Network asset attribute labeling method

Abstract

The invention discloses a network asset attribute labeling method. The method comprises the following steps: acquiring a plurality of pieces of target image information of a network asset target site in a preset period; processing the asset image based on a preset processing strategy to obtain a sample image; creating a sample set based on all the sample images; creating a cluster center point set which is a null set in an initial state, performing clustering based on a preset clustering strategy to obtain a representative image set, and then obtaining a cluster to which each sample image in the sample set belongs based on a similarity strategy; the associated labeled label information is acquired; labeling the label in the labeled label information as an asset attribute label corresponding to the associated sample image; and analyzing the labeled label information, and calling a corresponding labeling strategy to label the asset attributes corresponding to the unassociated sample images. According to the method, labeling requirements of different scales are met, and high-efficiency, low-cost and high-accuracy network asset attribute labeling can be realized.

Description

Technical Field

[0001] The present disclosure relates to the technical field of network asset management, and in particular to a network asset attribute labeling method. Background Art

[0002] Network assets are the infrastructure that supports the operation of various information systems. The accurate labeling of network asset attributes, as an important means to improve asset management efficiency and strengthen network security defenses, has received widespread attention in the industry.

[0003] In the prior art, manual labeling is usually used to label network asset attributes. This method can barely cope with the collection and organization of network asset attributes in small-scale and relatively simple network environments. However, as the scale of the network continues to expand and the complexity increases, more manual participation is required, resulting in low labeling efficiency and high cost. It is also unable to guarantee the accuracy of large-scale requirements and cannot meet the dual needs of modern network asset management for timeliness and economy. Summary of the invention

[0004] In view of this, the embodiments of the present disclosure provide a network asset attribute labeling method, which can solve the problems of low efficiency and high cost of manual labeling in the prior art and the inability to meet different network asset management requirements.

[0005] In a first aspect, an embodiment of the present disclosure provides a method for labeling network asset attributes, the method specifically comprising the following schemes:

[0006] Collecting a number of target image information of a network asset target site within a preset period, each of which includes an asset image and a corresponding image file number and a network asset identifier;

[0007] Processing the asset image based on a preset processing strategy to obtain a sample image;

[0008] Based on all the sample images, a sample set is created; a cluster center point set is created, and the cluster center point set is an empty set in an initial state;

[0009] Clustering the sample set and the cluster center point set based on a preset clustering strategy to obtain a representative image set;

[0010] Based on the similarity strategy, the representative image set, and the preset cluster set information, obtaining the cluster to which each sample image in the sample set belongs;

[0011] Based on the network asset identifier corresponding to each of the sample images, the associated annotated label information is obtained in a preset annotated network asset information library; wherein the total amount of labels of the obtained annotated label information is less than the total amount of all the sample images;

[0012] The annotated labels in the annotated label information are annotated as asset attribute labels corresponding to the associated sample images; the annotated label information is analyzed, and the corresponding annotation strategy is called to annotate the asset attributes corresponding to the unassociated sample images.

[0013] Optionally, the processing the asset image based on a preset processing strategy to obtain a sample image includes:

[0014] Processing each of the asset images to obtain an initial four-dimensional image data matrix;

[0015] The initial four-dimensional image data matrix includes a red dimension image data matrix, a green dimension image data matrix, a blue dimension image data matrix and a transparency dimension image data matrix;

[0016] All of the initial four-dimensional image data matrices have the same data type and channel order;

[0017] Performing scaling processing on all the initial four-dimensional image data matrices based on a preset scaling algorithm, and recording the processed initial four-dimensional image data matrix as a first image data matrix;

[0018] All of the first image data matrices have the same size;

[0019] Perform multi-level thresholding processing on the first image data matrix to obtain a sample image.

[0020] Optionally, performing multi-level thresholding processing on the first image data matrix to obtain a sample image includes:

[0021] Performing multi-level threshold processing on pixel point values ​​in each dimensional matrix in the first image data matrix to obtain a second image data matrix;

[0022] The multi-level thresholding process includes: d represents the corresponding pixel value in each dimension matrix, d' is the result of multi-level thresholding of the pixel, % is the remainder calculation, and E is the preset threshold parameter;

[0023] A sample image is acquired based on the second image data matrix.

[0024] Optionally, clustering the sample set and the cluster center point set based on a preset clustering strategy to obtain a representative image set includes:

[0025] S410, randomly extracting a sample image from the sample set as a current representative image and storing it in the cluster center point set to obtain an updated cluster center point set;

[0026] S420, obtaining the similarity between the current representative image and each remaining sample image in the sample set, and deleting image objects whose similarity is not less than a preset similarity threshold from the sample set to obtain the updated sample set;

[0027] S430, determining whether the updated sample set is an empty set, and if so, taking the updated cluster center point set as a representative image set;

[0028] If not, a sample image is randomly selected from the updated sample set as the current representative image and stored in the cluster center point set to obtain the updated cluster center point set, and then the process returns to execute S420.

[0029] Optionally, the acquiring the cluster to which each sample image in the sample set belongs based on the similarity strategy, the representative image set, and the preset cluster set information includes:

[0030] Acquire cluster set information based on the representative image set;

[0031] Obtaining the similarity between each sample image in the sample set and all representative images in the representative image set;

[0032] Based on the cluster set information, the cluster corresponding to the current representative image with the highest similarity is obtained, and the cluster is used as the cluster of the corresponding sample image.

[0033] Optionally, analyzing the annotated label information and calling a corresponding annotation strategy to annotate the unassociated sample image with corresponding asset attributes includes:

[0034] Based on the obtained associated labeled label information, determine the labeled label with the most votes of the same type as the target labeled label;

[0035] When the number of votes and the proportion of votes of the target annotation label meet the first preset condition, executing the first annotation strategy;

[0036] When the number of votes and the percentage of votes of the target label do not meet the first preset condition, but meet the second preset condition, execute the second labeling strategy;

[0037] When the number of votes and the proportion of votes of the target annotation label do not meet the first preset condition and do not meet the second preset condition, the third annotation strategy is executed.

[0038] Optionally, the first preset condition includes: the number of votes of the target annotation label is greater than a first vote threshold, and the proportion of the votes of the target annotation label is greater than a first proportion threshold;

[0039] The first labeling strategy is an automatic labeling strategy, including: using the labeled labels that meet the conditions as labels for the asset attributes corresponding to the unassociated sample images in the corresponding clusters;

[0040] The second preset condition includes: the number of votes for the target label is greater than a second vote threshold, and the proportion of votes for the target label is greater than a second proportion threshold;

[0041] The second labeling strategy is a semi-automatic labeling strategy, including: based on the obtained associated labeled label information, obtaining the labeled labels ranked TOP n by votes, and using the TOP n labeled labels as recommended labels, manually checking the recommended labels and determining a unique expected label from the recommended labels, and using the expected label as the label of the asset attribute corresponding to all unassociated sample images in the corresponding cluster;

[0042] The third labeling strategy is manual labeling, including: based on the obtained associated labeled label information, obtaining the labeled labels ranked TOP n in terms of votes, and using the TOP n labeled labels as recommended labels, manually analyzing the recommended labels, and determining a unique expected label based on an actual preset database, and using the expected label as a label for the asset attribute corresponding to all unassociated sample images in the corresponding cluster;

[0043] Among them, n≥2.

[0044] Optionally, the method further includes: if the highest similarity obtained is not greater than a preset similarity threshold, adding a new cluster to which the sample image belongs, and updating and adding the cluster to which it belongs to the cluster set information.

[0045] Optionally, the method further includes: updating the cluster center point of the representative image set;

[0046] The updating of the representative image set comprises:

[0047] Calculate the mean of the corresponding values ​​of all samples in the same cluster;

[0048] If the mean value is different from the value corresponding to the current representative image in the representative image set, the cluster center point of the current representative image in the representative image set is updated to the mean value.

[0049] Optionally, the processing the asset image based on a preset processing strategy to obtain a sample image further includes: processing the first image using a fuzzification algorithm, and performing multi-level thresholding processing on the processed first image to obtain a sample image.

[0050] In a second aspect, the embodiment of the present disclosure further provides a network asset attribute annotation system, which includes the following technical solutions:

[0051] A collection module, used to collect a plurality of target image information of a network asset target site within a preset period, each of which includes an asset image and a corresponding image file number and a network asset identifier;

[0052] A processing module, used for processing the asset image based on a preset processing strategy to obtain a sample image;

[0053] A creation module, used to create a sample set based on all the sample images; create a cluster center point set, wherein the cluster center point set is an empty set in an initial state;

[0054] A clustering module, used for clustering the sample set and the cluster center point set based on a preset clustering strategy to obtain a representative image set;

[0055] A clustering cluster acquisition module, used for acquiring the clustering cluster to which each sample image in the sample set belongs based on a similarity strategy, the representative image set, and preset clustering cluster set information;

[0056] A labeled label information acquisition module, used to acquire the associated labeled label information in a preset labeled network asset information database based on the network asset identifier corresponding to each of the sample images;

[0057] The total amount of the obtained labeled label information is less than the total amount of all the sample images;

[0058] A labeling module, used to label the label in the labeled label information as the asset attribute label corresponding to the associated sample image;

[0059] The annotated label information is analyzed, and the corresponding annotation strategy is called to annotate the asset attributes corresponding to the unassociated sample images.

[0060] In a third aspect, the embodiments of the present disclosure further provide a computer device, which adopts the following technical solution:

[0061] The computer device comprises:

[0062] at least one processor; and,

[0063] a memory communicatively connected to the at least one processor; wherein,

[0064] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute any of the above-mentioned network asset attribute labeling methods.

[0065] In a fourth aspect, an embodiment of the present disclosure further provides a computer-readable storage medium, which stores computer instructions, and the computer instructions are used to enable a computer to execute any of the network asset attribute labeling methods described above.

[0066] In a fifth aspect, an embodiment of the present disclosure further provides a computer program product, including a computer program / instruction, which implements the steps of any of the above methods when executed by a processor.

[0067] The network asset attribute labeling method disclosed in the present application can effectively reduce manual participation and improve data collection efficiency by automatically collecting target image information of network asset target sites within a preset period. Automatic collection can not only cover more samples, but also reduce omissions or errors caused by human factors, thereby improving the accuracy of subsequent processing; by processing asset images based on preset processing strategies, sample images are obtained. This step standardizes the image data, ensures the consistency of subsequent analysis, and provides a stable basis for subsequent clustering and labeling; creates a sample set. This process improves data management efficiency by systematically storing processed sample images, thereby making it easy to call and process massive data; uses a preset clustering strategy to cluster the sample set to form a representative image set, and uses a clustering algorithm to automatically identify Images with high similarity are identified and grouped to avoid errors that may occur during manual retrieval; according to the similarity strategy, each sample image is mapped to the cluster to which it belongs. By clarifying the relationship between samples, the understanding and recording of asset attributes are strengthened, and subjective judgment during manual designation is reduced; using the network asset identifier of each sample image, the associated label information is obtained from the annotated network asset information library, and the consistency and accuracy of information are improved through data association, reducing unnecessary re-labeling work; on the basis of associating the annotated label information, it is annotated as the asset attribute label of the relevant sample image. At the same time, by analyzing the unlabeled sample images, the corresponding labeling strategy is called to annotate them. In this way, large-scale asset image automatic labeling can be achieved, which significantly improves the labeling efficiency.

[0068] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the following preferred embodiments are specifically cited and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0070] Figure 1 A schematic diagram of a flow chart of a network asset attribute labeling method provided in an embodiment of the present disclosure.

[0071] Figure 2 A schematic flow chart of a method for acquiring a sample image provided in an embodiment of the present disclosure.

[0072] Figure 3 A flowchart of a method for performing multi-level thresholding processing provided in an embodiment of the present disclosure.

[0073] Figure 4 A schematic flowchart of a method for acquiring a representative image set provided in an embodiment of the present disclosure.

[0074] Figure 5 A schematic flow chart of a method for obtaining the cluster to which each sample image in a sample set belongs provided in an embodiment of the present disclosure.

[0075] Figure 6 A flowchart of a method for calling a corresponding annotation strategy to perform attribute annotation provided in an embodiment of the present disclosure.

[0076] Figure 7 A schematic diagram of the structure of a computer device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0077] The embodiments of the present disclosure are described in detail below with reference to the accompanying drawings.

[0078] It should be clear that the following embodiments of the present disclosure are described by specific specific examples, and those skilled in the art can easily understand other advantages and effects of the present disclosure from the contents disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of the present disclosure, rather than all of the embodiments. The present disclosure can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present disclosure. It should be noted that the following embodiments and features in the embodiments can be combined with each other in the absence of conflict. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present disclosure.

[0079] It should be noted that various aspects of the embodiments within the scope of the appended claims are described below. It should be apparent that the aspects described herein may be embodied in a wide variety of forms, and any specific structure and / or function described herein is merely illustrative. Based on the present disclosure, it should be understood by those skilled in the art that an aspect described herein may be implemented independently of any other aspect, and two or more of these aspects may be combined in various ways. For example, any number of aspects described herein may be used to implement the device and / or practice the method. In addition, other structures and / or functionalities other than one or more of the aspects described herein may be used to implement this device and / or practice this method.

[0080] It should also be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present disclosure. The drawings only show components related to the present disclosure rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0081] Additionally, in the following description, specific details are provided to facilitate a thorough understanding of the examples. However, it will be understood by those skilled in the art that the aspects described may be practiced without these specific details.

[0082] Reference Figure 1 The present application discloses a network asset attribute labeling method, which specifically includes:

[0083] S100, collecting a plurality of target image information of a network asset target site within a preset period, each target image information including an asset image and a corresponding image file number and a network asset identifier.

[0084] Specifically, site snapshot collectors can be deployed at multiple key nodes in the network to collect and store network asset site home page image data, that is, regularly capture and save screenshots of the target website's home page. This distributed collection system can ensure the comprehensiveness and accuracy of the collected data; among them, a site snapshot collector usually refers to a tool or program that can regularly access and take screenshots of a website's home page.

[0085] Specifically, the site snapshot collector will access the homepage of a specified website at a preset time interval. After the access, the site snapshot collector will take a screenshot of the homepage content of the webpage, and the screenshot will be stored in a specified database or storage system.

[0086] This step can ensure the timeliness and accuracy of sample data, facilitate subsequent processing, and achieve a comprehensive understanding of the target assets, providing reliable basic data for subsequent labeling and processing.

[0087] S200, processing the asset image based on a preset processing strategy to obtain a sample image.

[0088] This step is to improve the usability of the image and ensure the consistency of the sample.

[0089] Specifically, refer to Figure 2 , the method for obtaining the sample image specifically includes the following steps:

[0090] S210, processing each asset image to obtain an initial four-dimensional image data matrix;

[0091] The initial four-dimensional image data matrix includes a red dimension image data matrix, a green dimension image data matrix, a blue dimension image data matrix and a transparency dimension image data matrix;

[0092] All initial four-dimensional image data matrices have the same data type and channel order. Specifically, in each initial four-dimensional image data matrix, they are sorted according to the red dimension image data matrix, the green dimension image data matrix, the blue dimension image data matrix, and the transparency dimension image data matrix.

[0093] In this step, the storage type of the image data values ​​in the initial four-dimensional image data matrix can be selected as integer or floating point, and the arrangement order of each channel can be selected by yourself, but it is necessary to satisfy that all processed images have the same data type and channel order.

[0094] Among them, the red dimension image data matrix, the green dimension data image matrix, the blue dimension image data matrix, and the transparency dimension image data matrix are all two-dimensional matrices.

[0095] In this step, each asset image is read and converted into an initial four-dimensional image data matrix using an image processing library (such as OpenCV, PIL, etc.).

[0096] The initial four-dimensional image data matrix includes the following dimensions: 1) Red dimension: extract the pixel values ​​of the red channel in the image to obtain a separate red image data matrix; 2) Green dimension: extract the pixel values ​​of the green channel to form a green image data matrix; 3) Blue dimension: obtain the pixel values ​​of the blue channel to obtain a blue image data matrix; 4) Transparency dimension: if the image supports transparency, extract the pixel values ​​of the alpha channel to form a transparency image data matrix. Among them, the data type (such as uint8) and channel order (RGB or RGBA) of all initial four-dimensional image data matrices remain consistent.

[0097] In this step, by decomposing the channels, we can better analyze and process the different features of the image. The unified data type and channel order facilitate the processing and calculation of subsequent steps and improve the processing efficiency.

[0098] S220, performing scaling processing on all initial four-dimensional image data matrices based on a preset scaling algorithm, and recording the processed initial four-dimensional image data matrix as a first image data matrix;

[0099] All first image data matrices have the same size.

[0100] Among them, the preset scaling algorithm preferably includes one or more of a nearest neighbor interpolation algorithm, a bilinear interpolation algorithm, a bicubic interpolation algorithm, and a nonlinear interpolation algorithm.

[0101] Specifically, a preset scaling algorithm is applied to all initial four-dimensional image data matrices and they are uniformly adjusted to the same size. The uniform image size is very important for subsequent processing, as it avoids calculation errors caused by inconsistent sizes, ensures that the scaled images maintain the highest possible quality, and helps to improve the accuracy of subsequent processing steps (such as thresholding).

[0102] S230, performing multi-level thresholding processing on the first image data matrix to obtain a sample image.

[0103] Multi-level thresholding can more effectively extract important features, enhance image contrast, and improve the sensitivity of subsequent processing.

[0104] Furthermore, the present application also includes: using a blurring algorithm to process the first image, performing multi-level thresholding processing on the processed first image, obtaining a sample image, and reducing the impact of noise through blurring; specifically, the image blurring algorithm can be selected from Gaussian blur, Kawase blur, double blur, granular blur, etc.

[0105] Reference Figure 3 , the method for performing multi-level thresholding processing specifically includes:

[0106] S231, performing multi-level threshold processing on pixel point values ​​in each dimensional matrix in the first image data matrix to obtain a second image data matrix;

[0107] Multi-level thresholding includes: d represents the corresponding pixel value in each dimension matrix, d' is the result of the pixel after multi-level thresholding processing, % is the remainder calculation, and E is the preset threshold parameter.

[0108] Image multi-level thresholding refers to processing the continuous values ​​of image pixels into discrete values, reducing the impact of noise and increasing the distinction between the main body of the image and the background. The principle of the method is: set a preset threshold parameter E, perform multi-level thresholding calculations on each pixel in each channel of the original image, and obtain the multi-level thresholding results of the entire image. Specifically, for each dimensional matrix (red, green, blue and transparency), traverse each pixel in it, and perform multi-level processing on the value d of each pixel (assuming the value range is 0-255) according to the set threshold. Through the remainder calculation, the level of the pixel in different intervals can also be obtained; the results of each dimensional matrix after thresholding are collected to form a new image data matrix, that is, the second image data matrix.

[0109] Multi-level thresholding can subdivide the grayscale levels of an image so that pixels of the same type are classified into the same category, extracting features more effectively. The generated second image data matrix retains important information structures and is suitable for subsequent processing.

[0110] S232, acquiring a sample image based on the second image data matrix.

[0111] Specifically, the value of each pixel in the second image matrix can be mapped back to the range of 255 for visualization. For example, each level can correspond to a color or grayscale value, so that different values ​​appear in different colors. The processed pixel values ​​are then combined into the final sample image to form a clear image that highlights important features.

[0112] Through this step, the final generated sample image highlights the required features and is easier to analyze and label.

[0113] Table 1

[0114]

[0115] Referring to Table 1, assuming that the value of the original pixel before multi-level thresholding is 0-255, different E parameters are used to obtain the results after multi-level thresholding, which reduces the influence of noise and increases the distinction between the image subject and the background.

[0116] The multi-level thresholding processing method disclosed in this embodiment provides a richer feature representation than simple thresholding, making key features more distinct and improving the efficiency of subsequent analysis; through multi-level processing, it can effectively reduce the interference of background noise on key parts and highlight the main features of network assets; this method is applicable to different types of asset images and can adaptively process different image characteristics; it is presented in a color-coded form, which significantly improves the interpretability and intuitiveness of the image; through standardized sample images, it provides basic data for subsequent steps such as clustering and classification, further optimizing the entire network asset management and labeling process.

[0117] S300, creating a sample set based on all sample images;

[0118] Create a cluster center point set, which is initially an empty set.

[0119] Centralized management helps improve data access efficiency and also lays the foundation for subsequent clustering analysis.

[0120] S400, clustering the sample set and the cluster center point set based on a preset clustering strategy to obtain a representative image set.

[0121] Through clustering, similar sample images can be grouped into the same category, reducing the complexity of subsequent processing and annotation. This process makes asset identification and provenance clear, which facilitates subsequent information retrieval and use.

[0122] Reference Figure 4 , the method for obtaining the representative image set specifically includes the following steps:

[0123] S410, randomly extract a sample image from the sample set as the current representative image and store it in the cluster center point set to obtain an updated cluster center point set.

[0124] Assume that there are 10 images in the sample set, and the random selection result is the third image, then the cluster center point set is updated from 0 images to 1 image, and the newly added representative image is the third image. In this step, random sampling avoids human bias, helps ensure the diversity of clustering, generates initial representative images for clustering, and facilitates subsequent similarity calculation and sample aggregation.

[0125] S420, obtaining the similarity between the current representative image and each remaining sample image in the sample set, and deleting image objects whose similarity is not less than a preset similarity threshold from the sample set to obtain an updated sample set.

[0126] Specifically, similarity calculation is performed based on a similarity algorithm, for example, Euclidean distance, cosine similarity or other similarity measurement methods can be used. By eliminating dissimilar samples, the data set used for cluster analysis is streamlined, which can improve the efficiency of subsequent similarity calculations, so that the updated sample set is more focused on representative images, and the representativeness of each representative image is enhanced.

[0127] S430, determining whether the updated sample set is an empty set, and if so, taking the updated cluster center point set as the representative image set;

[0128] If not, a sample image is randomly selected from the updated sample set as the current representative image and stored in the cluster center point set to obtain an updated cluster center point set, and then the process returns to execute S420.

[0129] When the sample set is empty, a natural termination condition is provided to avoid the infinite loop problem, while ensuring the validity of the representative image set, which only contains relevant and representative samples.

[0130] The method for obtaining a representative image set disclosed in this embodiment improves the efficiency and effect of clustering by combining random selection with similarity filtering, and ensures the diversity and representativeness of representative images; by eliminating samples with low similarity, the amount of data for subsequent processing is reduced, and redundant calculations are avoided; when performing cluster analysis, it supports dynamic updating of cluster centers, improves adaptability, and makes the algorithm flexible when facing changing data sets; the clustering strategy can be adjusted according to specific needs and scenarios, such as modifying the similarity threshold or adopting different similarity calculation methods, which increases the flexibility of practical applications; the streamlined sample set more accurately represents the data characteristics, making the final representative image set more informative and effectively supporting subsequent analysis and decision-making.

[0131] By implementing the above steps, the sample set can be effectively divided into groups with similar features, thereby obtaining a higher quality representative image set, laying a solid foundation for subsequent tasks (such as labeling or classification).

[0132] S500, based on the similarity strategy, the representative image set, and the preset cluster set information, the cluster to which each sample image in the sample set belongs is obtained.

[0133] Reference Figure 5 , the method for obtaining the cluster to which each sample image in the sample set belongs includes:

[0134] S510: Acquire cluster set information based on the representative image set.

[0135] Assume that the representative image set contains multiple selected images (such as 5 images), each image corresponds to a cluster (for example: representative image 1 corresponds to cluster A, representative image 2 corresponds to cluster B, and so on). This information is stored in a related data structure, such as a dictionary or a mapping table, for easy subsequent query.

[0136] Through this step, the cluster to which each representative image belongs can be clearly identified, ensuring the uniqueness and consistency of the clustering information, which helps to ensure the accuracy of subsequent analysis.

[0137] S520, obtaining the similarity between each sample image in the sample set and all representative images in the representative image set.

[0138] Specifically, for each sample image in the sample set, all representative images in the representative image set are traversed to calculate the similarity between the sample image and each representative image; a preset similarity calculation method (such as Euclidean distance, cosine similarity, etc.) can be used to obtain the similarity value.

[0139] By calculating the similarity one by one, the relationship between the sample image and each cluster can be effectively evaluated, and the similarity results between each sample and the representative image can be retained to provide basic data for the next step of cluster allocation.

[0140] S530, based on the cluster set information, obtaining the cluster to which the current representative image with the highest similarity belongs, and using the cluster to which the corresponding sample image belongs as the cluster to which the corresponding sample image belongs.

[0141] Traverse the similarity data obtained in step S520 to find the representative image with the highest similarity between each sample image and the representative image; then use the cluster cluster set information previously constructed in S510 to determine the cluster cluster to which the representative image belongs, and assign it to the cluster cluster to which the sample image belongs, that is, use the cluster cluster as the cluster cluster to which the corresponding sample image belongs.

[0142] This step ensures that each sample image is assigned to the most appropriate cluster, thereby improving the accuracy of clustering.

[0143] The method for obtaining the cluster to which each sample image in the sample set disclosed in this embodiment determines the cluster of each sample image through clear steps, which helps to build a clear data classification system; the similarity calculation and clustering methods can be adjusted according to specific needs, so that the scheme can adapt to different types of image data and application scenarios; using representative images and their clustering information, the sample images can be quickly and accurately classified into the most suitable cluster, which improves the time efficiency of data processing; according to the clustering information, it is easier to understand the reason why each sample image is classified, which improves the depth and breadth of the result analysis. Through the implementation of the above steps, this scheme can effectively realize the clustering of each image in the sample set, improve the intelligent level of image classification, and then support the subsequent analysis and decision-making process.

[0144] Table 2

[0145] Serial number Network asset identification Belongs to cluster Image file number 1 192.168.0.1 Cluster A 283EAC 2 192.168.0.5 Cluster A 23F973 3 192.168.0.6 Cluster A 98EBF 4 172.10.1.13 Cluster B FF2111 5 172.10.2.29 Cluster B 4D143C

[0146] Specifically, for the sample image with the file number 283EAC, the cluster corresponding to the current representative image with the highest similarity calculated with all the representative images in the representative image set is cluster A, so the cluster to which the sample image belongs is cluster A; for the sample image with the file number 23F973, the cluster corresponding to the current representative image with the highest similarity calculated with all the representative images in the representative image set is cluster A, so the cluster to which the sample image belongs is cluster A; for the sample image with the file number 98EBF, the cluster corresponding to the current representative image with the highest similarity calculated with all the representative images in the representative image set is cluster A, so the cluster to which the sample image belongs is cluster A. The cluster to which the current representative image with the highest similarity belongs is cluster A, so the cluster to which the sample image belongs is cluster A; for the sample image with file number FF2111, the cluster to which the current representative image with the highest similarity calculated with all representative images in the representative image set belongs is cluster B, so the cluster to which the sample image belongs is cluster B; for the sample image with file number 4D143C, the cluster to which the current representative image with the highest similarity calculated with all representative images in the representative image set belongs is cluster B, so the cluster to which the sample image belongs is cluster B.

[0147] S600, based on the network asset identifier corresponding to each sample image, obtaining associated annotated label information in a preset annotated network asset information database;

[0148] The total amount of the obtained labeled label information is less than the total amount of all sample images.

[0149] In this application, there is no situation where all sample images have corresponding annotated labels, so the solution that this application wants to protect does not consider such a situation.

[0150] In this step, the efficiency and accuracy of labeling can be improved by using existing knowledge through the associated labeled label information.

[0151] S700, marking the marked label in the marked label information as the asset attribute label corresponding to the associated sample image;

[0152] Analyze the labeled label information and call the corresponding labeling strategy to label the corresponding asset attributes of the unassociated sample images.

[0153] Specifically, the annotated labels in the annotated label information are annotated as asset attribute labels corresponding to the associated sample images; for unassociated sample images, that is, for those sample images for which the associated annotated label information cannot be found in the preset annotated network asset information library, the corresponding annotation strategy is called to perform attribute annotation.

[0154] Specifically, refer to Figure 6, the method of calling the corresponding annotation strategy to perform attribute annotation specifically includes:

[0155] S710: Based on the acquired associated labeled tag information, determine the labeled tag with the most votes of the same type as the target labeled tag.

[0156] For sample images for which no labeled labels are found, the corresponding labeled labels can be left as empty fields.

[0157] For example, based on the network asset identifier corresponding to the sample image, the labeled label corresponding to the same network asset identifier is obtained from the preset labeled network asset information library. If all associated sample images are recorded as the first group of images, and all unassociated sample images are recorded as the second group of images, the labels of the asset attributes corresponding to the sample images in the first group of images can be found. For the attribute labels of the network assets corresponding to all sample images in the second group of images, the types of different asset attribute labels in the first group of images are analyzed, and the number of votes for each type of asset attribute label is obtained. In this way, the label with the most votes can be determined and then used as the target annotation label.

[0158] S720: When the number of votes and the proportion of votes of the target annotation label meet a first preset condition, execute a first annotation strategy.

[0159] Specifically, the first preset condition includes: the number of votes of the target annotation label is greater than the first vote threshold, and the proportion of the votes of the target annotation label is greater than the first proportion threshold;

[0160] The first labeling strategy is an automatic labeling strategy, which includes: using the labeled labels that meet the conditions as labels of the corresponding asset attributes of all unassociated sample images in the corresponding cluster.

[0161] Table 3

[0162]

[0163]

[0164] For example, in the preset labeled network asset information database, the labeled label corresponding to the sample image with the network asset identification of 192.168.0.1 is A, the labeled label corresponding to the sample image with the network asset identification of 192.168.0.6 is A, and the labeled label corresponding to the sample image with the network asset identification of 172.10.1.13 is B.

[0165] Table 4

[0166] Network asset identification Belongs to cluster The label of an attribute 192.168.0.1 Cluster A First 192.168.0.5 Cluster A 192.168.0.6 Cluster A First 172.10.1.13 Cluster B Second 172.10.2.29 Cluster B

[0167] For example, the asset attributes (A, A, B) corresponding to the associated sample images (192.168.0.1, 192.168.0.6, 172.10.1.13) can be used directly for annotation; for the unassociated sample images: 192.168.0.5, 172.10.2.29, determine that the annotated label with the most votes of the same type is "A", where the number of votes for "A" is 2, and the vote share is 0.67. If the first vote threshold is set to 1 and the first share threshold is 0.6, the first preset condition is met, and the automatic annotation strategy will be executed, and the annotated labels that meet the conditions will be used as labels for the corresponding asset attributes of all unassociated sample images in the corresponding cluster, that is, the label "A" is used as the label for the corresponding asset attributes of all unassociated sample images in cluster A (all unlabeled network assets), that is, "192.168.0.5" is automatically annotated as "A".

[0168] S730: When the number of votes and the percentage of votes of the target annotation label do not satisfy the first preset condition but satisfy the second preset condition, execute the second annotation strategy.

[0169] The second preset condition includes: the number of votes for the target label is greater than the second vote threshold, and the proportion of the votes for the target label is greater than the second proportion threshold;

[0170] The second labeling strategy is a semi-automatic labeling strategy, including: based on the obtained associated labeled label information, obtaining the labeled labels ranked TOP n in terms of votes, and using the TOP n labeled labels as recommended labels, manually verifying the recommended labels and determining a unique expected label from the recommended labels, and using the expected label as the label of the corresponding asset attribute of all unassociated sample images in the corresponding cluster.

[0171] S740: When the number of votes and the proportion of votes of the target annotation label do not meet the first preset condition and do not meet the second preset condition, execute the third annotation strategy.

[0172] Specifically, the third labeling strategy is manual labeling, including: based on the obtained associated labeled label information, obtaining the labeled labels ranked TOP n in terms of votes, and using the TOP n labeled labels as recommended labels, manually analyzing the recommended labels, and determining a unique expected label based on the actual preset database, and using the expected label as the label of the corresponding asset attribute of all unassociated sample images in the corresponding clustering cluster.

[0173] Among them, n≥2.

[0174] In this embodiment, when some associated annotated label information is found in the preset annotated network asset information library based on the network asset identification corresponding to each sample image, these existing resources can be utilized to improve the labeling efficiency and accuracy of the present method; when the associated annotated label information cannot be found in the preset annotated network asset information library, the algorithm will not trigger a class of conditions.

[0175] It should be noted that the vote threshold and percentage threshold can be flexibly set according to actual needs.

[0176] The solution disclosed in this application can effectively utilize existing information, greatly improve the degree of automation of labeling, and reduce labor costs; through strategic labeling, the consistency, accuracy, and efficiency of labeling can be guaranteed.

[0177] The labeling process disclosed in this application can flexibly determine which assets should be automatically labeled, semi-automatically labeled, or manually labeled, effectively utilize assets with existing labeling information, achieve an effective balance between automation and manual intervention, and is suitable for labeling of different scales. It is fast and efficient and can achieve the best labeling effect.

[0178] Furthermore, the network asset attribute labeling method disclosed in the present application also includes:

[0179] Specifically, if the highest similarity obtained is not greater than a preset similarity threshold, the cluster to which the sample image belongs is newly added, and the cluster to which it belongs is updated and added to the cluster set information.

[0180] By classifying the clusters to which the newly added sample images belong into new clusters, more diversity can be effectively introduced, avoiding the clusters from being too concentrated on certain types of samples, thereby enhancing the generalization ability of the model for network assets; this method can dynamically adjust the cluster structure according to the characteristics of the new data, so that the system can continuously adapt to and learn new asset types and maintain the advancement of the model; adding new samples to different clusters helps to improve the quality of asset attribute annotation and ensure that the annotation results are more accurate and representative; by setting the similarity threshold, irrelevant or low-similarity samples are avoided from being mistakenly classified into existing categories, which improves the robustness of the model and reduces misclassification; with the dynamic classification of new samples, the complexity of each calculation can be reduced, thereby improving the overall operation efficiency, especially when processing large-scale data. Overall, this method provides a more flexible and efficient means for the management and analysis of network assets, which helps to improve overall work efficiency and accuracy.

[0181] Furthermore, the network asset attribute labeling method disclosed in the present application also includes: updating the cluster center point representing the image set.

[0182] Specifically, the representative image set is updated, including:

[0183] Calculate the mean of the corresponding values ​​of all samples in the same cluster;

[0184] If the mean value is different from the value corresponding to the current representative image in the representative image set, the cluster center point corresponding to the current representative image in the representative image set is updated to the mean value.

[0185] This method of updating the cluster center points representing the image set can ensure that the cluster center more accurately reflects the characteristics of the samples of the same type by calculating the mean of all samples in the same type of cluster and updating the cluster center points, thereby improving the accuracy of the overall clustering; as the sample data continues to change, by dynamically updating the cluster center points, the model can adapt to the changes in data and maintain the rationality and timeliness of the clustering results; the mean update can effectively reduce the impact of noise data on the cluster center, making the clustering results more stable, thereby improving the reliability of the annotation results; by updating the cluster center points representing the image, the clustering is more concentrated and compact, and can better represent the characteristics within the class, making subsequent annotation and classification tasks more efficient; the calculation of the mean is usually simpler than classification based on complex models, which can significantly improve the computing efficiency in real-time or large-scale data processing; updating the cluster center points allows the model to be continuously optimized during the gradual learning process, improving the learning effect of the algorithm and enhancing the ability to process complex data.

[0186] In summary, this method not only maintains the flexibility and adaptability of the model, but also improves its stability and accuracy, providing more reliable support for the attribute labeling of network assets.

[0187] In this embodiment, the representative image set includes the values ​​corresponding to all the extracted current representative images.

[0188] The network asset attribute labeling method disclosed in this application can meet the labeling needs in different scenarios such as small-scale, medium-scale, and large-scale; specifically, a small-scale scenario refers to a scenario where the number of assets is less than 1,000, a medium-scale scenario refers to a scenario where the number of assets is between 1,000 and 100,000, and a large-scale scenario refers to a scenario where the number of assets is more than 100,000.

[0189] The network asset attribute labeling method disclosed in the present application can effectively reduce manual participation and improve data collection efficiency by automatically collecting target image information of network asset target sites within a preset period. Automatic collection can not only cover more samples, but also reduce omissions or errors caused by human factors, thereby improving the accuracy of subsequent processing; by processing asset images based on preset processing strategies, sample images are obtained. This step standardizes the image data, ensures the consistency of subsequent analysis, and provides a stable basis for subsequent clustering and labeling; creates a sample set. This process improves data management efficiency by systematically storing processed sample images, thereby making it easy to call and process massive data; uses a preset clustering strategy to cluster the sample set to form a representative image set, and uses a clustering algorithm to automatically identify Images with high similarity are identified and grouped to avoid errors that may occur during manual retrieval; according to the similarity strategy, each sample image is mapped to the cluster to which it belongs. By clarifying the relationship between samples, the understanding and recording of asset attributes are strengthened, and subjective judgment during manual designation is reduced; using the network asset identifier of each sample image, the associated label information is obtained from the annotated network asset information library, and the consistency and accuracy of information are improved through data association, reducing unnecessary re-labeling work; on the basis of associating the annotated label information, it is annotated as the asset attribute label of the relevant sample image. At the same time, by analyzing the unlabeled sample images, the corresponding labeling strategy is called to annotate them. In this way, large-scale asset image automatic labeling can be achieved, which significantly improves the labeling efficiency.

[0190] The solution of this application solves the problems of low efficiency, high cost and insufficient accuracy caused by manual labeling in the existing technology by realizing an automated and intelligent labeling process. In this way, it can not only reduce the cost of manual operation, but also meet the dual needs of timeliness and economy for asset management in the modern network environment; through this method, the management of network assets becomes more efficient, accurate and sustainable, and the overall network security defense line is improved.

[0191] The second aspect of the present application discloses a network asset attribute annotation system, including:

[0192] A collection module, used to collect a number of target image information of a network asset target site within a preset period, each target image information includes an asset image and a corresponding image file number and a network asset identifier;

[0193] A processing module, used to process the asset image based on a preset processing strategy to obtain a sample image;

[0194] A creation module is used to create a sample set based on all sample images; a cluster center point set is created, and the cluster center point set is an empty set in the initial state;

[0195] A clustering module is used to cluster the sample set and the cluster center point set based on a preset clustering strategy to obtain a representative image set;

[0196] The clustering cluster acquisition module is used to obtain the clustering cluster to which each sample image in the sample set belongs based on the similarity strategy, the representative image set, and the preset clustering cluster set information;

[0197] The labeled label information acquisition module is used to acquire the associated labeled label information in a preset labeled network asset information library based on the network asset identification corresponding to each sample image; wherein the total amount of labels of the acquired labeled label information is less than the total amount of all sample images;

[0198] A labeling module, used to label the label in the labeled label information as the asset attribute label corresponding to the associated sample image;

[0199] Analyze the labeled label information and call the corresponding labeling strategy to label the corresponding asset attributes of the unassociated sample images.

[0200] It should be noted that the scheme disclosed in the network asset attribute labeling method disclosed in the first aspect of this application is applicable to the network asset attribute labeling system disclosed in the second aspect of this application, so it will not be repeated here one by one.

[0201] The computer device according to the embodiment of the present disclosure includes a memory and a processor. The memory is used to store non-temporary computer-readable instructions. Specifically, the memory may include one or more computer program products, and the computer program product may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may, for example, include a random access memory (RAM) and / or a cache memory (cache), etc. The non-volatile memory may, for example, include a read-only memory (ROM), a hard disk, a flash memory, etc.

[0202] The processor may be a central processing unit (CPU) or other forms of processing units with data processing capabilities and / or instruction execution capabilities, and may control other components in the computer device to perform desired functions. In one embodiment of the present disclosure, the processor is used to run the computer-readable instructions stored in the memory, so that the computer device performs all or part of the steps of the network asset attribute labeling method of each embodiment of the present disclosure.

[0203] Those skilled in the art should be able to understand that in order to solve the technical problem of how to obtain a good user experience, the present embodiment may also include well-known structures such as a communication bus and an interface, and these well-known structures should also be included in the protection scope of the present disclosure.

[0204] like Figure 7 A schematic diagram of the structure of a computer device provided in an embodiment of the present disclosure is shown, which is a schematic diagram of the structure of a computer device suitable for implementing the embodiment of the present disclosure. Figure 7 The computer device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0205] like Figure 7 As shown, the computer device may include a processor (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) or a program loaded from a storage device into a random access memory (RAM). In the RAM, various programs and data required for the operation of the computer device are also stored. The processor, ROM, and RAM are connected to each other via a bus. An input / output (I / O) interface is also connected to the bus.

[0206] Typically, the following devices can be connected to the I / O interface: input devices such as sensors or visual information acquisition devices; output devices such as display screens; storage devices such as tapes, hard disks, etc.; and communication devices. The communication device can allow the computer device to communicate with other devices (such as edge computing devices) wirelessly or by wire to exchange data. Figure 7 A computer device having various devices is shown, but it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed instead.

[0207] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a non-transitory computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device, or installed from a ROM. When the computer program is executed by a processor, all or part of the steps of the network asset attribute labeling method of the embodiment of the present disclosure are executed.

[0208] For detailed description of this embodiment, reference may be made to the corresponding descriptions in the aforementioned embodiments, which will not be repeated here.

[0209] According to the computer-readable storage medium of the embodiment of the present disclosure, non-transitory computer-readable instructions are stored thereon. When the non-transitory computer-readable instructions are executed by a processor, all or part of the steps of the network asset attribute labeling method of each embodiment of the present disclosure are executed.

[0210] The above-mentioned computer-readable storage media include, but are not limited to: optical storage media (e.g., CD-ROM and DVD), magneto-optical storage media (e.g., MO), magnetic storage media (e.g., magnetic tape or mobile hard disk), media with built-in rewritable non-volatile memory (e.g., memory card) and media with built-in ROM (e.g., ROM box).

[0211] For detailed description of this embodiment, reference may be made to the corresponding descriptions in the aforementioned embodiments, which will not be repeated here.

[0212] The basic principles of the present disclosure are described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, effects, etc. mentioned in the present disclosure are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. are required by each embodiment of the present disclosure. In addition, the specific details disclosed above are only for the purpose of illustration and ease of understanding, and are not limitations. The above details do not limit the present disclosure to the necessity of adopting the above specific details to be implemented.

[0213] In the present disclosure, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that there is any such actual relationship or order between these entities or operations. The block diagrams of the devices, devices, equipment, and systems involved in the present disclosure are only illustrative examples and are not intended to require or imply that they must be connected, arranged, and configured in the manner shown in the block diagram. As will be appreciated by those skilled in the art, these devices, devices, equipment, and systems can be connected, arranged, and configured in any manner. Words such as "including", "comprising", "having", etc. are open words, referring to "including but not limited to", and can be used interchangeably with them. The words "or" and "and" used here refer to the words "and / or" and can be used interchangeably with them, unless the context clearly indicates otherwise. The words "such as" used here refer to the phrase "such as but not limited to", and can be used interchangeably with them.

[0214] Additionally, as used herein, "or" used in a list of items beginning with "at least one" indicates a separate list, so that, for example, a list of "at least one of A, B, or C" means A or B or C, or AB or AC or BC, or ABC (i.e., A and B and C). Furthermore, the word "exemplary" does not mean that the example described is preferred or better than other examples.

[0215] It should also be noted that in the system and method of the present disclosure, each component or each step can be decomposed and / or recombined. Such decomposition and / or recombination should be regarded as equivalent solutions of the present disclosure.

[0216] Various changes, substitutions, and modifications of the techniques described herein may be made without departing from the teachings defined by the appended claims. Furthermore, the scope of the claims of the present disclosure is not limited to the specific aspects of the processes, machines, manufactures, compositions of events, means, methods, and actions described above. Currently existing or later to be developed processes, machines, manufactures, compositions of events, means, methods, or actions that perform substantially the same functions or achieve substantially the same results as the corresponding aspects described herein may be utilized. Thus, the appended claims include such processes, machines, manufactures, compositions of events, means, methods, or actions within their scope.

[0217] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present disclosure. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present disclosure. Therefore, the present disclosure is not intended to be limited to the aspects shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

[0218] The above description has been given for the purpose of illustration and description. In addition, this description is not intended to limit the embodiments of the present disclosure to the forms disclosed herein. Although multiple example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, changes, additions and sub-combinations thereof.

Claims

1. A network asset attribute labeling method, characterized in that: include: Collecting a number of target image information of a network asset target site within a preset period, each of which includes an asset image and a corresponding image file number and a network asset identifier; Processing the asset image based on a preset processing strategy to obtain a sample image; Based on all the sample images, create a sample set; Creating a cluster center point set, wherein the cluster center point set is an empty set in an initial state; Clustering the sample set and the cluster center point set based on a preset clustering strategy to obtain a representative image set; Based on the similarity strategy, the representative image set, and the preset cluster set information, obtaining the cluster to which each sample image in the sample set belongs; Based on the network asset identifier corresponding to each of the sample images, the associated annotated label information is obtained in a preset annotated network asset information library; wherein the total amount of labels of the obtained annotated label information is less than the total amount of all the sample images; The annotated labels in the annotated label information are annotated as asset attribute labels corresponding to the associated sample images; the annotated label information is analyzed, and the corresponding annotation strategy is called to annotate the asset attributes corresponding to the unassociated sample images.

2. The network asset attribute labeling method according to claim 1, characterized in that: The step of processing the asset image based on a preset processing strategy to obtain a sample image includes: Processing each of the asset images to obtain an initial four-dimensional image data matrix; The initial four-dimensional image data matrix includes a red dimension image data matrix, a green dimension image data matrix, a blue dimension image data matrix and a transparency dimension image data matrix; All of the initial four-dimensional image data matrices have the same data type and channel order; Performing scaling processing on all the initial four-dimensional image data matrices based on a preset scaling algorithm, and recording the processed initial four-dimensional image data matrix as a first image data matrix; All of the first image data matrices have the same size; Perform multi-level thresholding processing on the first image data matrix to obtain a sample image.

3. The network asset attribute labeling method according to claim 2, characterized in that: The step of performing multi-level thresholding processing on the first image data matrix to obtain a sample image includes: Performing multi-level threshold processing on pixel point values ​​in each dimensional matrix in the first image data matrix to obtain a second image data matrix; The multi-level thresholding process includes: d represents the corresponding pixel value in each dimension matrix, d' is the result of multi-level thresholding of the pixel, % is the remainder calculation, and E is the preset threshold parameter; A sample image is acquired based on the second image data matrix.

4. The network asset attribute labeling method according to claim 1, characterized in that: The clustering of the sample set and the cluster center point set based on a preset clustering strategy to obtain a representative image set includes: S410, randomly extracting a sample image from the sample set as a current representative image and storing it in the cluster center point set to obtain an updated cluster center point set; S420, obtaining the similarity between the current representative image and each remaining sample image in the sample set, and deleting image objects whose similarity is not less than a preset similarity threshold from the sample set to obtain the updated sample set; S430, determining whether the updated sample set is an empty set, and if so, taking the updated cluster center point set as a representative image set; If not, a sample image is randomly selected from the updated sample set as the current representative image and stored in the cluster center point set to obtain the updated cluster center point set, and then the process returns to execute S420.

5. The network asset attribute labeling method according to claim 1, characterized in that: The obtaining, based on the similarity strategy, the representative image set, and the preset clustering cluster set information, the clustering cluster to which each sample image in the sample set belongs includes: Acquire cluster set information based on the representative image set; Obtaining the similarity between each sample image in the sample set and all representative images in the representative image set; Based on the cluster set information, the cluster corresponding to the current representative image with the highest similarity is obtained, and the cluster is used as the cluster of the corresponding sample image.

6. The network asset attribute labeling method according to claim 5, characterized in that: The analyzing the annotated label information and calling the corresponding annotation strategy to annotate the asset attributes corresponding to the unassociated sample image includes: Based on the obtained associated labeled label information, determine the labeled label with the most votes of the same type as the target labeled label; When the number of votes and the proportion of votes for the target label meet the first preset condition, Row-first labeling strategy; When the number of votes and the percentage of votes of the target label do not meet the first preset condition, but meet the second preset condition, execute the second labeling strategy; When the number of votes and the proportion of votes of the target annotation label do not meet the first preset condition and do not meet the second preset condition, the third annotation strategy is executed.

7. The network asset attribute labeling method according to claim 6, characterized in that: The first preset condition includes: the number of votes for the target label is greater than a first vote threshold, and the proportion of votes for the target label is greater than a first proportion threshold; The first labeling strategy is an automatic labeling strategy, including: using the labeled labels that meet the conditions as labels for the asset attributes corresponding to the unassociated sample images in the corresponding clusters; The second preset condition includes: the number of votes for the target label is greater than a second vote threshold, and the proportion of votes for the target label is greater than a second proportion threshold; The second labeling strategy is a semi-automatic labeling strategy, including: based on the obtained associated labeled label information, obtaining the labeled labels ranked TOP n by votes, and using the TOP n labeled labels as recommended labels, manually checking the recommended labels and determining a unique expected label from the recommended labels, and using the expected label as the label of the asset attribute corresponding to all unassociated sample images in the corresponding cluster; The third labeling strategy is manual labeling, including: based on the obtained associated labeled label information, obtaining the labeled labels ranked TOP n in terms of votes, and using the TOP n labeled labels as recommended labels, manually analyzing the recommended labels, and determining a unique expected label based on an actual preset database, and using the expected label as a label for the asset attribute corresponding to all unassociated sample images in the corresponding cluster; Among them, n≥2.

8. The network asset attribute labeling method according to claim 5, characterized in that: Also includes: If the highest similarity obtained is not greater than a preset similarity threshold, the cluster to which the sample image belongs is newly added, and the cluster to which it belongs is updated and added to the cluster set information.

9. The network asset attribute labeling method according to claim 4, characterized in that: Also includes: Updating the cluster center point of the representative image set; The updating of the representative image set comprises: Calculate the mean of the corresponding values ​​of all samples in the same cluster; If the mean value is different from the value corresponding to the current representative image in the representative image set, the cluster center point of the current representative image in the representative image set is updated to the mean value.

10. The network asset attribute labeling method according to claim 2, characterized in that: The method of processing the asset image based on a preset processing strategy to obtain a sample image also includes: processing the first image using a fuzzification algorithm, and performing multi-level thresholding processing on the processed first image to obtain a sample image.

Images