Target recognition method, device, electronic device and storage medium

Abstract

The present disclosure relates to a target identification method, device, electronic device and storage medium, the method comprising: obtaining detection data for a first target for each load among multiple loads; judging whether the first target actually exists based on the detection confidence and load credibility corresponding to each load; when it is judged that the first target actually exists, obtaining, for each load, the degree of matching between the detection attribute characteristics corresponding to the load and the preset attribute characteristics corresponding to each preset type among multiple preset types; determining the type of the first target based on the recognition confidence and matching degree corresponding to each load. In this way, when performing target detection, the detection data for the same target from multiple loads are considered at the same time, that is, the fused information can be fully utilized to make the considered information richer and more comprehensive. Compared with the method of using only a single system load for target detection, the "fusion" identification method provided by the present disclosure has higher recognition accuracy.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, and more specifically, to a target recognition method, device, electronic device, and storage medium. Background Art

[0002] When observing the same target, payloads of different systems may use different viewing angles, spectrum bands, and resolutions. For example, visible light information focuses on describing the geometric shape, color, and texture of the target; infrared information focuses on the thermal emission characteristics of the target's thermal spectrum; radar echoes focus on the electromagnetic spectrum reflection characteristics of the target; hyperspectral information focuses on the reflection characteristics of the target's material to the wavelengths of each spectrum band; and electronic reconnaissance focuses on capturing the characteristics of the target from the target's electromagnetic activity characteristics.

[0003] However, if only one type of payload is used for target detection, the accuracy of target detection may be poor because the features considered are relatively single and not comprehensive enough. Summary of the invention

[0004] The present disclosure provides a target recognition method, device, electronic device and storage medium to at least solve the problem in the above-mentioned related technologies that if only one type of payload is used for target detection, the characteristics considered are relatively single and not comprehensive, which may lead to poor accuracy of target detection.

[0005] According to a first aspect of an embodiment of the present disclosure, a target recognition method is provided, comprising: obtaining detection data for a first target for each load among a plurality of loads, wherein the detection data comprises detection confidence, recognition confidence and detection attribute characteristics, wherein the detection confidence is used to characterize the possibility that the first target actually exists, and the recognition confidence is used to characterize the possibility that the type of the first target is predicted to be each preset type among a plurality of preset types; judging whether the first target actually exists based on the detection confidence and load credibility corresponding to each load among the plurality of loads, wherein the load credibility is used to characterize the credibility of the detection capability of the corresponding load; in the case that it is judged that the first target actually exists, obtaining, for each load among the plurality of loads, the degree of matching between the detection attribute characteristics corresponding to the load and the preset attribute characteristics corresponding to each preset type among the plurality of preset types; determining the type of the first target based on the recognition confidence and the degree of matching corresponding to each load among the plurality of loads.

[0006] Optionally, judging whether the first target actually exists based on the detection confidence and load credibility corresponding to each of the multiple loads includes: calculating a single load existence probability function corresponding to each of the multiple loads based on the detection confidence and load credibility corresponding to each of the multiple loads, wherein the single load existence probability function corresponding to each load is used to indicate the probability that the first target predicted by the load actually exists, the probability that it does not exist, or the probability that it cannot be determined whether it exists; calculating a mixed existence probability function based on the single load existence probability function corresponding to each load, wherein the mixed existence probability function is used to characterize the total probability that the first target predicted by the multiple loads actually exists, the total probability that it does not exist, or the total probability that it cannot be determined whether it exists; judging whether the first target actually exists based on the mixed existence probability function.

[0007] Optionally, the single load existence probability function is expressed by the following formula:

[0008]

[0009]

[0010]

[0011] in, represents the probability that the first target predicted by the i-th load among the multiple loads actually exists, represents the probability that the first target predicted by the i-th load does not exist, represents the probability that the first target cannot be determined to exist as predicted by the i-th load, represents the detection confidence corresponding to the i-th load, represents the load credibility of the i-th load.

[0012] Optionally, the mixed existence probability function is expressed by the following formula:

[0013]

[0014]

[0015]

[0016]

[0017] in, represents the total probability that the first target actually exists, predicted by the multiple payloads. represents the total probability that the first target does not exist as predicted by the multiple payloads, represents the total probability of being unable to determine whether the first target exists, predicted by the multiple loads, n is the total number of the multiple loads, The value of is 1, 0 or -1. There is a conflict coefficient.

[0018] Optionally, determining the type of the first target based on the recognition confidence and the degree of matching corresponding to each load in the multiple loads includes: obtaining a single load type probability function corresponding to each load based on the recognition confidence and the degree of matching corresponding to each load in the multiple loads, wherein the single load type probability function corresponding to each load is used to indicate the probability that the first target predicted by the load belongs to each preset type in the multiple preset types or the probability that the type of the first target cannot be determined; calculating a mixed type probability function based on the single load type probability function corresponding to each load, wherein the mixed type probability function is used to characterize the probability that the first target jointly predicted by the multiple loads belongs to each preset type in the multiple preset types or the probability that the type of the first target cannot be determined; determining the type of the first target based on the mixed type probability function.

[0019] Optionally, the single load type probability function is obtained by the following formula:

[0020]

[0021]

[0022]

[0023] in, represents the probability that the i-th load among the multiple loads predicts the type of the first target to be the j-th preset type among the multiple preset types, represents the probability that the i-th load cannot determine the type of the first target, a recognition confidence level for predicting the type of the first target as the j-th preset type for the i-th load, is the matching degree between the detection attribute feature corresponding to the i-th load and the preset attribute feature corresponding to the j-th preset type, is the distance between the detection attribute feature of the i-th load and the preset attribute feature corresponding to the j-th preset type, represents the jth preset type among the multiple preset types.

[0024] Optionally, the mixed type probability function is obtained by the following formula:

[0025]

[0026]

[0027]

[0028] in, represents the probability that the first target predicted by the multiple loads belongs to the jth preset type among the multiple preset types, represents the probability that the type of the first target cannot be determined, as predicted by the multiple loads, n is the total number of the multiple loads, is the type conflict coefficient.

[0029] According to a second aspect of an embodiment of the present disclosure, a target recognition device is provided, comprising: a detection data acquisition module, configured to acquire detection data of each load in a plurality of loads for a first target, wherein the detection data comprises detection confidence, recognition confidence and detection attribute characteristics, wherein the detection confidence is used to characterize the possibility that the first target actually exists, and the recognition confidence is used to characterize the possibility that the type of the first target is predicted to be each preset type in a plurality of preset types; an existence judgment module, configured to judge whether the first target actually exists based on the detection confidence and load credibility corresponding to each load in the plurality of loads, wherein the load credibility is used to characterize the credibility of the detection capability of the corresponding load; a matching degree acquisition module, configured to, when it is judged that the first target actually exists, acquire, for each load in the plurality of loads, the matching degree between the detection attribute characteristics corresponding to the load and the preset attribute characteristics corresponding to each preset type in the plurality of preset types; a type determination module, configured to determine the type of the first target based on the recognition confidence and matching degree corresponding to each load in the plurality of loads.

[0030] Optionally, the existence judgment module is configured to: calculate a single load existence probability function corresponding to each of the multiple loads based on the detection confidence and load credibility corresponding to each of the multiple loads, wherein the single load existence probability function corresponding to each load is used to indicate the probability that the first target predicted by the load actually exists, the probability that it does not exist, or the probability that it cannot be determined whether it exists; calculate a mixed existence probability function based on the single load existence probability function corresponding to each load, wherein the mixed existence probability function is used to characterize the total probability that the first target predicted by the multiple loads actually exists, the total probability that it does not exist, or the total probability that it cannot be determined whether it exists; and judge whether the first target actually exists based on the mixed existence probability function.

[0031] Optionally, the single load existence probability function is expressed by the following formula:

[0032]

[0033]

[0034]

[0035] in, represents the probability that the first target predicted by the i-th load among the multiple loads actually exists, represents the probability that the first target predicted by the i-th load does not exist, represents the probability that the first target cannot be determined to exist as predicted by the i-th load, represents the detection confidence corresponding to the i-th load, represents the load credibility of the i-th load.

[0036] Optionally, the mixed existence probability function is expressed by the following formula:

[0037]

[0038]

[0039]

[0040]

[0041] in, represents the total probability that the first target actually exists, predicted by the multiple payloads. represents the total probability that the first target does not exist as predicted by the multiple payloads, represents the total probability of being unable to determine whether the first target exists, predicted by the multiple loads, n is the total number of the multiple loads, The value of is 1, 0 or -1. There is a conflict coefficient.

[0042] Optionally, the type determination module is configured to: obtain a single load type probability function corresponding to each load based on the recognition confidence and matching degree corresponding to each load in the multiple loads, wherein the single load type probability function corresponding to each load is used to indicate the probability that the first target predicted by the load belongs to each preset type in the multiple preset types or the probability that the type of the first target cannot be determined; calculate a mixed type probability function based on the single load type probability function corresponding to each load, wherein the mixed type probability function is used to characterize the probability that the first target jointly predicted by the multiple loads belongs to each preset type in the multiple preset types or the probability that the type of the first target cannot be determined; determine the type of the first target based on the mixed type probability function.

[0043] Optionally, the single load type probability function is obtained by the following formula:

[0044]

[0045]

[0046]

[0047] in, represents the probability that the i-th load among the multiple loads predicts the type of the first target to be the j-th preset type among the multiple preset types, represents the probability that the i-th load cannot determine the type of the first target, a recognition confidence level for predicting the type of the first target as the j-th preset type for the i-th load, is the matching degree between the detection attribute feature corresponding to the i-th load and the preset attribute feature corresponding to the j-th preset type, is the distance between the detection attribute feature of the i-th load and the preset attribute feature corresponding to the j-th preset type, represents the jth preset type among the multiple preset types.

[0048] Optionally, the mixed type probability function is obtained by the following formula:

[0049]

[0050]

[0051]

[0052] in, represents the probability that the first target predicted by the multiple loads belongs to the jth preset type among the multiple preset types, represents the probability that the type of the first target cannot be determined, as predicted by the multiple loads, n is the total number of the multiple loads, is the type conflict coefficient.

[0053] According to a third aspect of an embodiment of the present disclosure, there is provided an electronic device, comprising: a processor; and a memory for storing instructions executable by the processor; wherein the processor is configured to execute the instructions to implement a target recognition method according to the present disclosure.

[0054] According to a fourth aspect of an embodiment of the present disclosure, a computer-readable storage medium is provided. When instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to perform a target recognition method according to the present disclosure.

[0055] The technical solution provided by the embodiments of the present disclosure brings at least the following beneficial effects:

[0056] In the present disclosure, when performing target detection, the detection data of multiple payloads for the same target are considered at the same time, that is, the fusion information can be fully utilized to make the considered information richer and more comprehensive. Compared with the method of using only a single system payload for target detection, the "fusion" recognition method provided by the present disclosure has higher recognition accuracy. In addition, the method of the present disclosure has a strong adaptability to the number and type of payloads involved in the fusion, can adapt to loosely coupled intersatellite links and situations where individual payload data is missing, and has good robustness and anti-interference performance for target recognition.

[0057] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] The drawings herein are incorporated into and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the description are used to explain the principles of the present disclosure, and do not constitute improper limitations on the present disclosure.

[0059] Figure 1 is a flow chart illustrating a target recognition method according to an exemplary embodiment of the present disclosure;

[0060] Figure 2is a flowchart illustrating a detection target presence layer decision fusion algorithm according to an exemplary embodiment of the present disclosure;

[0061] Figure 3 is a flow chart showing a target identity recognition layer decision fusion algorithm according to an exemplary embodiment of the present disclosure;

[0062] Figure 4 is a block diagram illustrating an object recognition apparatus according to an exemplary embodiment of the present disclosure;

[0063] Figure 5 is a block diagram illustrating an electronic device according to an exemplary embodiment of the present disclosure. DETAILED DESCRIPTION

[0064] In order to enable ordinary persons in the art to better understand the technical solutions of the present disclosure, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the accompanying drawings.

[0065] It should be noted that the terms "first", "second", etc. in the specification and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The implementation methods described in the following examples do not represent all implementation methods consistent with the present disclosure. Instead, they are merely examples of devices and methods consistent with some aspects of the present disclosure as detailed in the attached claims.

[0066] It should be noted that the phrase "at least one of the items" in the present disclosure includes three types of parallel situations: "any one of the items", "a combination of any number of the items", and "all of the items". For example, "including at least one of A and B" includes the following three types of parallel situations: (1) including A; (2) including B; (3) including A and B. Another example is "executing at least one of step 1 and step 2" which means the following three types of parallel situations: (1) executing step 1; (2) executing step 2; (3) executing step 1 and step 2.

[0067] On-orbit detection and recognition of targets of interest in satellite remote sensing images is an important field of satellite remote sensing applications. On the one hand, it can effectively filter out a large amount of invalid data and reduce the pressure of on-board storage and satellite-to-ground data transmission; on the other hand, it can quickly focus on targets of interest and improve the timeliness of intelligence information, which is of great significance in emergency rescue and military affairs. In addition, when different payloads observe the same target, the viewing angles, spectrum bands, and resolutions they use may be different from each other. If only one type of payload is used for target detection, the accuracy of target detection may be poor because the features considered are relatively single and not comprehensive enough.

[0068] In order to solve the above-mentioned problems existing in the related art, the target recognition method, device, electronic device and storage medium provided by the present disclosure, when performing target detection, simultaneously consider the detection data of multiple payloads for the same target, that is, it can make full use of fusion information to make the considered information richer and more comprehensive. Compared with the method of using only a single system payload for target detection, the "fusion" recognition method provided by the present disclosure has higher recognition accuracy. In addition, the method of the present disclosure has a strong adaptability to the number and type of payloads involved in the fusion, can adapt to loosely coupled intersatellite links and situations where individual payload data is missing, and has good robustness and anti-interference performance for target recognition.

[0069] Figure 1 is a flowchart illustrating a target recognition method according to an exemplary embodiment of the present disclosure.

[0070] It should be noted that the target recognition method provided by the present disclosure mainly includes two levels of decision fusion, wherein the first level is "target existence decision fusion discrimination" and the second level is "target identity decision fusion recognition". The following will specifically describe these two implementation processes in conjunction with the accompanying drawings.

[0071] Reference Figure 1 In step 101, detection data of each load among multiple loads for the first target can be obtained, wherein the detection data can include detection confidence, recognition confidence and detection attribute characteristics, wherein "detection confidence" can be used to characterize the possibility that the first target does exist, and "recognition confidence" can be used to characterize the possibility of predicting the type of the first target as each preset type among multiple preset types.

[0072] It should be noted that in the present disclosure, it is assumed that all targets detected by the payloads have completed target association based on information such as location, that is, the present disclosure can only use the hierarchical decision method to identify the associated targets, that is, the input of the onboard multi-level hierarchical decision fusion target identification method can be the detection results of each payload in multiple payloads for the same target. In addition, "payload" can also be called "information source".

[0073] The present disclosure may also provide a "payload processing unit" and a "fusion processing unit". The "payload processing unit" can be used to process images or signals of a target acquired by a single payload. For example, the "payload processing unit" can be used to extract attribute features of the target, identify the identity of the target, determine the location of the target, and so on. In addition, the aforementioned "multiple payloads" may include but are not limited to: visible light satellite payloads, infrared satellite payloads, hyperspectral satellite payloads, SAR satellite payloads, AIS satellite payloads, and so on. The data acquired may be images or electromagnetic signals; the "fusion processing unit" is mainly used to aggregate the processing results of various payload processing units for information fusion.

[0074] It should be noted that the above-mentioned "payload processing unit" and "fusion processing unit" can be configured on the same satellite or on different satellites. If they are configured on the same satellite, the processing results of each "payload processing unit" can be directly aggregated through the intra-satellite data link; if they are configured on different satellites, the processing results of each "payload processing unit" can be aggregated to the "fusion processing unit" through the inter-satellite link. In addition, the aggregated information can include the processing results of each "payload processing unit".

[0075] Then, the "fusion processing unit" can associate the targets according to the detection positions of the targets contained in the processing results of different "load processing units" to ensure that the detection data of multiple loads for the same target are fused to avoid the erroneous fusion of detection data for different targets. Next, the "fusion processing unit" can adopt decision-level fusion and output the final decision recognition result.

[0076] In step 102, whether the first target actually exists can be determined based on the detection confidence and load credibility corresponding to each of the multiple loads, wherein the load credibility can be used to characterize the credibility of the detection capability of the corresponding load.

[0077] Exemplarily, after the detection data for the first target from different loads are aggregated, a DS evidence reasoning algorithm can be constructed based on the detection confidence corresponding to each load and the load credibility to determine whether the detected first target actually exists. For example, for each detected target, the detection data of all loads that detect the target can be fused to output the existence confidence of the target. In addition, a preset threshold can also be set. If the calculated existence confidence is lower than the preset threshold, the target can be filtered out as a false alarm. In this way, it can ensure that false alarms are effectively removed and the accuracy of target detection is improved.

[0078] According to an exemplary embodiment of the present disclosure, a single load existence probability function corresponding to each of the multiple loads can be calculated based on the detection confidence and load credibility corresponding to each of the multiple loads, wherein the single load existence probability function corresponding to each load can be used to indicate the probability that the first target predicted by the load actually exists, the probability that it does not actually exist, or the probability that it cannot be determined whether it exists.

[0079] For example, an "existence decision space" can be set, which can include three decision types: "definitely exists", "definitely does not exist", and "cannot determine whether it exists". In addition, "1" can be used to represent "definitely exists", "0" can be used to represent "definitely does not exist", and "-1" can be used to represent "cannot determine whether it exists". Assume that the loads involved in the fusion are: e1, e2, ..., en, and these loads can correspond to a "detection target accuracy rate" respectively: , ,… , the meaning of “target detection accuracy” can be: “the probability of confirming as the target among the detected suspicious targets”.

[0080] Detection target accuracy = number of correctly detected targets / total number of detected targets = detection rate / (detection rate + false alarm rate);

[0081] Detection rate = number of correctly detected targets / number of all truly correct targets

[0082] False alarm rate = number of wrongly detected targets / total number of correct targets

[0083] For example, suppose there are 10 cars on the ground, and a certain load detects 11 cars, and 8 of the 11 cars detected by this load are indeed cars, and 3 of them are actually rocks. Then the "number of correct targets detected" of this load is 8, the "number of all targets detected" of this load is 11, the "number of all truly correct targets" is 10, and the "number of wrong targets detected" of this load is 3. And, the "correct detection rate" of the target at this time is 8 / 11; the "detection rate" is 8 / 10; and the "false alarm rate" is 3 / 10.

[0084] It should be noted that the above-mentioned "target detection accuracy" can be an evaluation of the detection capability of a certain load obtained based on the actual statistics (detection rate, false alarm rate) of the load over a period of history, and the "target detection accuracy" can also be called "load credibility", that is, "source credibility".

[0085] Then, based on the single load existence probability function corresponding to each load, a mixed existence probability function can be calculated, wherein the mixed existence probability function can be used to characterize the total probability that the first target predicted by multiple loads does exist, the total probability that it does not exist, or the total probability that it cannot be determined whether it exists. Next, based on the mixed existence probability function, it can be determined whether the first target does exist.

[0086] Figure 2 FIG. 1 is a flow chart showing a target presence layer decision fusion algorithm according to an exemplary embodiment of the present disclosure. Figure 2 , for each of the n types of loads, the detection confidence and load credibility corresponding to the load can be obtained, and then the single load existence probability function corresponding to the load can be calculated based on the detection confidence and load credibility corresponding to the load. Then, the n single load existence probability functions corresponding to the n types of loads can be fused to obtain a mixed existence probability function. Next, based on the calculated mixed existence probability function, it can be judged whether the corresponding target is "really present", "really not present" or "cannot be determined whether it exists".

[0087] According to an exemplary embodiment of the present disclosure, it is assumed that the n detection confidences corresponding to the n types of loads are respectively: , , …, Then the "single load existence probability function" can be expressed by the following formula:

[0088]

[0089]

[0090]

[0091] in, It can represent the probability that the first target predicted by the i-th load among multiple loads actually exists. It can represent the probability that the first target predicted by the i-th load does not exist. It can represent the probability that the first target cannot be determined whether it exists, as predicted by the i-th load. It can represent the detection confidence corresponding to the i-th load, It can represent the load credibility of the i-th load, that is, the source credibility.

[0092] According to an exemplary embodiment of the present disclosure, the mixed existence probability function can be expressed by the following formula:

[0093]

[0094]

[0095]

[0096]

[0097] in, It can represent the total probability that the first target predicted by multiple payloads actually exists. It can represent the total probability that the first target predicted by multiple payloads does not exist. It can represent the total probability of being unable to determine whether the first target exists, predicted by multiple loads. n can be the total number of multiple loads. The value of can be 1 (definitely exists), 0 (definitely does not exist), or -1 (it is not certain whether it exists). There may be a conflict coefficient.

[0098] After calculating the value of the mixed existence probability function, the decision branch with the largest value can be determined, that is, , and Then, based on the decision branch with the largest value, it can be determined whether the corresponding target actually exists. For example, if the decision branch with the largest value is , it can be shown that the corresponding target does exist; if the decision branch with the largest value is , it can be shown that the corresponding target does not exist; if the decision branch with the largest value is , it means that it is impossible to determine whether the corresponding target exists.

[0099] In step 103, when it is determined that the first target does exist, for each of the multiple loads, the degree of matching between the detection attribute feature corresponding to the load and the preset attribute feature corresponding to each of the multiple preset types can be obtained.

[0100] Specifically, a “type decision space” may be set, which may include all possible target types. Exemplarily, the “type decision space” may be represented as: , -1), where It can indicate the identity of the target that each member star has determined, that is, It can indicate the type of target that has been determined by various loads; "-1" can indicate "the type of target cannot be determined". As mentioned above, assuming that there are n types of loads involved in the fusion, that is, n sources are: e1, e2, ..., en, then the matching degree between the detection attribute characteristics corresponding to the i-th load and the preset attribute characteristics corresponding to the j-th preset type can be defined as follows:

[0101]

[0102] in, is the distance between the detected attribute feature of the i-th load and the preset attribute feature corresponding to the j-th preset type. Exemplarily, the “distance” may be the Euclidean distance.

[0103] It should be noted that various payloads may have their own corresponding attribute features. For example, for imaging payloads, the corresponding attribute features may be, but are not limited to, simple shape features such as Hu moment and aspect ratio; or, more abundant feature descriptions such as HOG and Daisy.

[0104] Obtaining the matching degree between the detected attribute features and the preset attribute features is equivalent to re-testing the identification results of a single load, and the test results can be added to the identity discrimination basis, that is, the test results can be added to the type discrimination basis. In this way, the identification results can be properly corrected when the load capacity is uncertain, thereby ensuring the accuracy of the identity classification decision.

[0105] In step 104, the type of the first target may be determined based on the recognition confidence and matching degree corresponding to each of the multiple loads. Exemplarily, the maximum value of the mixed type probability function corresponding to each of the multiple prediction types predicted for the target may be determined, and then the prediction type corresponding to the maximum value may be used as the final type of the target.

[0106] In the present disclosure, the matching degree between the detection attribute features corresponding to various loads and the preset attribute features in the attribute library can be calculated based on whether the fusion detection target actually exists. At the same time, all loads participating in the fusion have their own judgment results on the identity of the target. At this time, based on the attribute matching degree and the judgment results of various loads on the identity of the target, a single load type probability function corresponding to various loads can be established, and then multiple single load type probability functions corresponding to multiple loads can be fused to obtain a mixed type probability function to achieve classification and identification of the target. Specifically, a decision reasoning can be performed based on multiple single load type probability functions using the "DS evidence reasoning algorithm" to output the type to which the target belongs and the confidence level of the type to which it belongs.

[0107] In this way, when determining the type of target, the determination results of various loads can be integrated as the basis for judgment, that is, the fused information can be fully utilized to make the considered information richer and more comprehensive, which can ensure the reliability and accuracy of determining the target type.

[0108] According to an exemplary embodiment of the present disclosure, a single load type probability function corresponding to each load can be obtained based on the recognition confidence and matching degree corresponding to each load among the multiple loads, wherein the single load type probability function corresponding to each load can be used to indicate the probability that the first target predicted by the load belongs to each preset type among multiple preset types or the probability that the type of the first target cannot be determined.

[0109] Then, based on the single load type probability function corresponding to each load, a mixed type probability function can be calculated, wherein the mixed type probability function can be used to characterize the probability that the first target predicted by multiple loads belongs to each preset type in multiple preset types or the probability that the type of the first target cannot be determined. Next, the type of the first target can be determined based on the mixed type probability function.

[0110] Figure 3 is a flow chart showing a target identity recognition layer decision fusion algorithm according to an exemplary embodiment of the present disclosure. Figure 3 For each of the n types of loads, that is, for each of the n types of sources, the degree of match between the detection attribute feature corresponding to the load and the preset attribute feature pre-stored in the attribute library can be calculated. Then, based on the recognition confidence and matching degree corresponding to each load, the single load type probability function corresponding to the load can be determined. Next, multiple single load type probability functions corresponding to multiple loads can be fused to obtain a mixed type probability function. Then, based on the calculated mixed type probability function, a classification decision for the target identity can be made.

[0111] According to an exemplary embodiment of the present disclosure, the above-mentioned “single load type probability function” can be obtained by the following formula:

[0112]

[0113]

[0114]

[0115] in, It can represent the probability that the i-th load among multiple loads predicts the type of the first target to be the j-th preset type among multiple preset types, It can represent the probability that the i-th load cannot determine the type of the first target, The recognition confidence that the type of the first target can be predicted as the jth preset type for the i-th load, It can be the matching degree between the detection attribute feature corresponding to the i-th load and the preset attribute feature corresponding to the j-th preset type. It can be the distance between the detection attribute feature of the i-th load and the preset attribute feature corresponding to the j-th preset type, It can represent the jth preset type among multiple preset types.

[0116] For example, if the target is found to exist by integrating the detection data of multiple loads, for example, if the target "car" is found to exist, the type of the "car" can be further determined. For example, it can be determined whether the "car" is a sedan, a bus, or an off-road vehicle. The value of can be: "car", "bus" or "SUV". Or, assuming that the target is found to exist by integrating the detection data of multiple loads, for example, if the target "ship" is found to exist, the type of the "ship" can be further determined. For example, it can also be determined whether the "ship" is a ship, a boat or a motorboat, etc. At this time, The value of can be: "ship", "boat" or "motorboat".

[0117] According to an exemplary embodiment of the present disclosure, the above-mentioned “mixed type probability function” can be obtained by the following formula:

[0118]

[0119]

[0120]

[0121] in, It can represent the probability that the first target predicted by multiple payloads belongs to the jth preset type among multiple preset types. It can represent the probability that the type of the first target cannot be determined, which is predicted by multiple loads. n can be the total number of multiple loads. Can be type conflict coefficient.

[0122] Furthermore, as described above, the maximum value of the mixed type probability function corresponding to each prediction type among the multiple prediction types predicted for the target can be determined, and then the prediction type corresponding to the maximum value can be used as the final type of the target.

[0123] It should be noted that there are many mature multi-source information fusion technologies, including data-level fusion, feature-level fusion, decision-level fusion, etc. This disclosure mainly aims at the special scenario of multi-satellite on-orbit joint observation, and proposes a hierarchical decision fusion algorithm that combines payload performance and target characteristics, which is suitable for multi-satellite and multi-payload information fusion target detection and recognition under low-bandwidth intersatellite link conditions.

[0124] That is, in the present disclosure, a multi-source information fusion target identification method that can be applied to the scenario of limited satellite computing power is proposed. This method fully considers the parameters such as false alarm rate and detection rate that can fully reflect the actual capabilities of each information source, that is, it can be used to measure the credibility of the detection capability of the corresponding payload, so that the positioning of each information source in information fusion is more scientific, and a better fusion effect is guaranteed. Moreover, through the strategy of hierarchical decision-making, it is possible to effectively utilize a variety of fusion information, thereby ensuring the accuracy of the final judgment result. In addition, this method has a strong adaptability to the number and type of payloads involved in the fusion, can adapt to loosely coupled intersatellite links and the situation where individual payload data is missing, and has good robustness and anti-interference performance for target identification.

[0125] Figure 4 is a block diagram illustrating an object recognition apparatus according to an exemplary embodiment of the present disclosure.

[0126] It should be noted that the target recognition method provided by the present disclosure mainly includes two levels of decision fusion, wherein the first level is "target existence decision fusion discrimination" and the second level is "target identity decision fusion recognition".

[0127] Reference Figure 4 The target recognition device 400 may include a detection data acquisition module 401, an existence judgment module 402, a matching degree acquisition module 403 and a type determination module 404.

[0128] The detection data acquisition module 401 can obtain detection data of each load among multiple loads for the first target, wherein the detection data can include detection confidence, recognition confidence and detection attribute characteristics, wherein "detection confidence" can be used to characterize the possibility that the first target does exist, and "recognition confidence" can be used to characterize the possibility of predicting the type of the first target as each preset type among multiple preset types.

[0129] It should be noted that in the present disclosure, it is assumed that all targets detected by the payloads have completed target association based on information such as location, that is, the present disclosure can only use the hierarchical decision method to identify the associated targets, that is, the input of the onboard multi-level hierarchical decision fusion target identification method can be the detection results of each payload in multiple payloads for the same target. In addition, "payload" can also be called "information source".

[0130] The present disclosure may also provide a "payload processing unit" and a "fusion processing unit". The "payload processing unit" can be used to process images or signals of a target acquired by a single payload. For example, the "payload processing unit" can be used to extract attribute features of the target, identify the identity of the target, determine the location of the target, and so on. In addition, the aforementioned "multiple payloads" may include but are not limited to: visible light satellite payloads, infrared satellite payloads, hyperspectral satellite payloads, SAR satellite payloads, AIS satellite payloads, and so on. The data acquired may be images or electromagnetic signals; the "fusion processing unit" is mainly used to aggregate the processing results of various payload processing units for information fusion.

[0131] It should be noted that the above-mentioned "payload processing unit" and "fusion processing unit" can be configured on the same satellite or on different satellites. If they are configured on the same satellite, the processing results of each "payload processing unit" can be directly aggregated through the intra-satellite data link; if they are configured on different satellites, the processing results of each "payload processing unit" can be aggregated to the "fusion processing unit" through the inter-satellite link. In addition, the aggregated information can include the processing results of each "payload processing unit".

[0132] Then, the "fusion processing unit" can associate the targets according to the detection positions of the targets contained in the processing results of different "load processing units" to ensure that the detection data of multiple loads for the same target are fused to avoid the erroneous fusion of detection data for different targets. Next, the "fusion processing unit" can adopt decision-level fusion and output the final decision recognition result.

[0133] The existence determination module 402 can determine whether the first target actually exists based on the detection confidence and load credibility corresponding to each of the multiple loads, wherein the load credibility can be used to characterize the credibility of the detection capability of the corresponding load.

[0134] Exemplarily, after the detection data for the first target from different loads are aggregated, a DS evidence reasoning algorithm can be constructed based on the detection confidence corresponding to each load and the load credibility to determine whether the detected first target actually exists. For example, for each detected target, the detection data of all loads that detect the target can be fused to output the existence confidence of the target. In addition, a preset threshold can also be set. If the calculated existence confidence is lower than the preset threshold, the target can be filtered out as a false alarm. In this way, it can ensure that false alarms are effectively removed and the accuracy of target detection is improved.

[0135] According to an exemplary embodiment of the present disclosure, the existence judgment module 402 can calculate a single load existence probability function corresponding to each of the multiple loads based on the detection confidence and load credibility corresponding to each of the multiple loads, wherein the single load existence probability function corresponding to each load can be used to indicate the probability that the first target predicted by the load actually exists, the probability that it does not actually exist, or the probability that it cannot be determined whether it exists.

[0136] For example, an "existence decision space" can be set, which can include three decision types: "definitely exists", "definitely does not exist", and "cannot determine whether it exists". In addition, "1" can be used to represent "definitely exists", "0" can be used to represent "definitely does not exist", and "-1" can be used to represent "cannot determine whether it exists". Assume that the loads involved in the fusion are: e1, e2, ..., en, and these loads can correspond to a "detection target accuracy rate" respectively: , ,… , the meaning of “target detection accuracy” can be: “the probability of confirming as the target among the detected suspicious targets”.

[0137] Detection target accuracy = number of correctly detected targets / total number of detected targets = detection rate / (detection rate + false alarm rate);

[0138] Detection rate = number of correctly detected targets / number of all truly correct targets

[0139] False alarm rate = number of wrongly detected targets / total number of correct targets

[0140] It should be noted that the above-mentioned "target detection accuracy" can be an evaluation of the detection capability of a certain load obtained based on the actual statistics (detection rate, false alarm rate) of the load over a period of history, and the "target detection accuracy" can also be called "load credibility", that is, "source credibility".

[0141] Then, the existence judgment module 402 can calculate a mixed existence probability function based on the single load existence probability function corresponding to each load, wherein the mixed existence probability function can be used to characterize the total probability that the first target predicted by multiple loads does exist, the total probability that it does not exist, or the total probability that it cannot be determined whether it exists. Next, the existence judgment module 402 can judge whether the first target does exist based on the mixed existence probability function.

[0142] According to an exemplary embodiment of the present disclosure, it is assumed that the n detection confidences corresponding to the n types of loads are respectively: , , …, Then the "single load existence probability function" can be expressed by the following formula:

[0143]

[0144]

[0145]

[0146] in, It can represent the probability that the first target predicted by the i-th load among multiple loads actually exists. It can represent the probability that the first target predicted by the i-th load does not exist. It can represent the probability that the first target cannot be determined whether it exists, as predicted by the i-th load. It can represent the detection confidence corresponding to the i-th load, It can represent the load credibility of the i-th load, that is, the source credibility.

[0147] According to an exemplary embodiment of the present disclosure, the mixed existence probability function can be expressed by the following formula:

[0148]

[0149]

[0150]

[0151]

[0152] in, It can represent the total probability that the first target predicted by multiple payloads actually exists. It can represent the total probability that the first target predicted by multiple payloads does not exist. It can represent the total probability of being unable to determine whether the first target exists, predicted by multiple loads. n can be the total number of multiple loads. The value of can be 1 (definitely exists), 0 (definitely does not exist), or -1 (it is not certain whether it exists). There may be a conflict coefficient.

[0153] After calculating the value of the mixed existence probability function, the decision branch with the largest value can be determined, that is, , and Then, based on the decision branch with the largest value, it can be determined whether the corresponding target actually exists. For example, if the decision branch with the largest value is , it can be shown that the corresponding target does exist; if the decision branch with the largest value is , it can be shown that the corresponding target does not exist; if the decision branch with the largest value is , it means that it is impossible to determine whether the corresponding target exists.

[0154] When it is determined that the first target does exist, the matching degree acquisition module 403 may acquire, for each of the multiple loads, a matching degree between the detection attribute feature corresponding to the load and the preset attribute feature corresponding to each of the multiple preset types.

[0155] It should be noted that various payloads may have their own corresponding attribute features. For example, for imaging payloads, the corresponding attribute features may be, but are not limited to, simple shape features such as Hu moment and aspect ratio; or, more abundant feature descriptions such as HOG and Daisy.

[0156] Obtaining the matching degree between the detected attribute features and the preset attribute features is equivalent to re-testing the identification results of a single load, and the test results can be added to the identity discrimination basis, that is, the test results can be added to the type discrimination basis. In this way, the identification results can be properly corrected when the load capacity is uncertain, thereby ensuring the accuracy of the identity classification decision.

[0157] The type determination module 404 may determine the type of the first target based on the recognition confidence and matching degree corresponding to each of the multiple loads. For example, the maximum value of the mixed type probability function corresponding to each of the multiple prediction types predicted for the target may be determined, and then the prediction type corresponding to the maximum value may be used as the final type of the target.

[0158] In the present disclosure, the matching degree between the detection attribute features corresponding to various loads and the preset attribute features in the attribute library can be calculated based on whether the fusion detection target actually exists. At the same time, all loads participating in the fusion have their own judgment results on the identity of the target. At this time, based on the attribute matching degree and the judgment results of various loads on the identity of the target, a single load type probability function corresponding to various loads can be established, and then multiple single load type probability functions corresponding to multiple loads can be fused to obtain a mixed type probability function to achieve classification and identification of the target. Specifically, a decision reasoning can be performed based on multiple single load type probability functions using the "DS evidence reasoning algorithm" to output the type to which the target belongs and the confidence level of the type to which it belongs.

[0159] In this way, when determining the type of target, the determination results of various loads can be integrated as the basis for judgment, that is, the fused information can be fully utilized to make the considered information richer and more comprehensive, which can ensure the reliability and accuracy of determining the target type.

[0160] According to an exemplary embodiment of the present disclosure, the type determination module 404 can obtain a single load type probability function corresponding to each load based on the recognition confidence and matching degree corresponding to each load among the multiple loads, wherein the single load type probability function corresponding to each load can be used to indicate the probability that the first target predicted by the load belongs to each preset type among multiple preset types or the probability that the type of the first target cannot be determined.

[0161] Then, the type determination module 404 may calculate a mixed type probability function based on the single load type probability function corresponding to each load, wherein the mixed type probability function may be used to characterize the probability that the first target predicted by the multiple loads belongs to each of the multiple preset types or the probability that the type of the first target cannot be determined. Next, the type determination module 404 may determine the type of the first target based on the mixed type probability function.

[0162] According to an exemplary embodiment of the present disclosure, the above-mentioned “single load type probability function” can be obtained by the following formula:

[0163]

[0164]

[0165]

[0166] in, It can represent the probability that the i-th load among multiple loads predicts the type of the first target to be the j-th preset type among multiple preset types, It can represent the probability that the i-th load cannot determine the type of the first target, The recognition confidence that the type of the first target can be predicted as the jth preset type for the i-th load, It can be the matching degree between the detection attribute feature corresponding to the i-th load and the preset attribute feature corresponding to the j-th preset type. It can be the distance between the detection attribute feature of the i-th load and the preset attribute feature corresponding to the j-th preset type, It can represent the jth preset type among multiple preset types.

[0167] For example, if the target is found to exist by integrating the detection data of multiple loads, for example, if the target "car" is found to exist, the type of the "car" can be further determined. For example, it can be determined whether the "car" is a sedan, a bus, or an off-road vehicle. The value of can be: "car", "bus" or "SUV". Or, assuming that the target is found to exist by integrating the detection data of multiple loads, for example, if the target "ship" is found to exist, the type of the "ship" can be further determined. For example, it can also be determined whether the "ship" is a ship, a boat or a motorboat, etc. At this time, The value of can be: "ship", "boat" or "motorboat".

[0168] According to an exemplary embodiment of the present disclosure, the above-mentioned “mixed type probability function” can be obtained by the following formula:

[0169]

[0170]

[0171]

[0172] in, It can represent the probability that the first target predicted by multiple payloads belongs to the jth preset type among multiple preset types. It can represent the probability that the type of the first target cannot be determined, which is predicted by multiple loads. n can be the total number of multiple loads. Can be type conflict coefficient.

[0173] Furthermore, as described above, the maximum value of the mixed type probability function corresponding to each prediction type among the multiple prediction types predicted for the target can be determined, and then the prediction type corresponding to the maximum value can be used as the final type of the target.

[0174] In the present disclosure, a multi-source information fusion target recognition method applicable to scenarios with limited satellite computing power is proposed. This method fully considers parameters such as false alarm rate and detection rate that can fully reflect the actual capabilities of each information source, that is, it can be used to measure the credibility of the detection capability of the corresponding payload, so that the positioning of each information source in information fusion is more scientific, ensuring a better fusion effect. Moreover, through the strategy of hierarchical decision-making, it is possible to effectively utilize a variety of fusion information, thereby ensuring the accuracy of the final judgment result. In addition, this method has a strong adaptability to the number and type of payloads involved in the fusion, can adapt to loosely coupled intersatellite links and situations where individual payload data is missing, and has good robustness and anti-interference performance for target recognition.

[0175] Figure 5 is a block diagram illustrating an electronic device 500 according to an exemplary embodiment of the present disclosure.

[0176] Reference Figure 5 The electronic device 500 includes at least one memory 501 and at least one processor 502. The at least one memory 501 stores instructions. When the instructions are executed by the at least one processor 502, the target recognition method according to the exemplary embodiment of the present disclosure is executed.

[0177] As an example, the electronic device 500 may be a PC, a tablet device, a personal digital assistant, a smart phone, or other device capable of executing the above instructions. Here, the electronic device 500 is not necessarily a single electronic device, but may also be any device or circuit collection capable of executing the above instructions (or instruction sets) individually or in combination. The electronic device 500 may also be part of an integrated control system or system manager, or may be configured as a portable electronic device interconnected with a local or remote (e.g., via wireless transmission) interface.

[0178] In the electronic device 500, the processor 502 may include a central processing unit (CPU), a graphics processing unit (GPU), a programmable logic device, a dedicated processor system, a microcontroller or a microprocessor. As an example and not limitation, the processor may also include an analog processor, a digital processor, a microprocessor, a multi-core processor, a processor array, a network processor, etc.

[0179] The processor 502 may execute instructions or codes stored in the memory 501, wherein the memory 501 may also store data. Instructions and data may also be sent and received over a network via a network interface device, wherein the network interface device may employ any known transmission protocol.

[0180] The memory 501 may be integrated with the processor 502, for example, by placing RAM or flash memory within an integrated circuit microprocessor or the like. In addition, the memory 501 may include a separate device, such as an external disk drive, a storage array, or any other storage device that can be used by a database system. The memory 501 and the processor 502 may be operatively coupled, or may communicate with each other, such as through an I / O port, a network connection, etc., so that the processor 502 can read files stored in the memory.

[0181] In addition, the electronic device 500 may further include a video display (such as a liquid crystal display) and a user interaction interface (such as a keyboard, a mouse, a touch input device, etc.) All components of the electronic device 500 may be connected to each other via a bus and / or a network.

[0182] According to an exemplary embodiment of the present disclosure, a computer-readable storage medium may also be provided, and when the instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to perform the above-mentioned target recognition method. Examples of computer-readable storage media here include: read-only memory (ROM), random access programmable read-only memory (PROM), electrically erasable programmable read-only memory (EEPROM), random access memory (RAM), dynamic random access memory (DRAM), static random access memory (SRAM), flash memory, non-volatile memory, CD-ROM, CD-R, CD+R, CD-RW, CD+RW, DVD-ROM, DVD-R, DVD+R, DVD-RW, DVD+RW, DVD-RAM, BD-ROM, BD-R, BD-R LTH, BD-RE, Blu-ray or optical disk storage, hard disk drive (HDD), solid state drive (SSD), card storage (such as, multimedia card, secure digital (SD) card or extreme digital (XD) card), magnetic tape, floppy disk, magneto-optical data storage device, optical data storage device, hard disk, solid state disk and any other device, any other device is configured to store computer programs and any associated data, data files and data structures in a non-transitory manner and provide the computer programs and any associated data, data files and data structures to a processor or computer so that the processor or computer can execute the computer program. The computer program in the above-mentioned computer-readable storage medium can be run in an environment deployed in a computer device such as a client, a host, an agent device, a server, etc. In addition, in one example, the computer program and any associated data, data files and data structures are distributed on a networked computer system, so that the computer program and any associated data, data files and data structures are stored, accessed and executed in a distributed manner by one or more processors or computers.

[0183] According to an exemplary embodiment of the present disclosure, a computer program product may also be provided, including a computer program, and when the computer program is executed by a processor, the target recognition method according to the present disclosure is implemented.

[0184] According to the target recognition method, device, electronic device and storage medium disclosed in the present invention, when performing target detection, the detection data of multiple payloads for the same target are taken into account at the same time, that is, the fusion information can be fully utilized to make the considered information richer and more comprehensive. Compared with the method of using only a single system payload for target detection, the "fusion" recognition method provided by the present invention has higher recognition accuracy. In addition, the method disclosed in the present invention has a strong adaptability to the number and type of payloads involved in the fusion, can adapt to loosely coupled intersatellite links and situations where individual payload data is missing, and has good robustness and anti-interference performance for target recognition.

[0185] According to an exemplary embodiment of the present disclosure, obtaining the matching degree between the detected attribute features and the preset attribute features is equivalent to re-testing the identification result of a single load, and the test result can be added to the identity discrimination basis, that is, the test result can be added to the type discrimination basis. In this way, when the load capacity is uncertain, its identification result can be properly corrected, thereby ensuring the accuracy of the identity classification decision.

[0186] According to an exemplary embodiment of the present disclosure, when determining the type of a target, the determination results of various loads can be fused as a basis for determination, that is, the fused information can be fully utilized to make the considered information richer and more comprehensive, thereby ensuring the reliability and accuracy of determining the target type.

[0187] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. The present disclosure is intended to cover any variations, uses or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The description and examples are to be considered exemplary only, and the true scope and spirit of the present disclosure are indicated by the following claims.

[0188] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A target recognition method, characterized in that: include: Acquire detection data of each of the multiple payloads for a first target, wherein the detection data includes detection confidence, recognition confidence, and detection attribute characteristics, wherein the detection confidence is used to characterize the possibility that the first target actually exists, and the recognition confidence is used to characterize the possibility that the type of the first target is predicted to be each of the multiple preset types, and the payload includes a satellite payload; Based on the detection confidence and load credibility corresponding to each of the multiple loads, determining whether the first target actually exists, wherein the load credibility is used to characterize the credibility of the detection capability of the corresponding load; In the case where it is determined that the first target does exist, for each of the multiple loads, obtaining a matching degree between a detection attribute feature corresponding to the load and a preset attribute feature corresponding to each of the multiple preset types; Determining the type of the first target based on the recognition confidence and matching degree corresponding to each of the multiple payloads; The determining whether the first target actually exists based on the detection confidence and the load credibility corresponding to each of the multiple loads includes: Based on the detection confidence and load credibility corresponding to each of the multiple loads, calculating a single load existence probability function corresponding to each of the multiple loads, wherein the single load existence probability function corresponding to each load is used to indicate the probability that the first target predicted by the load does exist, does not exist, or cannot be determined to exist; Based on the single load existence probability function corresponding to each load, a mixed existence probability function is calculated, wherein the mixed existence probability function is used to characterize the total probability that the first target predicted by the multiple loads does exist, the total probability that it does not exist, or the total probability that it is impossible to determine whether it exists; Based on the mixed existence probability function, determining whether the first target actually exists; The determining the type of the first target based on the recognition confidence and the matching degree corresponding to each of the multiple payloads includes: Based on the recognition confidence and matching degree corresponding to each load in the multiple loads, a single load type probability function corresponding to each load is obtained, wherein the single load type probability function corresponding to each load is used to indicate the probability that the first target predicted by the load belongs to each preset type in the multiple preset types or the probability that the type of the first target cannot be determined; Based on the single load type probability function corresponding to each load, a mixed type probability function is calculated, wherein the mixed type probability function is used to characterize the probability that the first target predicted by the multiple loads belongs to each preset type of the multiple preset types or the probability that the type of the first target cannot be determined; Based on the mixed type probability function, a type of the first target is determined.

2. The target recognition method according to claim 1, characterized in that: The single load existence probability function is expressed by the following formula: in, represents the probability that the first target predicted by the i-th load among the multiple loads actually exists, represents the probability that the first target predicted by the i-th load does not exist, represents the probability that the first target cannot be determined to exist as predicted by the i-th load, represents the detection confidence corresponding to the i-th load, represents the load credibility of the i-th load.

3. The target recognition method according to claim 2, characterized in that: The mixed existence probability function is expressed by the following formula: in, represents the total probability that the first target actually exists, predicted by the multiple payloads. represents the total probability that the first target does not exist as predicted by the multiple payloads, represents the total probability of being unable to determine whether the first target exists, predicted by the multiple loads, n is the total number of the multiple loads, The value of is 1, 0 or -1. There is a conflict coefficient.

4. The target recognition method according to claim 1, characterized in that: The single load type probability function is obtained by the following formula: in, represents the probability that the i-th load among the multiple loads predicts the type of the first target to be the j-th preset type among the multiple preset types, represents the probability that the i-th load cannot determine the type of the first target, the recognition confidence of predicting the type of the first target as the j-th preset type for the i-th load, is the matching degree between the detection attribute feature corresponding to the i-th load and the preset attribute feature corresponding to the j-th preset type, is the distance between the detection attribute feature of the i-th load and the preset attribute feature corresponding to the j-th preset type, represents the jth preset type among the multiple preset types.

5. The target recognition method according to claim 4, characterized in that: The mixed type probability function is obtained by the following formula: in, represents the probability that the first target predicted by the multiple loads belongs to the jth preset type among the multiple preset types, represents the probability that the type of the first target cannot be determined, as predicted by the multiple loads, n is the total number of the multiple loads, is the type conflict coefficient.

6. A target recognition device, characterized in that: include: a detection data acquisition module, configured to acquire detection data of each of the multiple payloads for a first target, wherein the detection data includes detection confidence, recognition confidence, and detection attribute characteristics, wherein the detection confidence is used to characterize the possibility that the first target actually exists, and the recognition confidence is used to characterize the possibility that the type of the first target is predicted to be each of the multiple preset types, and the payload includes a satellite payload; an existence judgment module, configured to judge whether the first target actually exists based on the detection confidence and the load credibility corresponding to each of the multiple loads, wherein the load credibility is used to characterize the credibility of the detection capability of the corresponding load; a matching degree acquisition module configured to, when it is determined that the first target does exist, acquire, for each of the multiple loads, a matching degree between a detection attribute feature corresponding to the load and a preset attribute feature corresponding to each of the multiple preset types; a type determination module, configured to determine the type of the first target based on the recognition confidence and matching degree corresponding to each of the multiple payloads; Wherein, the existence judgment module is specifically configured as follows: Based on the detection confidence and load credibility corresponding to each of the multiple loads, calculating a single load existence probability function corresponding to each of the multiple loads, wherein the single load existence probability function corresponding to each load is used to indicate the probability that the first target predicted by the load does exist, does not exist, or cannot be determined to exist; Based on the single load existence probability function corresponding to each load, a mixed existence probability function is calculated, wherein the mixed existence probability function is used to characterize the total probability that the first target predicted by the multiple loads does exist, the total probability that it does not exist, or the total probability that it is impossible to determine whether it exists; Based on the mixed existence probability function, determining whether the first target actually exists; Wherein, the type determination module is specifically configured as follows: Based on the recognition confidence and matching degree corresponding to each load in the multiple loads, a single load type probability function corresponding to each load is obtained, wherein the single load type probability function corresponding to each load is used to indicate the probability that the first target predicted by the load belongs to each preset type in the multiple preset types or the probability that the type of the first target cannot be determined; Based on the single load type probability function corresponding to each load, a mixed type probability function is calculated, wherein the mixed type probability function is used to characterize the probability that the first target predicted by the multiple loads belongs to each preset type of the multiple preset types or the probability that the type of the first target cannot be determined; Based on the mixed type probability function, a type of the first target is determined.

7. An electronic device, characterized in that: include: processor; a memory for storing instructions executable by the processor; The processor is configured to execute the instructions to implement the target recognition method as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that: When the instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute the target recognition method as claimed in any one of claims 1 to 5.

Images