Target characteristic identification method and device based on clustering analysis

Abstract

The invention relates to the technical field of image recognition, in particular to a target characteristic recognition method and device based on clustering analysis, and the method comprises the steps: collecting candidate unmanned aerial vehicle images; inputting the candidate unmanned aerial vehicle images into a deep learning network, and classifying the candidate unmanned aerial vehicle images to obtain a frequency spectrum and a panoramic unmanned aerial vehicle image; importing the initial spectrum and initial panoramic feature information base to obtain candidate spectrum and candidate panoramic feature information, and generating a fusion matching feature information base; and when a candidate airspace real-time state image is obtained, inputting the candidate airspace real-time state image into the information base to enable the fusion matching feature information base to process the candidate airspace real-time state image by using the candidate frequency spectrum to obtain a first record, and processing the candidate airspace real-time state image through the panoramic feature information base to obtain a second record. Based on the common characteristics of the characteristic signals of the unmanned aerial vehicle, the clustering analysis method is applied to the detection and recognition of the characteristic signals of the unmanned aerial vehicle, and the detection, discovery and recognition efficiency of the unmanned aerial vehicle can be remarkably improved.

Description

Technical Field

[0001] The present application relates to the technical field of image recognition, and in particular to a method and device for identifying target characteristics based on cluster analysis. Background Art

[0002] Drones can generally be divided into two categories, “cooperative” and “non-cooperative”, depending on whether there is communication between the drone target and the detection equipment.

[0003] Non-cooperative intrusion flight incidents of drones in low-altitude airspace have become common both at home and abroad, which has seriously restricted the development of drone industrialization and posed a huge threat to public safety. Therefore, it is urgent to use video image information to effectively detect "low, slow and small" targets such as low-altitude non-cooperative drones in order to achieve subsequent protection. Summary of the invention

[0004] To achieve the above objectives, this application provides the following technical solutions:

[0005] According to a first aspect of the present invention, the present invention claims a method for identifying target characteristics based on cluster analysis, the method comprising:

[0006] Collect candidate drone images;

[0007] Input the candidate drone image into a preconfigured deep learning network, so that the preconfigured deep learning network classifies the candidate drone image to obtain a spectrum drone image and a panoramic drone image; wherein the preconfigured deep learning network is obtained by configuring the deep learning network to be configured using a drone image reference set; the drone image reference set includes at least a historical spectrum drone image and a historical panoramic drone image;

[0008] Importing the spectrum drone image into the initial spectrum to obtain a candidate spectrum, and importing the panoramic drone image into the initial panoramic feature information library to obtain candidate panoramic feature information;

[0009] generating a fusion matching feature information library according to the candidate spectrum and the candidate panoramic feature information;

[0010] When a candidate airspace real-time status image transmitted by a requesting person is obtained, the candidate airspace real-time status image is input into the fusion matching feature information library, so that the fusion matching feature information library processes the candidate airspace real-time status image using the candidate spectrum to obtain at least one first record, and processes the candidate airspace real-time status image through the panoramic feature information library to obtain at least one second record;

[0011] The at least one first record and the at least one second record are output.

[0012] Furthermore, before outputting the at least one first record and the at least one second record, the method further includes:

[0013] For each of the first records, calculating the consistency between the first record and the candidate airspace real-time status image;

[0014] For each of the second records, calculating the consistency between the second record and the candidate airspace real-time status image;

[0015] According to the consistency of each of the first records and the consistency of each of the second records, priority deployment is performed on each of the first records and the second records from large to small to obtain a priority deployment record;

[0016] Wherein, outputting the at least one first record and the at least one second record comprises:

[0017] The priority deployment record is output, where the priority deployment record includes each of the first records and each of the second records after priority deployment from large to small according to consistency.

[0018] Furthermore, generating a fusion matching feature information library according to the candidate spectrum and the candidate panoramic feature information includes:

[0019] Encapsulating the candidate spectrum and the candidate panoramic feature information through a matching gateway application service to generate a fusion matching feature information library, wherein the fusion matching feature information library includes a matching fusion export;

[0020] Wherein, the outputting the priority deployment record includes:

[0021] The priority deployment record is outputted through the matching and fusion export.

[0022] Furthermore, configuring the deep learning network to be configured using the drone image reference set to obtain the preconfigured deep learning network includes:

[0023] Collecting a reference set of drone images; wherein the reference set of drone images includes historical spectral drone images and their reference labels, and historical panoramic drone images and their reference labels;

[0024] Inputting each of the drone image references into the deep learning network to be configured in sequence, so that the deep learning network to be configured processes each of the drone image references respectively, and taking the obtained processing records close to the reference labels corresponding to each of the drone image references as configuration candidates, iteratively configuring the deep learning network to be configured until the deep learning network to be configured reaches convergence, thereby obtaining a deep learning network;

[0025] The drone image reference is the historical spectrum drone image, or the historical panoramic drone image.

[0026] According to a second aspect of the present invention, the present invention claims protection for a target characteristic recognition device based on cluster analysis, characterized in that the device comprises:

[0027] A candidate drone image acquisition module, used to acquire candidate drone images;

[0028] A classification processing module, used for inputting the candidate drone image into a preconfigured deep learning network, so that the preconfigured deep learning network classifies the candidate drone image to obtain a spectrum drone image and a panoramic drone image; wherein the preconfigured deep learning network is configured based on a configuration module;

[0029] An import module, used for importing the spectrum drone image into the initial spectrum to obtain a candidate spectrum, and importing the panoramic drone image into the initial panoramic feature information library to obtain candidate panoramic feature information;

[0030] A generating module, used for generating a fusion matching feature information library according to the candidate spectrum and the candidate panoramic feature information;

[0031] an airspace real-time status image processing module, configured to, when acquiring a candidate airspace real-time status image transmitted by a requesting person, input the candidate airspace real-time status image into the fusion matching feature information library, so that the fusion matching feature information library processes the candidate airspace real-time status image using a candidate spectrum to obtain at least one first record, and processes the candidate airspace real-time status image using the panoramic feature information library to obtain at least one second record;

[0032] An output module is used to output the at least one first record and the at least one second record.

[0033] Furthermore, the device also includes:

[0034] A first calculation module, configured to calculate, for each of the first records, the consistency between the first record and the candidate airspace real-time status image;

[0035] A second calculation module, configured to calculate, for each of the second records, the consistency between the second record and the candidate airspace real-time status image;

[0036] A priority deployment module, used to perform priority deployment on each of the first record and the second record from large to small according to the consistency of each of the first record and the consistency of each of the second record, so as to obtain a priority deployment record;

[0037] The output module is specifically used to output a priority deployment record, wherein the priority deployment record includes each of the first records and each of the second records after priority deployment in descending order according to consistency.

[0038] Furthermore, the generation module includes:

[0039] A generating submodule, used for encapsulating the candidate spectrum and the candidate panoramic feature information through a matching gateway application service to generate a fused matching feature information library, wherein the fused matching feature information library includes a matching fusion export;

[0040] Wherein, the output module is further used to output the priority deployment record through the matching fusion export.

[0041] Furthermore, the configuration module includes:

[0042] A drone image acquisition module is used to acquire a drone image reference set; wherein the drone image reference set includes historical spectral drone images and their reference labels, and historical panoramic drone images and their reference labels;

[0043] an iterative configuration module, for sequentially inputting each of the drone image references into the deep learning network to be configured, so that the deep learning network to be configured processes each of the drone image references respectively, and takes the obtained processing records close to the reference labels corresponding to each of the drone image references as configuration candidates, and iteratively configuring the deep learning network to be configured until the deep learning network to be configured reaches convergence, thereby obtaining a deep learning network;

[0044] The drone image reference is the historical spectrum drone image, or the historical panoramic drone image.

[0045] The present application relates to the field of image recognition technology, and in particular to a method and device for identifying target characteristics based on cluster analysis, which includes collecting candidate drone images; inputting candidate drone images into a deep learning network, classifying candidate drone images to obtain spectrum and panoramic drone images; importing the initial spectrum and initial panoramic feature information library to obtain candidate spectrum and candidate panoramic feature information, and generating a fusion matching feature information library; when obtaining a candidate airspace real-time status image, input the information library so that the fusion matching feature information library processes the candidate airspace real-time status image using the candidate spectrum to obtain a first record, and processes the candidate airspace real-time status image through the panoramic feature information library to obtain a second record. Based on the common characteristics of unmanned aerial vehicle characteristic signals, the present invention applies cluster analysis methods to unmanned aerial vehicle characteristic signal detection and identification, which can significantly improve the efficiency of detection, discovery and identification of unmanned aerial vehicles. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 A schematic diagram of a flow chart of a target characteristic identification method based on cluster analysis claimed in an embodiment of the present application;

[0047] Figure 2 A schematic diagram of the structure of a target characteristic recognition device based on cluster analysis as claimed in an embodiment of the present application. DETAILED DESCRIPTION

[0048] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0049] The terms "first", "second" and "third" in this application are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features. Thus, the features defined as "first", "second" and "third" can explicitly or implicitly include at least one of the features. In the description of this application, the meaning of "multiple" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined. All directional indications in the embodiments of the present application (such as up, down, left, right, front, back...) are only used to explain the relative position relationship, movement, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication also changes accordingly. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions. For example, a process, method, device, product or equipment that includes a series of steps or modules is not limited to the listed steps or modules, but optionally also includes steps or modules that are not listed, or optionally also includes other steps or modules inherent to these processes, methods, products or equipment.

[0050] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0051] The target characteristic recognition method based on cluster analysis provided by the present invention can be used in the field of airspace monitoring or other fields. The above is only an example and does not limit the application field of the target characteristic recognition method based on cluster analysis provided by the present invention.

[0052] Cluster analysis, also known as clustering, is one of the most important techniques in unsupervised learning. The purpose of cluster analysis is to find clusters composed of similar samples in the data, and it has been developed in machine learning, pattern recognition, data mining, statistics, computer science, biomedicine, social sciences, economics and other fields.

[0053] In the military field, target feature recognition is the key to the combat effectiveness of weapon systems such as early warning, interception, reconnaissance, and long-range strike. With the widespread application of various types of small drones in the civil and military fields, various threats and challenges brought by them have also followed. Therefore, the effective detection and identification of small and medium-sized drones in low-altitude areas has become one of the hot topics in the field of target recognition. Small drones have the characteristics of small size and complex flight environment. For such low-altitude, slow, and weak targets, traditional technical means are difficult to detect, let alone accurately identify and locate them.

[0054] See also Figure 1 , shows a schematic flow chart of a target characteristic recognition method based on cluster analysis provided by an embodiment of the present invention, and the target characteristic recognition method based on cluster analysis specifically comprises the following steps:

[0055] S101: Collect candidate drone images.

[0056] In the specific process of executing step S101, candidate drone images that need to be classified are collected, wherein the candidate drone images include spectrum drone images and panoramic drone images.

[0057] It should be noted that the spectrum drone image includes information with certain logical relationships, for example, information including level, requesting person, and signal width, and the level, requesting person, and signal width in the information have certain logical relationships.

[0058] S102: Input the candidate drone images into a preconfigured deep learning network, so that the preconfigured deep learning network classifies the candidate drone images to obtain spectral drone images and panoramic drone images.

[0059] In an embodiment of the present application, a deep learning network is pre-configured, and the pre-configured deep learning network is obtained by configuring the deep learning network to be configured using a drone image reference set. The drone image reference set includes historical spectral drone images and their reference labels, and historical panoramic drone images and their reference labels.

[0060] Optionally, the process of using the drone image reference set to configure the deep learning network to obtain a preconfigured deep learning network is specifically as follows: collect the drone image reference set, input each drone image reference into the deep learning network to be configured in turn, so that the deep learning network to be configured processes each drone image reference respectively, and use the processing record of each drone image reference and the reference label corresponding to each drone image reference to construct a corresponding loss function, and use each loss function to iteratively configure the deep learning network to be configured until the deep learning network to be configured converges to obtain a deep learning network.

[0061] The drone image reference is a historical spectrum drone image or a historical panoramic drone image. When the drone image reference is a historical spectrum drone image, its corresponding reference label is spectrum; correspondingly, when the drone image reference is a historical panoramic drone image, its corresponding reference label is panoramic.

[0062] In the specific process of executing step S102, after the deep learning network is configured, the collected candidate drone images are input into the preconfigured deep learning network, so that the preconfigured deep learning network can classify the input candidate drone images to obtain each drone image and its corresponding label, that is, obtain a spectral drone image and a panoramic drone image.

[0063] S103: Importing the spectrum drone image into the initial spectrum to obtain a candidate spectrum, and importing the panoramic drone image into the initial panoramic feature information library to obtain candidate panoramic feature information.

[0064] In the specific execution of step S103, after the candidate drone images are classified by using the preconfigured deep learning network to obtain the spectrum drone images and the panoramic drone images, the spectrum drone images are processed by a pre-set logic program to obtain structured information and unstructured information; the corresponding trajectory path information is extracted from the unstructured information by using the pre-configured extraction network; the corresponding trajectory path relationship and feature information are extracted from the structured information by using the pre-set signal width rule; the trajectory path relationship and feature information, as well as the trajectory path information are imported into the initial spectrum by a pre-set batch import program to obtain the spectrum.

[0065] For panoramic drone images, the panoramic drone images are sorted to obtain multiple key-value pair information and text information, and the key-value pair information and text information are imported into the initial panoramic feature information library to obtain candidate panoramic feature information. Among them, the structure of the key-value pair information is structured, and there is no association relationship between such key-value pair information.

[0066] It should be noted that the initial panoramic feature information database may be a relational information database.

[0067] In the embodiment of the present application, the preconfigured extraction network is configured using historical signal width text information, wherein the structure of the historical signal width text information is unstructured.

[0068] S104: Generate a fusion matching feature information library according to the candidate spectrum and the candidate panoramic feature information.

[0069] In the embodiment of the present application, technicians can write corresponding matching application services through packaging programming experience.

[0070] In the specific process of executing step S103, after the obtained spectrum drone image is imported into the initial spectrum to obtain the candidate spectrum, and the obtained panoramic drone image is imported into the initial panoramic feature information library to obtain the candidate panoramic feature information, the candidate spectrum and the candidate panoramic feature information are encapsulated through the pre-written matching gateway application service to generate a fused matching feature information library with a matching export.

[0071] It should be noted that the matching fusion export is used to output the answer corresponding to the spatial real-time status image, that is, the record.

[0072] S105: When the candidate airspace real-time status image transmitted by the requesting person is obtained, the candidate airspace real-time status image is input into the fusion matching feature information library, so that the fusion matching feature information library processes the candidate airspace real-time status image using the candidate spectrum to obtain at least one first record, and processes the candidate airspace real-time status image through the panoramic feature information library to obtain at least one second record.

[0073] In the specific process of executing step S105, after generating a fusion matching feature information library based on the candidate spectrum and the candidate panoramic feature information, it is detected in real time whether the candidate airspace real-time status image transmitted by the requesting person is obtained. When the candidate airspace real-time status image transmitted by the requesting person is obtained, the candidate airspace real-time status image is input into the fusion matching feature information library, so that the fusion matching feature information library matches each record corresponding to the candidate airspace real-time status image from the spectrum drone image stored in itself through the candidate spectrum (for the convenience of interval, the record corresponding to the candidate airspace real-time status image matched by the candidate spectrum is called the first record).

[0074] And the candidate airspace real-time status image is analyzed through the panoramic feature information library, and combined with the contextual relationship, each record corresponding to the candidate airspace real-time status image is matched from the panoramic drone images stored in itself (for the convenience of interval, the record corresponding to the candidate airspace real-time status image matched with the candidate panoramic feature information is called the second record).

[0075] S106: Output each first record and each second record.

[0076] In the specific process of executing step S106, each first record and each second record may be output through a matching fusion export.

[0077] Optionally, before outputting each first record and each second record through the matching fusion export, the embodiment of the present application can also calculate the consistency between each first record and the candidate airspace real-time status image by calling the secondary priority deployment algorithm, and calculate the consistency between each second record and the candidate airspace real-time status image. After that, according to the consistency of each first record and the consistency of each second record, priority deployment is performed on each first record and second record from large to small to obtain a priority deployment record, and then the priority deployment record is output through the matching fusion export.

[0078] The present invention provides a target characteristic recognition method based on cluster analysis. First, a deep learning network to be configured can be configured by using a drone image reference set to obtain a deep learning network. After a candidate drone image is collected, the collected candidate drone image can be input into a preconfigured deep learning network, so that the preconfigured deep learning network classifies the candidate drone image to obtain a spectrum drone image and a panoramic drone image; the spectrum drone image is imported into an initial spectrum to obtain a candidate spectrum, and the panoramic drone image is imported into an initial panoramic feature information library to obtain candidate panoramic feature information; a fusion matching feature information library is generated according to the candidate spectrum and the candidate panoramic feature information; when a candidate airspace real-time status image transmitted by a requesting person is obtained, the candidate airspace real-time status image is input into a fusion matching feature information library, so that the fusion matching feature information library processes the candidate airspace real-time status image using the candidate spectrum to obtain at least one first record, and processes the candidate airspace real-time status image through the panoramic feature information library to obtain at least one second record; and at least one first record and at least one second record are output. The technical solution provided by the present invention can first classify the candidate drone images, and then classify and store the drone images of different types, so that the drone images can be stored in a suitable matching feature information library in an optimal mode, which can not only improve the matching accuracy of each matching feature information library, but also realize the fusion management of drone images. And according to the candidate spectrum and candidate panoramic feature information, the generated fusion matching feature information library can be integrated to provide matching services externally, and the same matching feature information library can realize the matching of different types of drone images.

[0079] Corresponding to the target characteristic recognition method based on cluster analysis disclosed in the above embodiment of the present invention, reference is made to Figure 2 The embodiment of the present invention further provides a schematic diagram of the structure of a target characteristic recognition device based on cluster analysis, and the target characteristic recognition device based on cluster analysis includes:

[0080] A candidate drone image acquisition module 21 is used to acquire candidate drone images;

[0081] The classification processing module 22 is used to input the candidate drone image into a preconfigured deep learning network so that the preconfigured deep learning network classifies the candidate drone image to obtain a spectrum drone image and a panoramic drone image; wherein the preconfigured deep learning network is configured based on the configuration module;

[0082] An import module 23 is used to import the spectrum drone image into the initial spectrum to obtain a candidate spectrum, and import the panoramic drone image into the initial panoramic feature information library to obtain candidate panoramic feature information;

[0083] A generating module 24, used to generate a fusion matching feature information library according to the candidate spectrum and the candidate panoramic feature information;

[0084] The airspace real-time status image processing module 25 is used to, when obtaining the candidate airspace real-time status image transmitted by the requesting person, input the candidate airspace real-time status image into the fusion matching feature information library, so that the fusion matching feature information library processes the candidate airspace real-time status image using the candidate spectrum to obtain at least one first record, and processes the candidate airspace real-time status image through the panoramic feature information library to obtain at least one second record;

[0085] The output module 26 is used to output at least one first record and at least one second record.

[0086] The specific principles and execution processes of each module in the target characteristic recognition device based on cluster analysis disclosed in the above embodiment of the present invention are similar to those in the above embodiment of the present invention. Figure 1 The disclosed target characteristic identification method based on cluster analysis is the same as that disclosed in the embodiment of the present invention. Figure 1 The corresponding parts of the disclosed target feature recognition method based on cluster analysis will not be described in detail here.

[0087] The present invention provides a target characteristic recognition device based on cluster analysis. First, a deep learning network to be configured can be configured by using a drone image reference set to obtain a deep learning network. After a candidate drone image is collected, the collected candidate drone image can be input into a preconfigured deep learning network, so that the preconfigured deep learning network classifies the candidate drone image to obtain a spectrum drone image and a panoramic drone image; the spectrum drone image is imported into an initial spectrum to obtain a candidate spectrum, and the panoramic drone image is imported into an initial panoramic feature information library to obtain candidate panoramic feature information; a fusion matching feature information library is generated according to the candidate spectrum and the candidate panoramic feature information; when a candidate airspace real-time status image transmitted by a requesting person is obtained, the candidate airspace real-time status image is input into the fusion matching feature information library, so that the fusion matching feature information library processes the candidate airspace real-time status image using the candidate spectrum to obtain at least one first record, and processes the candidate airspace real-time status image through the panoramic feature information library to obtain at least one second record; and at least one first record and at least one second record are output. The technical solution provided by the present invention can first classify the candidate drone images, and then classify and store the drone images of different types, so that the drone images can be stored in a suitable matching feature information library in an optimal mode, which can not only improve the matching accuracy of each matching feature information library, but also realize the fusion management of drone images. And according to the candidate spectrum and candidate panoramic feature information, the generated fusion matching feature information library can be integrated to provide matching services externally, and the same matching feature information library can realize the matching of different types of drone images.

[0088] Furthermore, the target characteristic recognition device based on cluster analysis provided by the embodiment of the present invention further includes:

[0089] A first calculation module, used for calculating, for each first record, the consistency between the first record and the candidate airspace real-time status image;

[0090] A second calculation module is used to calculate the consistency between each second record and the candidate airspace real-time status image;

[0091] A priority deployment module, used to perform priority deployment on each first record and each second record from large to small according to the consistency of each first record and the consistency of each second record, to obtain a priority deployment record;

[0092] Among them, the output module is specifically used to output the priority deployment record, and the priority deployment record includes each first record and each second record after priority deployment from large to small according to consistency.

[0093] Optionally, generate modules, including:

[0094] A generation submodule, used for encapsulating the candidate spectrum and the candidate panoramic feature information through the matching gateway application service, and generating a fusion matching feature information library, wherein the fusion matching feature information library includes a matching fusion export;

[0095] Among them, the output module is also used to output the priority deployment record by matching the fusion export.

[0096] Optional, configuration modules, including:

[0097] The drone image acquisition module is used to acquire a drone image reference set; wherein the drone image reference set includes historical spectral drone images and their reference labels, and historical panoramic drone images and their reference labels;

[0098] An iterative configuration module is used to sequentially input each drone image reference into the deep learning network to be configured, so that the deep learning network to be configured processes each drone image reference respectively, and takes the obtained processing record close to the reference label corresponding to each drone image reference as the configuration candidate, and iteratively configures the deep learning network to be configured until the deep learning network to be configured reaches convergence, thereby obtaining a deep learning network;

[0099] The drone image reference is a historical spectrum drone image, or a historical panoramic drone image reference.

[0100] In the several embodiments provided in the present application, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of modules is only a logical function division, and there may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or modules, which can be electrical, mechanical or other forms.

[0101] In addition, each functional module in each embodiment of the present application can be integrated into a processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The above integrated modules can be implemented in the form of hardware or in the form of software functional modules. The above is only an implementation method of the present application, and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the specification and drawings of this application, or directly or indirectly used in other related technical fields, are also included in the patent protection scope of the present application.

[0102] The specific implementation methods of the invention are described in detail above, but they are only examples, and the present application is not limited to the specific implementation methods described above. For those skilled in the art, any equivalent modification or substitution of the invention is also within the scope of the present application, and therefore, the equalization, modification, and improvement made without departing from the spirit and principle of the present application should be included in the scope of the present application.

Claims

1. A target feature recognition method based on cluster analysis, characterized in that: The method comprises: Collect candidate drone images; Input the candidate drone image into a preconfigured deep learning network, so that the preconfigured deep learning network classifies the candidate drone image to obtain a spectrum drone image and a panoramic drone image; wherein the preconfigured deep learning network is obtained by configuring the deep learning network to be configured using a drone image reference set; the drone image reference set includes at least a historical spectrum drone image and a historical panoramic drone image; Importing the spectrum drone image into the initial spectrum to obtain a candidate spectrum, and importing the panoramic drone image into the initial panoramic feature information library to obtain candidate panoramic feature information; generating a fusion matching feature information library according to the candidate spectrum and the candidate panoramic feature information; When a candidate airspace real-time status image transmitted by a requesting person is obtained, the candidate airspace real-time status image is input into the fusion matching feature information library, so that the fusion matching feature information library processes the candidate airspace real-time status image using the candidate spectrum to obtain at least one first record, and processes the candidate airspace real-time status image through the panoramic feature information library to obtain at least one second record; The at least one first record and the at least one second record are output.

2. The method according to claim 1, characterized in that Before outputting the at least one first record and the at least one second record, the method further includes: For each of the first records, calculating the consistency between the first record and the candidate airspace real-time status image; For each of the second records, calculating the consistency between the second record and the candidate airspace real-time status image; According to the consistency of each of the first records and the consistency of each of the second records, priority deployment is performed on each of the first records and the second records from large to small to obtain a priority deployment record; Wherein, outputting the at least one first record and the at least one second record comprises: The priority deployment record is output, where the priority deployment record includes each of the first records and each of the second records after priority deployment from large to small according to consistency.

3. The method according to claim 2, characterized in that The step of generating a fusion matching feature information library according to the candidate spectrum and the candidate panoramic feature information includes: Encapsulating the candidate spectrum and the candidate panoramic feature information through a matching gateway application service to generate a fused matching feature information library, wherein the fused matching feature information library includes a matching fusion export; Wherein, the outputting the priority deployment record includes: The priority deployment record is outputted through the matching and fusion export.

4. The method according to claim 1, characterized in that: The method of configuring the deep learning network to be configured using the drone image reference set to obtain the preconfigured deep learning network includes: Collecting a reference set of drone images; wherein the reference set of drone images includes historical spectral drone images and their reference labels, and historical panoramic drone images and their reference labels; Inputting each of the drone image references into the deep learning network to be configured in sequence, so that the deep learning network to be configured processes each of the drone image references respectively, and taking the obtained processing records close to the reference labels corresponding to each of the drone image references as configuration candidates, iteratively configuring the deep learning network to be configured until the deep learning network to be configured reaches convergence, thereby obtaining a deep learning network; The drone image reference is the historical spectrum drone image, or the historical panoramic drone image.

5. A target characteristic recognition device based on cluster analysis, characterized in that: The device comprises: A candidate drone image acquisition module, used to acquire candidate drone images; A classification processing module, used for inputting the candidate drone image into a preconfigured deep learning network, so that the preconfigured deep learning network classifies the candidate drone image to obtain a spectrum drone image and a panoramic drone image; wherein the preconfigured deep learning network is configured based on a configuration module; An import module, used for importing the spectrum drone image into the initial spectrum to obtain a candidate spectrum, and importing the panoramic drone image into the initial panoramic feature information library to obtain candidate panoramic feature information; A generating module, used for generating a fusion matching feature information library according to the candidate spectrum and the candidate panoramic feature information; an airspace real-time status image processing module, configured to, when acquiring a candidate airspace real-time status image transmitted by a requesting person, input the candidate airspace real-time status image into the fusion matching feature information library, so that the fusion matching feature information library processes the candidate airspace real-time status image using a candidate spectrum to obtain at least one first record, and processes the candidate airspace real-time status image using the panoramic feature information library to obtain at least one second record; An output module is used to output the at least one first record and the at least one second record.

6. The device according to claim 5, characterized in that The device also includes: A first calculation module, configured to calculate, for each of the first records, the consistency between the first record and the candidate airspace real-time status image; A second calculation module, configured to calculate, for each of the second records, the consistency between the second record and the candidate airspace real-time status image; A priority deployment module, used to perform priority deployment on each of the first record and the second record from large to small according to the consistency of each of the first record and the consistency of each of the second record, so as to obtain a priority deployment record; The output module is specifically used to output a priority deployment record, wherein the priority deployment record includes each of the first records and each of the second records after priority deployment in descending order according to consistency.

7. The device according to claim 6, characterized in that The generating module comprises: A generating submodule, used for encapsulating the candidate spectrum and the candidate panoramic feature information through a matching gateway application service to generate a fused matching feature information library, wherein the fused matching feature information library includes a matching fusion export; Wherein, the output module is further used to output the priority deployment record through the matching fusion export.

8. The device according to claim 5, characterized in that The configuration module includes: A drone image acquisition module is used to acquire a drone image reference set; wherein the drone image reference set includes historical spectral drone images and their reference labels, and historical panoramic drone images and their reference labels; an iterative configuration module, for sequentially inputting each of the drone image references into the deep learning network to be configured, so that the deep learning network to be configured processes each of the drone image references respectively, and takes the obtained processing records close to the reference labels corresponding to each of the drone image references as configuration candidates, and iteratively configuring the deep learning network to be configured until the deep learning network to be configured reaches convergence, thereby obtaining a deep learning network; The drone image reference is the historical spectrum drone image, or the historical panoramic drone image.

Images