corresponding to each multi-tuple.

[0116] The correlation cost matrix between multiple optical sensor measurements constructed based on the correlation cost matrix corresponding to each multi-tuple is as follows:

[0117]

[0118] in, is an indicator function. When (i1,i2,i3) form a tuple, otherwise represents the cost of the tuple (i1,i2,i3),

[0119]

[0120] Formula (14) is a standard generalized 3D assignment problem, which can be solved by Lagrangian relaxation technique. The cost function expressed in Formula (16) can effectively solve the association ambiguity problem.

[0121] The so-called cost matrix contains three dimensions, and each element records the cost corresponding to a certain association relationship (i.e., a multi-tuple). Obviously, the association cost matrix CM can be directly constructed by traversing all (i1, i2, i3) and passing equations (10) and (16). However, the association cost matrix constructed in this way is dense. The "dense" here means that all elements in the matrix have values. Although theoretically speaking, this "density" does not affect the attribution of the correct association result, for the Lagrangian relaxation technology, this "density" seriously affects its optimization quality. This is mainly because the "dense" matrix makes the search space for feasible solutions huge, resulting in greater uncertainty in the optimization results.

[0122] The method of constructing a correlation cost matrix between the angle measurements of any two optical sensors based on the angular distance between the angle measurements of any two optical sensors includes:

[0123] Construct the correlation cost matrix CM between any two sensor measurements (1,2) , C.M. (1,3) and CM (2,3) ,as follows:

[0124]

[0125] The m-best association results of the association cost matrix between any two optical sensor measurements are calculated respectively, including:

[0126] Using Murty's method, CM is calculated separately. (1,2) , C.M. (1,3) and CM (2,3) The m-best association results (j,k)∈{(1,2),(1,3),(2,3)}; is the i-th correlation result between the measurements of sensors j and k.

[0127] The method of obtaining a set of multi-tuples having a greater possibility of correlation between angle measurements of any two optical sensors based on the m-best correlation results includes:

[0128] For each Perform the following operations

[0129]

[0130] Get the union of the multi-tuples in the m-best association results measured in optical sensors j and k (j,k) ; The obtained ε (j,k) It contains the multi-tuples with a high probability of association between the measurements of sensors j and k, and this probability is set by the size of the m-best value.

[0131] For each tuple in the tuple set, calculating the associated cost corresponding to each tuple includes:

[0132] Traverse all (i1, i2, i3) and calculate their corresponding association costs by the following formula

[0133]

[0134] in, Represents the cost of the tuple (i1,i2,i3).

[0135] After the above processing, the multi-optical sensor association cost matrix CM becomes sparse, in which a large number of element values ​​are equal to Infinity, and the corresponding number of feasible solutions is greatly reduced, which reduces the difficulty of optimization and effectively improves the association accuracy of the algorithm. In general, the smaller the value of m-best, the easier it is to find the optimal solution. When m-best is equal to 1, the number of feasible solutions in the solution space is 1, and the association result can be obtained without optimization. However, this may miss some possible association results (for example, for scenarios where there is association ambiguity between two sensors), resulting in a decrease in the association accuracy. Therefore, the value of m-best is usually set to be greater than 1, and it gradually increases with the increase in the number of targets.

[0136] In one embodiment, referring to Figure 6 , Figure 6 The following is a design diagram for a two-sensor scenario experiment, which includes two industrial cameras, denoted as S1 and S2. Fig.10 , each camera is equipped with a posture sensor to measure the camera posture. Fig. 9 , use luminous beads to simulate the target, refer to Figure 8 , tied to the rack with a transparent thin line. The rack is 1.8m away from the camera. The side view of the scene is shown in Figure 6 (a). In order to verify the effectiveness of the algorithm under different observation angles, four different sensor pose combinations were designed for the experiment. Figure 6 (b)-(e). In pose combination 1, the two cameras face the target group, with a distance of about 0.1m and a height of about 0.3m from the ground. Compared with pose combination 1, the sensor S1 in pose combination 2 rolls about 30 degrees around the optical axis. Pose combinations 3 and 4 gradually increase the angle between the optical axes of the sensors.

[0137] The actual experimental scene is as follows Figure 7 As shown. The image resolution of both cameras is 1600 pixels * 1200 pixels. The internal parameters are obtained by Zhang Zhengyou calibration method. The specific values ​​are shown in Table 1, where cu and cv represent the coordinates of the camera principal point, and fu and fv represent the dimensions of the focal length converted into pixel coordinates in the u and v directions. The attitude sensor uses the WT9011DCL-BT50 attitude sensor produced by Shenzhen Weite Intelligent Company. The attitude sensor readings of sensors S1 and S2 in the posture combination 1-4 are shown in Table 2. The diameter of the target luminous beads is 0.08m, and they are distinguished by round stickers of different colors.

[0138] Table 1 Camera internal parameters

[0139]

[0140] Table 2 Attitude sensor readings

[0141]

[0142] The experimental process includes 3 steps

[0143] Step 1: Turn off all lights in the room and use two cameras to take photos of the target (photo examples are as follows: Fig.11 As shown), the target on the image is then detected by threshold segmentation and connected domain analysis, and the target angle measurement is calculated using the method described in the present invention.

[0144] Step 2: Use the algorithm proposed in the present invention to solve the correlation between the target angle measurements obtained by the two sensors, and number the targets in the scene according to the correlation results.

[0145] Step 3: Turn on the lights in the room and compare the data association results with the real scene. If the stickers corresponding to the targets with the same numbers have the same color, the association results are correct.

[0146] Fig.11 The experimental process and experimental results of four posture combination scenarios are shown. The following takes sensor posture combination 4 as an example (i.e. Fig.12 (d)) is introduced in detail.

[0147] Step 1 Experimental results: In a dark environment, S1 and S2 detect the target in their field of view respectively and obtain the target angle measurement set

[0148] Step 2 Experimental results: After calculation by the algorithm proposed in this patent, the association result a = {(1,3), (2,1), (3,4), (4,2), (5,5)} is obtained. The targets in the scene are numbered accordingly, and the correspondence between the angle measurement of each sensor and the target is shown in Table 3, where the targets are numbered according to the order of the binary groups in the association result.

[0149] Table 3. Correspondence between angle measurement and target

[0150]

[0151] Step 3 Experimental results: Check whether the sticker colors corresponding to the targets with the same numbers in the two sensors are the same. The inspection results are shown in Table 4.

[0152] Table 4 Correlation result correctness checklist

[0153]

[0154] The inspection results show that the stickers corresponding to the same numbered targets have the same color and the association results are correct. The experimental results show that the data association algorithm assigns the same number to the same target detected by different sensors, so that different sensors form a unified understanding of the targets in the scene. The association results of the other three pose combinations are all correct through the same method.

[0155] In one embodiment, referring to Fig.13 , Fig.13 The following is a design diagram for the 3-sensor scenario experiment, which includes 3 industrial cameras, denoted as S1, S2, and S3. The experimental method is similar to that of the 2-sensor scenario and will not be repeated here. Fig.14 As shown, the internal parameters of the three cameras are shown in Table 5, and their corresponding attitude sensor readings are shown in Table 6.

[0156] The same steps as the two-sensor scenario are used.

[0157] Step 1 Experimental results: In a dark environment, the three cameras detect the targets in their fields of view and obtain the target angle measurement set.

[0158] Step 2 Experimental results: After calculation by the algorithm proposed in this patent, the association result a = {(1,2,2), (2,4,4), (3,1,1), (4,3,3), (5,5,5)} is obtained. The targets in the scene are numbered accordingly, and the correspondence between the angle measurement of each sensor and the target is shown in Table 5, where the targets are numbered according to the order of the multi-tuple in the association result.

[0159] Step 3 Experimental results: Check whether the sticker colors corresponding to the targets with the same numbers in the three sensors are the same. The inspection results are shown in Table 6.

[0160] Table 5. Correspondence between angle measurement and target

[0161]

[0162] Table 6 Correlation result correctness checklist

[0163]

[0164] Reference Fig.15 ,The inspection results show that the stickers corresponding to the targets with the same ,number have the same color and the association result is correct.

[0165] Matters not covered by the present invention are known technologies.

[0166] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0167] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention. It should be pointed out that, for a person of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

[0168] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. Angle measurement data association method based on minimum angular distance and position unknown condition, characterized in that: include: Input angle measurements of all optical sensors; Calculating the angular distance between the angle measurements of any two optical sensors based on the angle measurements of each optical sensor; To find the minimum angular distance corresponding to the correlation relationship, an association relationship objective function is constructed; Solve the objective function to obtain the best correlation result.

2. The method for associating angle measurement data under unknown position conditions based on the minimum angular distance and the position as claimed in claim 1, characterized in that: For the jth and kth optical sensors, j=1,...,N S ,k=1,...,N S , and j≠k, the angular distance between any two angle measurements of the jth optical sensor and the kth optical sensor is calculated by the following steps: Get the pth angle measurement of the jth optical sensor and the qth angle measurement of the kth optical sensor Where p = 1, ..., M j ,q=1,...,M k ;M j represents the number of angle measurements in the jth optical sensor angle measurement set, M k represents the number of angle measurements in the k-th optical sensor angle measurement set; Get The corresponding angle measurement direction vector Their corresponding equivalent target sight lines are based on calculate The angular distance between Among them, dot represents the vector dot product, |·| represents the vector modulus; Represents equivalent target sight The angle formed.

3. The method for associating angle measurement data under unknown position conditions based on the minimum angular distance and the position of claim 1, characterized in that: In the current scenario, there are N S Optical sensors, where N S =2, with the goal of finding the minimum angular distance corresponding to the association relationship, the association relationship objective function is constructed as follows: Among them, a* represents the best correlation result; is a binary indicator function. When (i1,i2) forms a tuple, otherwise 4. The method for associating angle measurement data under unknown position conditions based on the minimum angular distance and the position as claimed in claim 3, characterized in that: The solution method of the association relationship objective function includes the Hungarian algorithm and the auction algorithm.

5. The method for associating angle measurement data under unknown position conditions based on the minimum angular distance and according to claim 2, characterized in that: In the current scenario, there are N S Optical sensors, where N S =3, with the goal of finding the minimum angular distance corresponding to the association relationship, the association relationship objective function is constructed, including: Based on the angular distance between the angular measurements of any two optical sensors, a correlation cost matrix between the measurements of any two optical sensors is constructed; Calculate the m-best correlation results of the correlation cost matrix between any two optical sensor measurements respectively; Based on the m-best correlation results, a set of multi-tuples with a high probability of correlation between angle measurements of any two optical sensors is obtained; For each tuple in the tuple set, calculate the associated cost corresponding to each tuple; The association cost matrix between multi-optical sensor measurements is constructed based on the association cost corresponding to each multi-tuple.

6. The method for associating angle measurement data under unknown position conditions based on the minimum angular distance and the position as claimed in claim 5, characterized in that: The method of constructing a correlation cost matrix between the angle measurements of any two optical sensors based on the angular distance between the angle measurements of any two optical sensors includes: Construct the correlation cost matrix CM between any two sensor measurements (1,2) , C.M. (1,3) and CM (2,3) ,as follows:

7. The method for associating angle measurement data under unknown position conditions based on the minimum angular distance and the position as claimed in claim 5, characterized in that: The m-best association results of the association cost matrix between any two optical sensor measurements are calculated respectively, including: Using Murty's method, CM is calculated separately. (1,2) , C.M. (1,3) and CM (2,3) The m-best association results (j,k)∈{(1,2),(1,3),(2,3)}; is the i-th correlation result between the measurements of sensors j and k.

8. The method for associating angle measurement data under unknown position conditions based on the minimum angular distance and as claimed in claim 5, characterized in that: The method of obtaining a set of multi-tuples having a greater possibility of correlation between angle measurements of any two optical sensors based on the m-best correlation results includes: For each Perform the following operations Get the union of the multi-tuples in the m-best association results measured in optical sensors j and k (j,k) .

9. The method for associating angle measurement data under unknown position conditions based on the minimum angular distance and as claimed in claim 5, characterized in that: For each tuple in the tuple set, calculating the associated cost corresponding to each tuple includes: Traverse all (i1, i2, i3) and calculate their corresponding association costs by the following formula in, Represents the cost of the tuple (i1,i2,i3).

10. The method for associating angle measurement data under unknown position conditions based on minimum angular distance and as claimed in claim 5, characterized in that: The correlation cost matrix between multiple optical sensor measurements constructed based on the correlation cost corresponding to each multi-tuple is as follows: Among them, ri1i2i3 is an indicator function. When (i1, i2, i3) forms a multi-tuple, r i1i2i3 =1, otherwise r i1i2i3 =0.

Images