A method and device for infrared target type recognition based on emission position guided sampling matching

Abstract

The present invention relates to the fields of remote sensing image processing and machine vision applications, and provides an infrared target type recognition method and device based on emission position guided sampling matching, the main purpose of which is to solve the problem of excessive consumption of computing resources in traditional global template matching. The main scheme includes pre-establishing a mapping relationship between the emission position and the possible types of the target, first estimating the emission position in combination with the target motion information and energy information during the prediction process, and then reducing the number of types to be recognized based on the mapping relationship, reducing the templates to be matched, and improving the accuracy and speed of target type recognition. The main technologies include establishing a mapping relationship between the emission position and the possible types of the target; combining target motion information, energy information, or multi-target collaborative information to estimate the longitude and latitude of the emission position; using the obtained longitude and latitude of the emission position, and combining the established mapping model to sample the global type matching template to obtain a type vector, thereby achieving the final infrared target type recognition.

Description

Technical Field

[0001] The present invention belongs to the field of remote sensing image processing and machine vision applications, and in particular relates to an infrared target type recognition method and device based on emission position guided sampling matching. Background Art

[0002] Infrared target type recognition is the core of infrared search and tracking systems. Currently, this is typically achieved by building a global library of type templates and matching them one by one. However, due to the vast number of target types within the global domain, matching them one by one would waste significant computing resources. Therefore, the key is to filter out a subset of target types from the vast array of target types. Each target must originate from a specific launch location within a fixed area. Therefore, a mapping model can be pre-established based on the regional latitude and longitude range, the launch location's latitude and longitude, and the target type. For the target to be predicted, the longitude and latitude coordinates of the launch point can be inferred based on known information combined with physical processes. This determines the region and location of the launch point. The mapping model then filters out the required types and sorts them from closest to farthest distance. Therefore, by establishing a mapping model and predicting the longitude and latitude of the launch point of the target to be identified, the number of types to be identified can be reduced. Summary of the Invention

[0003] The purpose of the present invention is to solve the problem that a large amount of computing resources will be wasted when the existing method needs to construct a global type template library and match it one by one to achieve type recognition by guiding the sampling matching library through the emission position. While ensuring the recognition accuracy, the speed of infrared target type recognition is improved.

[0004] The technical solution adopted in the present invention is as follows:

[0005] The present invention provides an infrared target type recognition method based on emission position guided sampling matching, comprising the following steps:

[0006] Step 1: Pre-establish a mapping model. Pre-establish the mapping relationship between launch locations and possible target types. Specifically, this includes mapping between real launch locations and type lists, mapping between virtual launch locations and type lists, and mapping between regional geographic information and launch location lists. Finally, establish a mapping between regional geographic information and launch location lists and type lists to obtain a mapping model and proceed to step 2.

[0007] Step 2: Estimation of the launch position, which is achieved by combining target motion information and energy information, or multi-target collaborative information to estimate the launch position latitude and longitude, obtain the launch position latitude and longitude, and proceed to step 3;

[0008] Step 3: Use the longitude and latitude of the emission position obtained in step 2 and the mapping model established in step 1 to sample the global type matching template to obtain the type vector and realize the final infrared target type recognition.

[0009] In the above scheme, step 1 includes the following steps:

[0010] Create a mapping between real launch location and type list:

[0011] Step 1.1: Create a mapping between the actual launch location and the type list based on the actual launch information:

[0012] f1: L1→M1

[0013] Where f1 represents a function that maps elements in set L1 to elements in set M1, (rx, ry)∈L1 represents the longitude and latitude of the actual launch location, and M1 is a set of all type lists contained in the actual launch location;

[0014] Create a virtual launch location-type list mapping:

[0015] Step 1.2: Combine historical target launch data to establish a virtual launch position-type list mapping, and calculate the launch time t of the i-th data of each type i :

[0016] t i =inf{t i |E i (t) = Fitting(T i , E i )<θ1}

[0017] Among them, T i 、E i Respectively represent the time variation information and energy variation information of the i-th data, Fitting(T i , E i ) indicates T i The horizontal axis E i Fitting is done for the vertical axis, inf represents the first value that meets the requirements, t i According to T i and E i The fitting relationship predicts the time forward until the energy is less than the threshold θ1;

[0018] Step 1.3: Calculate the longitude cx of the location where the i-th data of each type is transmitted i and latitude cy i :

[0019] cx i ={λ i (t i)|λ i (t) = Fitting(T i ,λ i )}

[0020]

[0021] Among them, λ i 、 Respectively represent the longitude change information and latitude change information of the i-th data;

[0022] Step 1.4: Calculate the outlier Z of each type of emission position using the following formula. If Z is less than the threshold θ2, proceed to step 1.5. If Z is greater than the threshold θ2, proceed to step 1.6.

[0023]

[0024] Among them, Z represents the type of emission position outliers, n represents the total number of data contained in the type, and The types respectively indicate the average longitude and latitude of the launch location containing all data. Indicates calculation (cx i ,cy i )arrive distance, Indicates that all data points to The average distance

[0025] Step 1.5: Using the geometric center of all data as the center point, establish the first batch of virtual emission position-type list mappings:

[0026] f2: L2→M2

[0027] Among them, f2 represents the function that maps the elements in the set L2 to the elements in the set M2, (vx G ,vy G )∈L2 represent the longitude and latitude of the virtual launch position with the geometric center of all data, and M2 is the set of all type lists contained in the first batch of virtual launch positions;

[0028] Step 1.6: Clustering to obtain the second batch of virtual emission position-type list mappings:

[0029] f3: L3→M3

[0030] Among them, f3 represents the function that maps the elements in the set L3 to the elements in the set M3, (vx C ,vy C)∈L3 represent the longitude and latitude of the virtual launch position obtained by clustering, M3 is the set of all type lists contained in the second batch of virtual launch positions;

[0031] Step 1.7: Gather the first batch of virtual launch position-type list mappings obtained in step 1.5 and the second batch of virtual launch position-type list mappings obtained in step 1.6 to establish a virtual launch position-type list mapping:

[0032] f4: L4→M4

[0033] Wherein, f4 represents a function that maps elements in set L4 to elements in set M4, (vx, vy)∈L4=L2∪L3 represent the longitude and latitude of the virtual launch position respectively, and M4=M2∪M3 is the set of all type lists contained in the virtual launch position;

[0034] Create a mapping between regional geographic information and launch location list:

[0035] Step 1.8: Combine the actual launch location longitude and latitude, the virtual launch location longitude and latitude obtained in step 1.7, and the regional geographic information to create a mapping between regional geographic information and the launch location list. Divide the regional territory into a number of rectangles such that the union of the rectangles covers the regional territory and minimizes the total area.

[0036]

[0037] Among them, Area k represents the latitude and longitude range covered by region k, m represents the number of rectangles divided into region k, R j Indicates the latitude and longitude range covered by the jth rectangle, It means that m rectangular areas R j merged together to form a larger area;

[0038] Step 1.9: Calculate the minimum distance min between the virtual launch location longitude and latitude (vx, vy) and the latitude and longitude of all real launch locations in the region k to which it belongs d (vx,vy):

[0039]

[0040] Where (rx, ry)∈Area k They respectively indicate that region k contains the longitude and latitude of all real launch locations. The region to which the real launch location belongs and the region to which the virtual launch location belongs are directly determined by the data;

[0041] Step 1.10: Combine all real launch positions and virtual launch positions in the area, and use the minimum distance min obtained in step 1.9 d(vx, vy) is compared with the threshold θ3, if min d (vx,vy) is less than the threshold θ3, then the virtual launch position is considered to coincide with the real launch position, and the longitude and latitude of the virtual launch position are not retained; if min d If (vx, vy) is greater than the threshold θ3, the virtual launch position is considered to be a new launch position different from the real launch position, and the longitude and latitude of the virtual launch position are retained. The formula is as follows:

[0042] L k =(L1∈Area k )∪((vx,vy)∈L4∈Area k and min d (vx,vy)>θ3)

[0043] Among them, (x k ,y k )∈L k Respectively represent the longitude and latitude of all launch locations contained in area k, L k Contains all real launch positions and virtual launch positions that meet the conditions in area k;

[0044] Step 1.11: Create a mapping between regional geographic information and the launch location list:

[0045] f5: A→L

[0046] Among them, f5 represents the function that maps the elements in set A to the elements in set L. Represents all regional geographic information, represents the set of all launch locations, p is the total number of regions, and (x, y)∈L represents the longitude and latitude of the launch location respectively;

[0047] Create a mapping between regional geographic information, launch location list, and type list:

[0048] Step 1.12: Combine the real launch location-type list mapping obtained in step 1.1, the virtual launch location-type list mapping obtained in step 1.7, and the regional geographic information-launch location list mapping obtained in step 1.11 to ultimately establish a regional geographic information-launch location list-type list mapping;

[0049] If, in step 1.10, a virtual launch position coincides with a real launch position, then extract the types not present in the real launch position from the target type list of the virtual launch position and add these types to the real launch position-type list mapping established in step 1.1. Combine the supplemented real launch position-type list mapping with the virtual launch position-type list mapping obtained in step 1.7, and map all types to the set L of all launch positions obtained in step 1.11 to establish a launch position-type list mapping:

[0050] f6: L→M

[0051] Where f6 represents a function that maps elements in set L to elements in set M, where M=M1∪M4 is the set of all type lists;

[0052] Step 1.13: Combine the regional geographic information-transmission location list mapping obtained in step 1.11 with the transmission location-type list mapping obtained in step 1.12 to create a regional geographic information-transmission location list-type list mapping:

[0053] f7: A→L→M

[0054] Among them, f7 is a composite mapping function from set A to set L, and then from set L to set M.

[0055] In the above scheme, step 2 includes the following steps:

[0056] Step 2.1: Calculate the launch time t of target o o :

[0057] t o =inf{t o |E o (t) = Fitting(T o , E o )<θ1}where, T o 、E o Represent the time variation information and energy variation information of target o respectively, Fitting(T o , E o ) indicates T o The horizontal axis E o Fitting is done for the vertical axis, inf represents the first value that meets the requirements, t o According to T o and E o The fitting relationship predicts the time forward until the energy is less than the threshold θ1;

[0058] Step 2.2: Calculate the longitude x of the launch point of target o o and latitude y o :

[0059] x o ={λ o (t o )|λ o (t) = Fitting(T o ,λ o )}

[0060] Among them, λ o 、 They respectively represent the longitude change information and latitude change information of the target o.

[0061] In the above scheme, step 3 includes the following steps:

[0062] Step 3.1: Based on all the regional geographic information A obtained in step 1.11, obtain the longitude and latitude of the launch point obtained in step 2.2 (x o ,y o ) belongs to the area list A((x o ,y o )):

[0063]

[0064] Step 3.2: Based on the regional geographic information-emission location list-type list mapping established in step 1.13, obtain (x o ,y o ) corresponds to the launch position list L((x o ,y o )):

[0065] L((x o ,y o ))=f7(A((x o ,y o )))→Lwhere, (ox,oy)∈L((x o ,y o )) respectively represent (x o ,y o ) The area to which it belongs corresponds to the longitude and latitude of the transmitting location;

[0066] Step 3.3: Calculate (x o ,y o ) and the corresponding launch position list L((x o ,y o )) The distance of each launch position in the sequence is sorted from smallest to largest, and the sorted launch position list L is obtained. S ((x o ,y o )):

[0067] L S ((x o ,y o ))=Sort((kx,ky)|(kx,ky)∈L((x o ,y o )), d((x o ,y o ), (kx,ky)))

[0068] Among them, (sx,sy)∈L S ((x o ,y o )) respectively represent the L((x o ,y o )) After sorting, the longitude and latitude of the launch position;

[0069] Step 3.4: According to the sorted launch position list L S ((x o ,y o ), combined with the launch location list-type list mapping in the regional geographic information-launch location list-type list mapping established in step 1.13, we get (x o ,y o ) corresponds to the type vector M((x o ,y o )):

[0070] M((x o ,y o ))=f7(L S ((x o ,y o )))→M

[0071] M((x o ,y o )) contains (x o ,y o ) All types contained in all emission positions after sorting the area to which they belong, to achieve global type matching template sampling;

[0072] Step 3.5: Based on the type vector obtained in step 3.4, use the type matching algorithm from top to bottom to calculate the target o and the type vector M((x o ,y o )) and select the type Model(o) with the highest similarity as the type of target o to achieve infrared target type recognition:

[0073]

[0074] Among them, S(0,mdl) indicates that the type matching algorithm is used to calculate the target o and M((x o ,y o )) in the type mdl similarity.

[0075] The present invention also provides an infrared target type recognition device based on emission position guided sampling matching, comprising the following steps:

[0076] Mapping model pre-establishment module: Pre-establish the mapping model, pre-establish the mapping relationship between the launch position and the possible target type, specifically including the real launch position-type list mapping, the virtual launch position-type list mapping, the regional geographic information-launch position list mapping, and finally establish the regional geographic information-launch position list-type list mapping to obtain the mapping model;

[0077] Transmitting position estimation module: Transmitting position estimation is achieved by combining target motion information and energy information, or multi-target collaborative information to estimate the longitude and latitude of the transmitting position to obtain the longitude and latitude of the transmitting position;

[0078] Type recognition module: Using the longitude and latitude of the emission position and combining it with the established mapping model to sample the global type matching template, the type vector is obtained to achieve the final infrared target type recognition.

[0079] In the above device, the specific implementation of the mapping model pre-establishment module includes the following steps:

[0080] Create a mapping between real launch location and type list:

[0081] Step 1.1: Create a mapping between the actual launch location and the type list based on the actual launch information:

[0082] f1: L1→M1

[0083] Where f1 represents a function that maps elements in set L1 to elements in set M1, (rx, ry)∈L1 represents the longitude and latitude of the actual launch location, and M1 is a set of all type lists contained in the actual launch location;

[0084] Create a virtual launch location-type list mapping:

[0085] Step 1.2: Combine historical target launch data to establish a virtual launch position-type list mapping, and calculate the launch time t of the i-th data of each type i :

[0086] t i =inf{t i |E i (t) = Fitting(T i , E i )<θ1}

[0087] Among them, T i 、E i Respectively represent the time variation information and energy variation information of the i-th data, Fitting(T i , E i ) indicates T i The horizontal axis E i Fitting is done for the vertical axis, inf represents the first value that meets the requirements, t i According to T i and E i The fitting relationship predicts the time forward until the energy is less than the threshold θ1;

[0088] Step 1.3: Calculate the longitude cx of the location where the i-th data of each type is transmitted i and latitude cy i :

[0089] cx i ={λ i (t i )|λ i (t) = Fitting(T i ,λ i )}

[0090]

[0091] Among them, λ i 、 Respectively represent the longitude change information and latitude change information of the i-th data;

[0092] Step 1.4: Calculate the outlier Z of each type of emission position using the following formula. If Z is less than the threshold θ2, proceed to step 1.5. If Z is greater than the threshold θ2, proceed to step 1.6.

[0093]

[0094] Among them, Z represents the type of emission position outliers, n represents the total number of data contained in the type, and The types respectively indicate the average longitude and latitude of the launch location containing all data. Indicates calculation (cx i ,cy i )arrive distance, Indicates that all data points to The average distance;

[0095] Step 1.5: Using the geometric center of all data as the center point, establish the first batch of virtual emission position-type list mappings:

[0096] f2: L2→M2

[0097] Among them, f2 represents the function that maps the elements in the set L2 to the elements in the set M2, (vx G ,vy G )∈L2 represent the longitude and latitude of the virtual launch position with the geometric center of all data, and M2 is the set of all type lists contained in the first batch of virtual launch positions;

[0098] Step 1.6: Clustering to obtain the second batch of virtual emission position-type list mappings:

[0099] f3: L3→M3

[0100] Among them, f3 represents the function that maps the elements in the set L3 to the elements in the set M3, (vx C ,vy C )∈L3 represent the longitude and latitude of the virtual launch position obtained by clustering, M3 is the set of all type lists contained in the second batch of virtual launch positions;

[0101] Step 1.7: Gather the first batch of virtual launch position-type list mappings obtained in step 1.5 and the second batch of virtual launch position-type list mappings obtained in step 1.6 to establish a virtual launch position-type list mapping:

[0102] f4: L4→→→M4

[0103] Wherein, f4 represents a function that maps elements in set L4 to elements in set M4, (vx, vy)∈L4=L2∪L3 represent the longitude and latitude of the virtual launch position respectively, and M4=M2∪M3 is the set of all type lists contained in the virtual launch position;

[0104] Create a mapping between regional geographic information and launch location list:

[0105] Step 1.8: Combine the actual launch location longitude and latitude, the virtual launch location longitude and latitude obtained in step 1.7, and the regional geographic information to create a mapping between regional geographic information and the launch location list. Divide the regional territory into a number of rectangles such that the union of the rectangles covers the regional territory and minimizes the total area.

[0106]

[0107] Among them, Area k represents the latitude and longitude range covered by region k, m represents the number of rectangles divided into region k, R jIndicates the latitude and longitude range covered by the jth rectangle, It means that m rectangular areas R j merged together to form a larger area;

[0108] Step 1.9: Calculate the minimum distance min between the virtual launch location longitude and latitude (vx, vy) and the latitude and longitude of all real launch locations in the region k to which it belongs d (vx,vy):

[0109]

[0110] Among them, (rx,ry)∈Area k They respectively indicate that region k contains the longitude and latitude of all real launch locations. The region to which the real launch location belongs and the region to which the virtual launch location belongs are directly determined by the data;

[0111] Step 1.10: Combine all real launch positions and virtual launch positions in the area, and use the minimum distance min obtained in step 1.9 d (vx,vy) is compared with the threshold θ3. If min d (vx, vy) is less than the threshold θ3, then the virtual launch position is considered to coincide with the real launch position, and the longitude and latitude of the virtual launch position are not retained; if min d If (vx, vy) is greater than the threshold θ3, the virtual launch position is considered to be a new launch position different from the real launch position, and the longitude and latitude of the virtual launch position are retained. The formula is as follows:

[0112] L k =(L1∈Area k )∪((vx,vy)∈L4∈Area k and min d (vx, vy)>θ3)

[0113] Among them, (x k ,y k )∈L k Respectively represent the longitude and latitude of all launch locations contained in area k, L k Contains all real launch positions and virtual launch positions that meet the conditions in area k;

[0114] Step 1.11: Create a mapping between regional geographic information and the launch location list:

[0115] f5: A→L

[0116] Among them, f5 represents the function that maps the elements in set A to the elements in set L. Represents all regional geographic information, represents the set of all launch locations, p is the total number of regions, and (x, y)∈L represents the longitude and latitude of the launch location respectively;

[0117] Create a mapping between regional geographic information, launch location list, and type list:

[0118] Step 1.12: Combine the real launch location-type list mapping obtained in step 1.1, the virtual launch location-type list mapping obtained in step 1.7, and the regional geographic information-launch location list mapping obtained in step 1.11 to ultimately establish a regional geographic information-launch location list-type list mapping;

[0119] If, in step 1.10, a virtual launch position coincides with a real launch position, then extract the types not present in the real launch position from the target type list of the virtual launch position and add these types to the real launch position-type list mapping established in step 1.1. Combine the supplemented real launch position-type list mapping with the virtual launch position-type list mapping obtained in step 1.7, and map all types to the set L of all launch positions obtained in step 1.11 to establish a launch position-type list mapping:

[0120] f6: L→M

[0121] Where f6 represents a function that maps elements in set L to elements in set M, where M=M1∪M4 is the set of all type lists;

[0122] Step 1.13: Combine the regional geographic information-transmission location list mapping obtained in step 1.11 with the transmission location-type list mapping obtained in step 1.12 to create a regional geographic information-transmission location list-type list mapping:

[0123] f7: A→L→M

[0124] Among them, f7 is a composite mapping function from set A to set L, and then from set L to set M.

[0125] In the above device, the implementation of the transmission position estimation module includes the following steps:

[0126] Step 2.1: Calculate the launch time t of target o o :

[0127] t o =inf{t o |E o (t) = Fitting(T o , E o )<θ1}where, T o 、E oRepresent the time variation information and energy variation information of target o respectively, Fitting(T o , E o ) indicates T o The horizontal axis E o Fitting is done for the vertical axis, inf represents the first value that meets the requirements, t o According to T o and E o The fitting relationship predicts the time forward until the energy is less than the threshold θ1;

[0128] Step 2.2: Calculate the longitude x of the launch point of target o o and latitude y o :

[0129] x o ={λ o (t o )|λ o (t) = Fitting(T o ,λ o ))

[0130] Among them, λ o 、 They respectively represent the longitude change information and latitude change information of the target o.

[0131] In the above device, the implementation of the type identification module includes the following steps:

[0132] Step 3.1: Based on all the regional geographic information A obtained in step 1.11, obtain the longitude and latitude of the launch point obtained in step 2.2 (x o ,y o ) belongs to the area list A((x o ,y o )):

[0133]

[0134] Step 3.2: Based on the regional geographic information-emission location list-type list mapping established in step 1.13, obtain (x o ,y o ) corresponds to the launch position list L((x o ,y o )):

[0135] L((x o ,y o ))=f7(A((x o ,y o )))→Lwhere, (ox,oy)∈L((x o,y o )) respectively represent (x o ,y o ) The area to which it belongs corresponds to the longitude and latitude of the transmitting location;

[0136] Step 3.3: Calculate (x o ,y o ) and the corresponding launch position list L((x o ,y o )) The distance of each launch position in the sequence is sorted from smallest to largest, and the sorted launch position list L is obtained. S ((x o ,y o )):

[0137] L S ((x o ,y o ))=Sort(((kx, ky)|(kx, ky)∈L((x o ,y o )), d((x o ,y o ), (kx, ky)))

[0138] Among them, (sx,sy)∈L S ((x o ,y o )) respectively represent the L((x o ,y o )) After sorting, the longitude and latitude of the launch position;

[0139] Step 3.4: According to the sorted launch position list L S ((x o ,y o ), combined with the launch location list-type list mapping in the regional geographic information-launch location list-type list mapping established in step 1.13, we get (x o ,y o ) corresponds to the type vector M((x o ,y o )):

[0140] M((x o ,y o ))=f7(L S ((x o ,y o )))→M

[0141] M((x o ,y o )) contains (x o ,y o) All types contained in all emission positions after sorting the area to which they belong, to achieve global type matching template sampling;

[0142] Step 3.5: Based on the type vector obtained in step 3.4, use the type matching algorithm from top to bottom to calculate the target o and the type vector M((x o ,y o )) and select the type Model(o) with the highest similarity as the type of target o to achieve infrared target type recognition:

[0143]

[0144] Among them, S(o,mdl) represents the type matching algorithm used to calculate the target o and M((x o ,y o )) similarity of type mdl.

[0145] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0146] 1. The core function of the present invention is to sample the type matching library, which greatly improves the recognition speed while ensuring a high target recognition rate.

[0147] 2. The emission position guided sampling matching library method proposed in the present invention has a wide range of applications and can be widely embedded in various infrared aerial target type recognition tasks as a pre-processing for recognition.

[0148] 3. The mapping model establishment method and emission point estimation strategy are comprehensively designed to cover a variety of special cases, giving it strong adaptability when facing various complex scenarios.

[0149] 4. The mapping model establishment method fully considers the missing of real information and historical data, efficiently integrates data, and constructs a complete mapping relationship.

[0150] 5. The launch point estimation strategy addresses the problems of difficult detection and incomplete data in the early stages of target launch. It uses existing information to accurately estimate the longitude and latitude of the target launch point, laying the foundation for accurate sampling of the subsequent type matching library. BRIEF DESCRIPTION OF THE DRAWINGS

[0151] Figure 1 It is the overall flow chart of the present invention;

[0152] Figure 2 This is a schematic diagram of the identification of the present invention. DETAILED DESCRIPTION

[0153] The present invention will be further described in detail below in conjunction with test examples and specific embodiments. However, this should not be understood as limiting the scope of the present invention to the following embodiments, and all technologies implemented based on the present invention fall within the scope of the present invention.

[0154] The present invention provides an infrared target type recognition method and device based on emission position guided sampling matching. The technical problem to be solved is how to sample some templates to be matched from a large number of target types, thereby achieving a high target recognition rate and high speed infrared target type recognition method and device based on emission position guided sampling matching library. The entire algorithm design structure framework is as follows Figure 1 As shown, the steps include:

[0155] Create a real launch location-type list mapping.

[0156] Step 1.1: Create a mapping between the actual launch location and the type list based on the actual launch information:

[0157] f1: L1→M1

[0158] Among them, (rx, ry)∈L1 represents the longitude and latitude of the actual transmission location respectively, and M1 is the set of all type lists contained in the actual transmission location.

[0159] Create a virtual launch location-type list mapping.

[0160] Step 1.2: Combined with historical target launch data, a virtual launch position-type list mapping can be established. Calculate the launch time t of the i-th data of each type i :

[0161] t i =inf{t i |E i (t) = Fitting(T i , E i )<θ1}

[0162] Among them, T i 、E i Respectively represent the time variation information and energy variation information of the i-th data, Fitting(T i , E i ) indicates T i The horizontal axis E i Fitting is done for the vertical axis, inf represents the first value that meets the requirements, t i According to T i and E i The fitting relationship predicts the time forward until the energy is less than the threshold θ1. In this example, a quadratic polynomial fitting is used, θ1 = 1000;

[0163] Step 1.3: Calculate the longitude cx of the location where the i-th data of each type is transmitted i and latitude cy i :

[0164] cx i ={λ i (t i )|λ i (t) = Fitting(T i ,λ i )}

[0165]

[0166] Among them, λ i 、 Respectively represent the longitude change information and latitude change information of the i-th data;

[0167] Step 1.4: Calculate the outlier Z of each type of emission position using the following formula. If Z is less than the threshold θ2, proceed to step 1.5. If Z is greater than the threshold θ2, proceed to step 1.6.

[0168]

[0169] Among them, Z represents the type of emission position outliers, n represents the total number of data contained in the type, and The types respectively indicate the average longitude and latitude of the launch location containing all data. Indicates calculation (cx i ,cy i )arrive distance, Indicates that all data points to In this example, the Haversine formula is used to calculate the distance between two points on the Earth's surface, with θ2 = 0.5;

[0170] Step 1.5: Using the geometric center of all data as the center point, establish the first batch of virtual emission position-type list mappings:

[0171] f2: L2→M2

[0172] Among them, (vx G ,vy G )∈L2 represent the longitude and latitude of the virtual launch position with the geometric center of all data, and M2 is the set of all type lists contained in the first batch of virtual launch positions;

[0173] Step 1.6: Clustering to obtain the second batch of virtual emission position-type list mappings:

[0174] f3: L3→M3

[0175] Among them, (vx C ,vy C )∈L3 represent the longitude and latitude of the virtual launch location obtained by clustering, and M3 is the set of all types of lists contained in the second batch of virtual launch locations. In this example, DBSCAN clustering is used;

[0176] Step 1.7: Gather the first batch of virtual launch position-type list mappings obtained in step 1.5 and the second batch of virtual launch position-type list mappings obtained in step 1.6 to establish a virtual launch position-type list mapping:

[0177] f4: L4→M4

[0178] Among them, (vx, vy)∈L4=L2∪L3 respectively represent the longitude and latitude of the virtual launch position, and M4=M2∪M3 is the set of all type lists contained in the virtual launch position.

[0179] Establish regional geographic information - launch location list mapping.

[0180] Step 1.8: Combine the actual launch location longitude and latitude, the virtual launch location longitude and latitude obtained in step 1.7, and the regional geographic information to create a mapping between regional geographic information and the launch location list. Divide the regional territory into a number of rectangles such that the union of the rectangles covers the regional territory and minimizes the total area:

[0181]

[0182] Among them, Area k represents the latitude and longitude range covered by region k, m represents the number of rectangles divided into region k, R j Indicates the latitude and longitude range covered by the jth rectangle;

[0183] Step 1.9: Calculate the minimum distance min between the virtual launch location longitude and latitude (vx, vy) and the latitude and longitude of all real launch locations in the region k to which it belongs d (vx, vy):

[0184]

[0185] Among them, (rx,ry)∈Area k The longitude and latitude of region k respectively indicate that the region k contains all real launch locations. The region to which the real launch location belongs and the region to which the virtual launch location belongs can be directly determined from the data. In this example, the Haversine formula is used to calculate the distance between two points on the earth's surface;

[0186] Step 1.10: Combine all the real launch positions and virtual launch positions in the region by the following formula. d (vx,vy) is compared with the threshold θ3. If min d (vx, vy) is less than the threshold θ3, then the virtual launch position is considered to coincide with the real launch position, and the longitude and latitude of the virtual launch position are not retained; if min d If (vx, vy) is greater than the threshold θ3, the virtual launch position is considered to be a new launch position different from the real launch position, and the longitude and latitude of the virtual launch position are retained.

[0187] L k =(L1∈Area k )∪((vx,vy)∈L4∈Area k and min d (vx, vy)>θ3)

[0188] Among them, (x k ,y k )∈L k Respectively represent the longitude and latitude of all emission locations contained in area k. In this example, θ3 = 100;

[0189] Step 1.11: Create a mapping between regional geographic information and the launch location list:

[0190] f5: A-→L

[0191] in, Represents all regional geographic information, represents the set of all launch locations, p is the total number of regions, and (x, y)∈L represents the longitude and latitude of the launch location respectively.

[0192] Establish regional geographic information-transmission location list-type list mapping.

[0193] Step 1.12: Combine the real launch location-type list mapping obtained in step 1.1, the virtual launch location-type list mapping obtained in step 1.7, and the regional geographic information-launch location list mapping obtained in step 1.11 to ultimately establish a regional geographic information-launch location list-type list mapping. If there is a virtual launch location that overlaps with a real launch location in step 1.10, extract the other types contained in the virtual launch location that are different from the types contained in the overlapping real launch location to supplement the real launch location-type list mapping established in step 1.1. Combine the supplemented real launch location-type list mapping with the virtual launch location-type list mapping obtained in step 1.7, and correspond all types to the set L of all launch locations obtained in step 1.11 to establish a launch location-type list mapping:

[0194] f6: L→M

[0195] Among them, M=M1∪M4 is the set of all type lists;

[0196] Step 1.13: Combine the regional geographic information-transmission location list mapping obtained in step 1.11 with the transmission location-type list mapping obtained in step 1.12 to create a regional geographic information-transmission location list-type list mapping:

[0197] f7: A→L→M

[0198] Step 2.1: Calculate the launch time t of target o o :

[0199] t o =inf{t o |E o (t) = Fitting(T o , E o )<θ1}

[0200] Among them, T o 、E o Represent the time variation information and energy variation information of target o respectively, Fitting(T o , E o ) indicates T o The horizontal axis E o Fitting is done for the vertical axis, inf represents the first value that meets the requirements, t o According to T o and E o The fitting relationship predicts the time forward until the energy is less than the threshold θ1. In this example, a quadratic polynomial fitting is used, θ1 = 1000;

[0201] Step 2.2: Calculate the longitude x of the launch point of target o o and latitude y o :

[0202] x o ={λ o (t o )|λ o (t) = Fitting(T o ,λ o )}

[0203]

[0204] Among them, λ o 、 They respectively represent the longitude change information and latitude change information of the target o.

[0205] Step 3.1: Based on all the regional geographic information A obtained in step 1.11, obtain the longitude and latitude of the launch point obtained in step 2.2 (x o ,y o ) belongs to the area list A((x o ,y o )):

[0206]

[0207] Because the longitude and latitude ranges of the regions are divided into blocks, there may be overlaps between the ranges, resulting in the target launch point belonging to more than one region. Therefore, a list is used to store the regions to which the longitude and latitude of the launch point belong.

[0208] Step 3.2: Based on the regional geographic information-emission location list-type list mapping established in step 1.13, obtain (x o ,y o ) corresponds to the launch position list L((x o ,y o )):

[0209] L((x o ,y o ))=f7(A((x o ,y o )))→L

[0210] Among them, (ox,oy)∈L((x o ,y o )) respectively represent (x o ,y o ) The area to which it belongs corresponds to the longitude and latitude of the transmitting location;

[0211] Step 3.3: Calculate (x o ,y o ) and the corresponding launch position list L((x o ,y o )) The distance of each launch position in the sequence is sorted from smallest to largest, and the sorted launch position list L is obtained. S ((x o ,y o )):

[0212] L S ((x o ,y o ))=Sort({(kx,ky)|(kx,ky)∈L((x o ,y o ))},d((x o ,y o ), (kx,ky)))

[0213] Among them, (sx,sy)∈L S ((x o ,y o )) respectively represent the L((x o ,y o )) After sorting, the longitude and latitude of the location are transmitted. In this example, the Haversine formula is used to calculate the distance between two points on the earth's surface;

[0214] Step 3.4: According to the sorted launch position list L S ((x o ,y o ), combined with the launch location-type list mapping in the regional geographic information-launch location list-type list mapping established in step 1.13, we get (x o ,y o ) corresponds to the type vector M((x o ,y o )):

[0215] M((x o ,y o ))=f7(L S ((x o ,y o )))→M

[0216] M((x o ,y o )) contains (x o ,y o ) All types contained in all emission positions after sorting the area to which they belong, to achieve global type matching template sampling;

[0217] Step 3.5: Based on the type vector obtained in step 3.4, use the type matching algorithm from top to bottom to calculate the target o and the type vector M((x o ,y o )) and select the type Model(o) with the highest similarity as the type of target o to achieve infrared target type recognition:

[0218]

[0219] Among them, S(o,mdl) represents the type matching algorithm used to calculate the target o and M((x o ,y o )) in the type mdl. In this example, the type matching uses the dynamic time warping algorithm.

[0220] In summary, the present invention has the following characteristics:

[0221] 1. By establishing a mapping relationship between emission locations and possible target types, and combining target motion and energy information to estimate emission locations, this approach addresses the problem of a large number of target types across the entire domain, which wastes computing resources by matching each type individually. This approach significantly reduces the complexity of the matching templates by reducing the number of types to be identified, thereby improving the accuracy and speed of target type recognition.

[0222] 2. By constructing mappings between real launch locations and target types, and virtual launch locations and target types, and incorporating regional geographic information, we address the issues of missing historical data or incomplete data from the initial launch of targets. This approach integrates real and historical data to create a comprehensive mapping, ensuring efficient target identification even in complex scenarios.

[0223] 3. By clustering and calculating the geometric center, a mapping between virtual launch positions and target type lists is established, addressing the issue of uneven target launch position distribution or a high number of outliers. This approach optimizes the distribution of virtual launch positions through clustering and calculating the geometric center, ensuring the robustness and adaptability of the mapping model and further improving the accuracy of target type recognition.

[0224] 4. By combining regional geographic information with a mapping of emission location lists and type lists, the issue of unclear target emission point regions is resolved. This approach divides the region into rectangles and, by combining the relationship between emission locations and regions, accurately determines the target emission point region, thus enabling precise sampling of global type matching templates.

[0225] 5. By combining type matching algorithms such as the Dynamic Time Warping algorithm with a sorted list of emission locations, the complex similarity calculation process in target type identification is resolved. This approach, by sorting emission locations by distance and calculating similarity sequentially, ensures that identification accuracy is maintained while significantly improving recognition speed, making it suitable for all types of infrared aerial target identification tasks.

Claims

1. A method for infrared target type recognition based on emission position guided sampling matching, characterized in that: The steps include: Step 1: Pre-establish a mapping model. Pre-establish the mapping relationship between launch locations and possible target types. Specifically, this includes mapping between real launch locations and type lists, mapping between virtual launch locations and type lists, and mapping between regional geographic information and launch location lists. Finally, establish a mapping between regional geographic information and launch location lists and type lists to obtain a mapping model and proceed to step 2. Step 2: Estimation of the launch position, which is achieved by combining target motion information and energy information, or multi-target collaborative information to estimate the launch position latitude and longitude, obtain the launch position latitude and longitude, and proceed to step 3; Step 3: Using the longitude and latitude of the emission position obtained in step 2 and the mapping model established in step 1, sample the global type matching template to obtain the type vector to achieve the final infrared target type recognition; The step 3 comprises the following steps: Step 3.1: Based on all regional geographic information A, obtain the latitude and longitude of the launch point (x o ,y o ) belongs to the area list A((x o ,y o )): p is the total number of areas, Area k Indicates the latitude and longitude range covered by region k; Step 3.2: According to the regional geographic information-transmission location list-type list mapping, obtain (x o ,y o ) corresponds to the launch position list L((x o ,y o )): L((x o ,and o ))=f7(A((x o ,and o )))→L (ox,oy)∈L((x o ,y o )) respectively represent (x o ,y o ) The area to which it belongs corresponds to the longitude and latitude of the transmitting location; f7 is a composite mapping function from set A to set L, and then from set L to set M; Step 3.3: Calculate (x o ,y o ) and the corresponding launch position list L((x o ,y o )) The distance of each launch position (ox,oy) in )) is sorted from small to large according to the distance, and the sorted launch position list L is obtained S ((x o ,y o )): L S ((x o ,y o ))=Sort((kx,ky)|(kx,ky)∈L((x o ,y o )),d((x o ,y o ),(kx,ky))) (sx,sy)∈L S ((x o ,y o )) respectively represent the L((x o ,y o )) After sorting, the longitude and latitude of the launch position; Step 3.4: According to the sorted launch position list L S ((x o ,y o ), combined with the launch location list-type list mapping in the regional geographic information-launch location list-type list mapping, we get (x o ,y o ) corresponds to the type vector M((x o ,y o )): M((x o ,and o ))=f7(L S ((x o ,and o )))→M M((x o ,y o )) contains (x o ,y o ) After sorting the area to which it belongs, all types contained in all emission positions are used to achieve global type matching template sampling, where M is a set; Step 3.5: Based on the type vector obtained in step 3.4, use the type matching algorithm from top to bottom to calculate the target o and the type vector M((x o ,y o )) and select the type Model(o) with the highest similarity as the type of target o to achieve infrared target type recognition: Among them, S(o,mdl) represents the type matching algorithm used to calculate the target o and M((x o ,y o )) in the type mdl similarity.

2. The infrared target type recognition method based on emission position guided sampling matching according to claim 1 is characterized in that: The step 1 comprises the following steps: Create a mapping between real emission location and type list: Step 1.1: Create a mapping between the actual launch location and the type list based on the actual launch information: f1:L1→M1 Where f1 represents the function that maps the elements in the set L1 to the elements in the set M1, (rx, ry) ∈ L1 represents the longitude and latitude of the actual launch location, and M1 is the set of all type lists contained in the actual launch location; Create a virtual launch location-type list mapping: Step 1.2: Combine historical target launch data to establish a virtual launch position-type list mapping, and calculate the launch time t of the i-th data of each type i : t i =inf{t i |E i (t)=Fitting(T i ,E i )<θ1} Among them, T i 、E i Respectively represent the time variation information and energy variation information of the i-th data, Fitting(T i ,E i ) indicates T i The horizontal axis E i Fitting is done for the vertical axis, inf represents the first value that meets the requirements, t i According to T i and E i The fitting relationship predicts the time forward until the energy is less than the threshold θ1; Step 1.3: Calculate the longitude cx of the location where the i-th data of each type is transmitted i and latitude cy i : cx i ={λ i (t i )|λ i (t)=Fitting(T i ,λ i )} Among them, λ i 、 Respectively represent the longitude change information and latitude change information of the i-th data; Step 1.4: Calculate the outlier Z of each type of emission position using the following formula. If Z is less than the threshold θ2, proceed to step 1.

5. If Z is greater than the threshold θ2, proceed to step 1.

6. Among them, Z represents the type of emission position outlier, n represents the total number of data contained in the type, and The types respectively indicate the average longitude and latitude of the launch location containing all data. Indicates calculation (cx i ,cy i )arrive distance, Indicates that all data points to The average distance; Step 1.5: Using the geometric center of all data as the center point, establish the first batch of virtual emission position-type list mappings: f2:L2→M2 Among them, f2 represents the function that maps the elements in the set L2 to the elements in the set M2, (vx G ,vy G )∈L2 represent the longitude and latitude of the virtual launch position with the geometric center of all data, and M2 is the set of all type lists contained in the first batch of virtual launch positions; Step 1.6: Clustering to obtain the second batch of virtual emission position-type list mappings: f3:L3→M3 Among them, f3 represents the function that maps the elements in the set L3 to the elements in the set M3, (vx C ,vy C )∈L3 represent the longitude and latitude of the virtual launch position obtained by clustering, M3 is the set of all type lists contained in the second batch of virtual launch positions; Step 1.7: Gather the first batch of virtual launch position-type list mappings obtained in step 1.5 and the second batch of virtual launch position-type list mappings obtained in step 1.6 to establish a virtual launch position-type list mapping: f4:L4→M4 Where f4 represents a function that maps elements in set L4 to elements in set M4, (vx, vy)∈L4=L2∪L3 represent the longitude and latitude of the virtual launch location, and M4=M2∪M3 is the set of all type lists contained in the virtual launch location; Create a mapping between regional geographic information and launch location list: Step 1.8: Combine the actual launch location longitude and latitude, the virtual launch location longitude and latitude obtained in step 1.7, and the regional geographic information to create a mapping between regional geographic information and the launch location list. Divide the regional territory into a number of rectangles such that the union of the rectangles covers the regional territory and minimizes the total area. Among them, Area k represents the latitude and longitude range covered by region k, m represents the number of rectangles divided into region k, R j Indicates the latitude and longitude range covered by the jth rectangle, It means that m rectangular areas R j merged together to form a larger area; Step 1.9: Calculate the minimum distance min between the virtual launch location longitude and latitude (vx, vy) and the latitude and longitude of all real launch locations in the region k to which it belongs d (vx,vy): Where, (rx,ry)∈Area k They respectively indicate that region k contains the longitude and latitude of all real launch locations. The region to which the real launch location belongs and the region to which the virtual launch location belongs are directly determined by the data; Step 1.10: Combine all real launch positions and virtual launch positions in the area, and use the minimum distance min obtained in step 1.9 d (vx,vy) is compared with the threshold θ3. If min d (vx,vy) is less than the threshold θ3, then the virtual launch position is considered to coincide with the real launch position, and the longitude and latitude of the virtual launch position are not retained; if min d If (vx,vy) is greater than the threshold θ3, the virtual launch position is considered to be a new launch position different from the real launch position, and the latitude and longitude of the virtual launch position are retained. The formula is as follows: L k =(L1∈Area k )∪((vx,vy)∈L4∈Area k and min d (vx,vy)>θ3) Among them, (x k ,y k )∈L k Respectively represent the longitude and latitude of all launch locations contained in area k, L k Contains all real launch positions and virtual launch positions that meet the conditions in area k; Step 1.11: Create a mapping between regional geographic information and the launch location list: f5:A→L Among them, f5 represents the function that maps the elements in set A to the elements in set L. Represents all regional geographic information, represents the set of all launch locations, p is the total number of regions, and (x, y)∈L represents the longitude and latitude of the launch location respectively; Create a mapping between regional geographic information, launch location list, and type list: Step 1.12: Combine the real launch location-type list mapping obtained in step 1.1, the virtual launch location-type list mapping obtained in step 1.7, and the regional geographic information-launch location list mapping obtained in step 1.11 to ultimately establish a regional geographic information-launch location list-type list mapping; If, in step 1.10, a virtual launch position coincides with a real launch position, then extract the types not present in the real launch position from the target type list of the virtual launch position and add these types to the real launch position-type list mapping established in step 1.

1. Combine the supplemented real launch position-type list mapping with the virtual launch position-type list mapping obtained in step 1.7, and map all types to the set L of all launch positions obtained in step 1.11 to establish a launch position-type list mapping: f6:L→M Where f6 represents a function that maps elements in set L to elements in set M, where M=M1∪M4 is the set of all type lists; Step 1.13: Combine the regional geographic information-transmission location list mapping obtained in step 1.11 with the transmission location-type list mapping obtained in step 1.12 to create a regional geographic information-transmission location list-type list mapping: f7:A→L→M Among them, f7 is a composite mapping function from set A to set L, and then from set L to set M.

3. The infrared target type recognition method based on emission position guided sampling matching according to claim 2 is characterized in that: The step 2 comprises the following steps: Step 2.1: Calculate the launch time t of target o o : t o =inf{t o |E o (t) = Fitting(T o ,E o )<θ1}where, T o 、E o Represent the time variation information and energy variation information of target o respectively, Fitting(T o ,E o ) indicates T o The horizontal axis E o Fitting is done for the vertical axis, inf represents the first value that meets the requirements, t o According to T o and E o The fitting relationship predicts the time forward until the energy is less than the threshold θ1; Step 2.2: Calculate the longitude x of the launch point of target o o and latitude y o : x o ={λ o (t o )|λ o (t)=Fitting(T o ,λ o )} Among them, λ o 、 They respectively represent the longitude change information and latitude change information of the target o.

4. An infrared target type recognition device guided by emission position sampling matching, characterized in that: The steps include: Mapping model pre-establishment module: Pre-establish the mapping model, pre-establish the mapping relationship between the launch position and the possible target type, specifically including the real launch position-type list mapping, the virtual launch position-type list mapping, the regional geographic information-launch position list mapping, and finally establish the regional geographic information-launch position list-type list mapping to obtain the mapping model; Transmitting position estimation module: Transmitting position estimation is achieved by combining target motion information and energy information, or multi-target collaborative information to estimate the longitude and latitude of the transmitting position to obtain the longitude and latitude of the transmitting position; Type recognition module: uses the longitude and latitude of the emission location and combines it with the established mapping model to sample the global type matching template to obtain the type vector and realize the final infrared target type recognition; The implementation of the type identification module includes the following steps: Step 3.1: Based on all regional geographic information A, obtain the latitude and longitude of the launch point (x o ,y o ) belongs to the area list A((x o ,y o )): p is the total number of areas, Area k Indicates the latitude and longitude range covered by region k; Step 3.2: According to the regional geographic information-transmission location list-type list mapping, obtain (x o ,y o ) corresponds to the launch position list L((x o ,y o )): L((x o ,and o ))=f7(A((x o ,and o )))→L (ox,oy)∈L((x o ,y o )) respectively represent (x o ,y o ) The area to which it belongs corresponds to the longitude and latitude of the transmitting location; f7 is a composite mapping function from set A to set L, and then from set L to set M; Step 3.3: Calculate (x o ,y o ) and the corresponding launch position list L((x o ,y o )) The distance of each launch position in the sequence is sorted from smallest to largest, and the sorted launch position list L is obtained. S ((x o ,y o )): L s ((x o ,y o ))=Sort((kx,ky)|(kx,ky)∈L((x o ,y o )),d((x o ,y o ),(kx,ky))) (sx,sy)∈L S ((x o ,y o )) respectively represent the L((x o ,y o )) After sorting, the longitude and latitude of the launch position; Step 3.4: According to the sorted launch position list L S ((x o ,y o ), combined with the launch location list-type list mapping in the regional geographic information-launch location list-type list mapping, we get (x o ,y o ) corresponds to the type vector M((x o ,y o )): M((x o ,and o ))=f7(L S ((x o ,and o )))→M M((x o ,y o )) contains (x o ,y o ) After sorting the area to which it belongs, all types contained in all emission positions are used to achieve global type matching template sampling, where M is a set; Step 3.5: Based on the type vector obtained in step 3.4, use the type matching algorithm from top to bottom to calculate the target o and the type vector M((x o ,y o )) and select the type Model(o) with the highest similarity as the type of target o to achieve infrared target type recognition: Among them, S(o,mdl) represents the type matching algorithm used to calculate the target o and M((x o ,y o )) in the type mdl similarity.

5. The infrared target type recognition device guided by emission position sampling matching according to claim 4, characterized in that: The specific implementation of the mapping model pre-establishment module includes the following steps: Create a mapping between real emission location and type list: Step 1.1: Create a mapping between the actual launch location and the type list based on the actual launch information: f1:L1→M1 Where f1 represents the function that maps the elements in the set L1 to the elements in the set M1, (rx, ry) ∈ L1 represents the longitude and latitude of the actual launch location, and M1 is the set of all type lists contained in the actual launch location; Create a virtual launch location-type list mapping: Step 1.2: Combine historical target launch data to establish a virtual launch position-type list mapping, and calculate the launch time t of the i-th data of each type i : t i =inf{t i |E i (t)=Fitting(T i ,E i )<θ1} Among them, T i 、E i Respectively represent the time variation information and energy variation information of the i-th data, Fitting(T i ,E i ) indicates T i The horizontal axis E i Fitting is done for the vertical axis, inf represents the first value that meets the requirements, t i According to T i and E i The fitting relationship predicts the time forward until the energy is less than the threshold θ1; Step 1.3: Calculate the longitude cx of the location where the i-th data of each type is transmitted i and latitude cy i : cx i ={λ i (t i )|λ i (t)=Fitting(T i ,λ i )} Among them, λ i 、 Respectively represent the longitude change information and latitude change information of the i-th data; Step 1.4: Calculate the outlier Z of each type of emission position using the following formula. If Z is less than the threshold θ2, proceed to step 1.

5. If Z is greater than the threshold θ2, proceed to step 1.

6. Among them, Z represents the type of emission position outlier, n represents the total number of data contained in the type, and The types respectively indicate the average longitude and latitude of the launch location containing all data. Indicates calculation (cx i ,cy i )arrive distance, Indicates that all data points to The average distance Step 1.5: Using the geometric center of all data as the center point, establish the first batch of virtual emission position-type list mappings: f2:L2→M2 Among them, f2 represents the function that maps the elements in the set L2 to the elements in the set M2, (vx G ,vy G )∈L2 respectively represent the longitude and latitude of the virtual launch position with the geometric center of all data as the virtual launch position, and M2 is the set of all type lists contained in the first batch of virtual launch positions; Step 1.6: Clustering to obtain the second batch of virtual emission position-type list mappings: f3:L3→M3 Among them, f3 represents the function that maps the elements in the set L3 to the elements in the set M3, (vx C ,vy C )∈L3 represent the longitude and latitude of the virtual launch position obtained by clustering, M3 is the set of all type lists contained in the second batch of virtual launch positions; Step 1.7: Gather the first batch of virtual launch position-type list mappings obtained in step 1.5 and the second batch of virtual launch position-type list mappings obtained in step 1.6 to establish a virtual launch position-type list mapping: f4:L4→M4 Where f4 represents a function that maps elements in set L4 to elements in set M4, (vx, vy)∈L4=L2∪L3 represent the longitude and latitude of the virtual launch location, and M4=M2∪M3 is the set of all type lists contained in the virtual launch location; Create a mapping between regional geographic information and launch location list: Step 1.8: Combine the actual launch location longitude and latitude, the virtual launch location longitude and latitude obtained in step 1.7, and the regional geographic information to create a mapping between regional geographic information and the launch location list. Divide the regional territory into a number of rectangles such that the union of the rectangles covers the regional territory and minimizes the total area. Among them, Area k represents the latitude and longitude range covered by region k, m represents the number of rectangles divided into region k, R j Indicates the latitude and longitude range covered by the jth rectangle, It means that m rectangular areas R j merged together to form a larger area; Step 1.9: Calculate the minimum distance min between the virtual launch location longitude and latitude (vx, vy) and the latitude and longitude of all real launch locations in the region k to which it belongs d (vx,vy): Where, (rx,ry)∈Area k They respectively indicate that region k contains the longitude and latitude of all real launch locations. The region to which the real launch location belongs and the region to which the virtual launch location belongs are directly determined by the data; Step 1.10: Combine all real launch positions and virtual launch positions in the area, and use the minimum distance min obtained in step 1.9 d (vx,vy) is compared with the threshold θ3. If min d (vx,vy) is less than the threshold θ3, then the virtual launch position is considered to coincide with the real launch position, and the longitude and latitude of the virtual launch position are not retained; if min d If (vx,vy) is greater than the threshold θ3, the virtual launch position is considered to be a new launch position different from the real launch position, and the latitude and longitude of the virtual launch position are retained. The formula is as follows: L k =(L1∈Area k )∪((vx,vy)∈L4∈Area k andmin d (vx,vy)>θ3) Among them, (x k ,y k )∈L k Respectively represent the longitude and latitude of all launch locations contained in area k, L k Contains all real launch positions and virtual launch positions that meet the conditions in area k; Step 1.11: Create a mapping between regional geographic information and the launch location list: f5:a→L Among them, f5 represents the function that maps the elements in set A to the elements in set L. Represents all regional geographic information, represents the set of all launch locations, p is the total number of regions, and (x, y)∈L represents the longitude and latitude of the launch location respectively; Create a mapping between regional geographic information, launch location list, and type list: Step 1.12: Combine the real launch location-type list mapping obtained in step 1.1, the virtual launch location-type list mapping obtained in step 1.7, and the regional geographic information-launch location list mapping obtained in step 1.11 to ultimately establish a regional geographic information-launch location list-type list mapping; If, in step 1.10, a virtual launch position coincides with a real launch position, then extract the types not present in the real launch position from the target type list of the virtual launch position and add these types to the real launch position-type list mapping established in step 1.

1. Combine the supplemented real launch position-type list mapping with the virtual launch position-type list mapping obtained in step 1.7, and map all types to the set L of all launch positions obtained in step 1.11 to establish a launch position-type list mapping: f6:L→M Where f6 represents a function that maps elements in set L to elements in set M, where M=M1∪M4 is the set of all type lists; Step 1.13: Combine the regional geographic information-transmission location list mapping obtained in step 1.11 with the transmission location-type list mapping obtained in step 1.12 to create a regional geographic information-transmission location list-type list mapping: f7:A→L→M Among them, f7 is a composite mapping function from set A to set L, and then from set L to set M.

6. The infrared target type recognition device guided by emission position sampling matching according to claim 5, characterized in that: The implementation of the transmit position estimation module includes the following steps: Step 2.1: Calculate the launch time t of target o o : t o =inf{t o |E o (t) = Fitting(T o ,E o )<θ1}where, T o 、E o Represent the time variation information and energy variation information of target o respectively, Fitting(T o ,E o ) indicates T o The horizontal axis E o Fitting is done for the vertical axis, inf represents the first value that meets the requirements, t o According to T o and E o The fitting relationship predicts the time forward until the energy is less than the threshold θ1; Step 2.2: Calculate the longitude x of the launch point of target o o and latitude y o : x o ={λ o (t o )|λ o (t)=Fitting(T o ,λ o )} Among them, λ o 、 They respectively represent the longitude change information and latitude change information of the target o.

Images