Product Capture Image Data Security Management Method and System

Abstract

This invention belongs to the field of information security and discloses a method and system for secure management of product-captured image data. The method includes: Step 1, obtaining a usage tag based on the borrowing application information and usage description submitted by the user when borrowing the equipment, and generating structured data record tuples based on the usage tag; Step 2, obtaining the legality judgment result of the user's use of the images based on the usage tag; Step 3, obtaining the spatial location of the borrowed equipment when capturing the image, and obtaining a comprehensive judgment result based on the spatial location; Step 4, obtaining the user's identity fingerprint of the captured image, and recording an operation log for the user's captured image based on the identity fingerprint; Step 5, calculating a risk score based on the operation log, and executing a risk-driven cleanup strategy based on the risk score. This invention significantly enhances the compliance control and data security of shooting behavior while ensuring the normal and convenient use of imaging equipment.

Description

Technical Field

[0001] This invention relates to the field of information security, and in particular to a method and system for secure management of product image data. Background Technology

[0002] In enterprises and research institutions with increasingly stringent information security requirements, employees are generally prohibited from bringing personal devices with photographic capabilities into office areas to prevent unauthorized photographing of confidential documents, screen content, or the internal environment. However, legitimate and necessary photographic needs still exist in the workplace, such as recording on whiteboards during meetings, collecting data during field visits, or providing on-site feedback. Therefore, some companies have implemented a "shared imaging equipment borrowing management system," whereby the organization provides photographic equipment (such as dedicated mobile phones, portable cameras, etc.), and employees submit an application in advance, specifying the purpose, time period, location, and other key information, which is then approved by the relevant person in charge before borrowing. This system superficially achieves controllability and process review of image data, but in actual implementation, numerous security vulnerabilities and regulatory blind spots still exist.

[0003] First, current systems often manage devices solely based on "process compliance." This means that as long as an application for the intended use is submitted and approved, employees can use the device to take pictures. However, the system itself cannot determine whether the content captured by the user truly matches the applied purpose. For example, an employee might borrow the device under the pretext of "meeting minutes" but actually use it to photograph computer screens or sensitive documents; such violations of "inconsistent purpose" are difficult to detect. Second, devices are shared by multiple people, and image data lacks a strong link to specific responsible parties. Even if the system records who borrowed the device, it is still difficult to identify who took which images after the images are exported or leaked. Third, current devices cannot perceive differences in the security level of the usage environment. Users entering "prohibited areas" such as server rooms or confidential meeting rooms will not trigger any restrictions or alarms. Finally, image data management largely relies on a "fixed-time automatic deletion" strategy, failing to incorporate more intelligent lifecycle management based on image usage frequency, access behavior, and data value. These problems collectively result in a situation where, even if the process appears controllable on the surface, the use and storage of image data remains "out of control during the process and difficult to trace afterward," failing to meet the actual needs of high-security scenarios. Summary of the Invention

[0004] The purpose of this invention is to disclose a method and system for secure management of product image data, thereby solving the technical problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] In a first aspect, the present invention provides a method for secure management of product captured image data, comprising:

[0007] Step 1: Obtain usage tags based on the borrowing application information and usage description submitted by the user when borrowing the device, and generate structured data record tuples based on the usage tags;

[0008] Step 2: Obtain the legality judgment result of the purpose of the image taken by the user based on the purpose tag;

[0009] Step 3: Obtain the spatial location of the device used when capturing the image, and obtain a comprehensive judgment result based on the spatial location;

[0010] Step 4: Obtain the user's identity fingerprint from the image taken by the user, and record an operation log for the image taken by the user based on the identity fingerprint;

[0011] Step 5: Calculate the risk score based on the operation log, and execute the risk-driven cleanup strategy based on the risk score;

[0012] Specifically, structured usage tags are obtained based on the borrowing application and usage description submitted by the user when borrowing the device, including:

[0013] Retrieve the usage tag corresponding to the usage description;

[0014] The system determines whether to approve a user's borrowing application based on the purpose tag and the borrowing application information. If so, it generates a structured data record tuple based on the purpose tag.

[0015] Furthermore, the borrowing application information includes user information u, device information d, and borrowing time period rt. r The purpose is described as D.

[0016] Furthermore, obtain the usage tags corresponding to the usage description, including:

[0017] Calculate the similarity between D and each tag in the tag library;

[0018] The tag with the highest similarity in the tag library is taken as the usage tag T corresponding to D.

[0019] Furthermore, the decision to approve a user's borrowing application is based on the usage tag and borrowing application information, including:

[0020] Obtain the job level λ(u) contained in the user information u, and determine whether λ(u) meets the minimum job level requirement corresponding to the purpose label T. If yes, approve the user's borrowing application; otherwise, disapprove the user's borrowing application.

[0021] Furthermore, structured data record tuples are generated based on usage tags, including:

[0022] R = (u, T, d, rt) r )

[0023] R represents a structured data record tuple.

[0024] Furthermore, step 2 includes:

[0025] Obtain the feature vector v of the image I taken by the user. I ;

[0026] Obtain the feature vector v corresponding to the purpose label T T ;

[0027] Calculate v I and v T The similarity S between them;

[0028] Let y represent the result of the legality judgment of the purpose of the image taken by the user;

[0029]

[0030] δ represents the preset similarity threshold.

[0031] Furthermore, step 3 includes:

[0032] Let L represent the spatial location of the device used by the user when capturing image I; let λ(L) represent the minimum level at which the image can be captured at L; let y final This indicates the result of a comprehensive judgment;

[0033]

[0034] allowed_areas represents areas where shooting is permitted; restricted_areas represents areas where shooting is not permitted; λ(u)≥λ(L) means that the user's job level is greater than or equal to the minimum job level at which shooting is permitted in L; λ(u)<λ(L) means that the user's job level is less than the minimum job level at which shooting is permitted in L.

[0035] If the overall judgment result is valid, image I will be stored in the database; otherwise, a warning will be issued according to the preset warning mechanism.

[0036] Furthermore, the identity fingerprint of the user's captured image is obtained, including:

[0037] Perform content hashing on image I to obtain the hash value H of the content of image I. I ;

[0038] The time t for capturing image I r The device's spatial location L, device information d, and user information u are concatenated into a string, and a hash calculation is performed to generate a metadata hash value H. M ;

[0039] For H I and H M By stitching the images together, we can obtain the identity fingerprint F of image I;

[0040] Embed the identity fingerprint F into the metadata of image I.

[0041] Furthermore, the operation log includes the operation type when operating on image I in the database, the identity fingerprint of image I, the user performing the operation, and the start and end times of the operation.

[0042] Secondly, the present invention provides a product image data security management system, including a first acquisition module, a second acquisition module, a third acquisition module, a fourth acquisition module, and a risk control module;

[0043] The first acquisition module is used to obtain usage tags based on the borrowing application information and usage description submitted by the user when borrowing the equipment, and to generate structured data record tuples based on the usage tags;

[0044] The second acquisition module is used to obtain the result of the legality judgment of the purpose of the image taken by the user based on the purpose tag;

[0045] The third acquisition module is used to acquire the spatial location of the device used when capturing images, and to obtain a comprehensive judgment result based on the spatial location;

[0046] The fourth acquisition module is used to acquire the identity fingerprint of the image captured by the user, and record operation logs about the image captured by the user based on the identity fingerprint;

[0047] The risk control module is used to calculate risk scores based on operation logs and execute risk-driven cleanup strategies based on the risk scores.

[0048] Specifically, structured usage tags are obtained based on the borrowing application and usage description submitted by the user when borrowing the device, including:

[0049] Retrieve the usage tag corresponding to the usage description;

[0050] The system determines whether to approve a user's borrowing application based on the purpose tag and the borrowing application information. If so, it generates a structured data record tuple based on the purpose tag.

[0051] Beneficial effects:

[0052] The core of this invention lies in constructing a multi-dimensional security control mechanism centered on compliance review of shooting behavior, traceability of shooting content, and secure area perception and control. The innovations of this invention are reflected in the following aspects: First, the system can determine in real time whether the image content matches the intended use when a user uses the device to shoot, thereby identifying potential abuse and solving the problem of "disconnect between application content and actual use" in existing systems. Second, the system automatically embeds responsible person identification information when the image is generated, achieving a strong binding between the image and the individual's behavior. Even if the image is exported, renamed, or forwarded, it can still be traced back to its source, overcoming the core defect of "unbinding the image from the responsible person" in current systems. Third, the system has environmental perception capabilities, able to identify whether the user is currently in a high-sensitivity area where shooting is prohibited based on visual and sensor data, and dynamically adjust device permissions accordingly, filling the loophole of insufficient spatial restrictions in existing systems. Fourth, this invention provides a data cleaning strategy that combines image usage behavior and access trajectory, differentiating itself from simple "periodic deletion," and realizing a dynamic assessment of the value of image data and a risk-driven cleaning mechanism. Through the aforementioned multi-layered and collaborative security control technologies, this invention significantly enhances the compliance control and data security of shooting behavior while ensuring the normal and convenient use of imaging equipment. It is applicable to various enterprise, scientific research, and government scenarios with strict requirements for information shooting security. Attached Figure Description

[0053] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0054] Figure 1 This is a schematic diagram of the image data security management method for the product of the present invention. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0056] like Figure 1As shown in one embodiment, the present invention provides a method for secure management of product captured image data, including:

[0057] Step 1: Obtain usage tags based on the borrowing application information and usage description submitted by the user when borrowing the device, and generate structured data record tuples based on the usage tags;

[0058] Step 2: Obtain the legality judgment result of the purpose of the image taken by the user based on the purpose tag;

[0059] Step 3: Obtain the spatial location of the device used when capturing the image, and obtain a comprehensive judgment result based on the spatial location;

[0060] Step 4: Obtain the user's identity fingerprint from the image taken by the user, and record an operation log for the image taken by the user based on the identity fingerprint;

[0061] Step 5: Calculate the risk score based on the operation log, and execute the risk-driven cleanup strategy based on the risk score;

[0062] Specifically, structured usage tags are obtained based on the borrowing application and usage description submitted by the user when borrowing the device, including:

[0063] Retrieve the usage tag corresponding to the usage description;

[0064] The system determines whether to approve a user's borrowing application based on the purpose tag and the borrowing application information. If so, it generates a structured data record tuple based on the purpose tag.

[0065] The main task of Step 1 is to transform the user-submitted borrowing requests and usage descriptions into structured usage tags, providing standardized input data for subsequent steps such as image compliance determination, identity tracing, and spatial compliance detection. This process not only helps the present invention automatically identify the rationality of each borrowing request, but also provides a foundation for the system to achieve comprehensive auditing while ensuring data consistency.

[0066] Furthermore, the borrowing application information includes user information u, device information d, and borrowing time period rt. r The purpose is described as D.

[0067] User information includes employee ID, job title, and other identity information;

[0068] Equipment information d includes equipment number, type, etc.;

[0069] Borrowing time period rt r This includes the time period for which the equipment will be used (including start and end times);

[0070] Purpose Description D includes a natural language description of the user's purpose for borrowing the device, such as "to take meeting minutes" or "to take product problem feedback".

[0071] Furthermore, obtain the usage tags corresponding to the usage description, including:

[0072] Calculate the similarity between D and each tag in the tag library;

[0073] The tag with the highest similarity in the tag library is taken as the usage tag T corresponding to D.

[0074] For example, if the purpose description is "I want to take meeting minutes," the similarity between this purpose description and each tag in the tag library will be calculated to find the matching tag. Assuming there are multiple tags in the tag library, the tag "meeting minutes" has the highest similarity to the user description; therefore, the purpose tag T = meeting minutes will be assigned to this application.

[0075] Furthermore, the similarity between D and each tag in the tag library is calculated separately, including:

[0076] The usage description is transformed into a high-dimensional semantic vector v using a semantic understanding model (such as MiniLM). D ;

[0077] Use v Ti Represents the i-th tag T in the tag library. i The semantic vectors of D and T are then... i The formula for calculating the similarity between them is:

[0078] S i =cos(v D ,v Ti )

[0079] S i This indicates that the cosine similarity calculation algorithm is used to calculate v. D and v Ti The similarity obtained through calculation.

[0080] Furthermore, the decision to approve a user's borrowing application is based on the usage tag and borrowing application information, including:

[0081] Obtain the job level λ(u) contained in the user information u, and determine whether λ(u) meets the minimum job level requirement corresponding to the purpose label T. If yes, approve the user's borrowing application; otherwise, disapprove the user's borrowing application.

[0082] Once the purpose tag T is determined, the job level in the user information u will be further verified to ensure it meets the tag requirements. Each tag in the tag library has a corresponding minimum job level, ensuring that only eligible users can apply for equipment for a specific purpose.

[0083] For example, the "Photographing Confidential Documents" tag might require users with a managerial or higher rank to apply for the device. If a user's rank λ(u) is a regular employee, which is lower than the minimum rank corresponding to the "Photographing Confidential Documents" tag, then the minimum rank requirement is not met, and the borrowing application will be considered rejected.

[0084] Furthermore, structured data record tuples are generated based on usage tags, including:

[0085] R = (u, T, d, rt) r )

[0086] R represents a structured data record tuple.

[0087] Once a user's borrowing application passes semantic matching and job level verification, a structured data record tuple R is generated. R contains user information, usage tags, device information, and the borrowing period. R serves as input for subsequent steps (image compliance assessment, identity fingerprint embedding, etc.). This information ensures that all subsequent image review, behavior auditing, and data cleaning are based on an accurate and structured data foundation.

[0088] In step 1, the present invention has generated a usage label T, which defines the intended use scenario when the user applies for the device (such as "meeting minutes" or "customer feedback"). The task of step 2 is to confirm, through image analysis, whether the captured image conforms to the usage label T.

[0089] Furthermore, step 2 includes:

[0090] Obtain the feature vector v of the image I taken by the user. I ;

[0091] The image I captured by the user's borrowed device is input into an image recognition model (such as ResNet or Inception). The image recognition model has been trained on a large-scale dataset and is able to extract the semantic features of the image. These semantic features can reflect the objects, scenes and other high-level information in the image.

[0092] Through this process, image I is transformed into a high-dimensional feature vector v. I This vector contains visual feature information of image I, such as the text, objects, environment, and their relative positions in image I.

[0093] For example, if an image contains a whiteboard and several attendees, the model will extract features related to "meeting minutes" and generate a feature vector v. I .

[0094] By extracting complex features from images through deep learning, compared with traditional image processing methods based on low-level features such as color and texture, it can more comprehensively understand the high-level semantics of images and significantly improve the accuracy of judgment.

[0095] Obtain the feature vector v corresponding to the purpose label T T ;

[0096] In step 1, the usage description provided by the user when borrowing the device is transformed into a structured usage label T. Each label has a corresponding feature vector predefined in the label library, representing the typical features of the image category.

[0097] In this step, the feature vector v of image I is... I The feature vector v corresponding to the tag T for this purpose in the tag library T A comparison is made to evaluate whether image I meets the application requirements. This is done by calculating the image feature vector v. I The feature vector v corresponding to the purpose label T T By assessing the similarity between images, this invention can determine the degree of fit between image content and usage tags.

[0098] Calculate v I and v T The similarity S between them;

[0099]

[0100] cos represents the calculation of cosine similarity;

[0101] Let y represent the result of the legality judgment of the purpose of the image taken by the user;

[0102]

[0103] δ represents the preset similarity threshold.

[0104] By calculating v I and v T Based on the similarity S between the two images, this invention will determine whether the image content of image I conforms to the user's declared purpose according to a preset similarity threshold δ. If S≥δ, it means that the image content is consistent with the user's declared purpose and is judged as "legal"; if S<δ, it means that the image content does not conform to the user's declared purpose and is judged as "suspected abuse".

[0105] For example, suppose a user requests to borrow the equipment for the purpose of "taking meeting minutes," and the feature vector v of image I is extracted using an image recognition model. I And compare it with the feature vector v corresponding to the purpose label T. TA comparison is made. Assume the calculated cosine similarity S = 0.92, and the threshold δ = 0.85. Because S is greater than δ, the image is deemed "legal".

[0106] If the similarity is low, for example, S=0.72, the image will be marked as "suspected misuse" and may trigger subsequent manual review or other monitoring measures.

[0107] Step 2 is the core of the entire system's behavior review process, ensuring that no non-compliant shooting behavior occurs during equipment borrowing. Through semantic matching and legality determination, this invention can automatically identify and mark images that do not meet the intended use requirements, thereby preventing unauthorized use of image data. In this way, Step 2 provides an image behavior compliance review mechanism for the entire system, which is a key link in image data security management.

[0108] The core task of step 3 is to ensure the compliance of the image capture environment through spatial recognition and shooting permission judgment. Specifically, this step needs to confirm that the shooting behavior occurs within a legal area and that the user has the permission to shoot within that area. Step 3 relies on the output y of step 2 to further determine spatial legality and ultimately determine whether the image acquisition behavior is legal.

[0109] Furthermore, step 3 includes:

[0110] Let L represent the spatial location of the device used by the user when capturing image I;

[0111] While the device is capturing images, it collects spatial location data through multiple sensors such as GPS, barometer, and Bluetooth to generate a spatial location L. For example, the latitude and longitude coordinates returned by GPS, combined with the triangulation results from multiple Bluetooth devices set up in the company, can determine whether the device is in a defined area such as a "meeting room," "confidential area," or "rest area."

[0112] The system maintains a spatial region database, assigning a legal shooting marker and job level requirements to each region. This information will be used for subsequent permission checks.

[0113] Let λ(L) denote the minimum rank that allows shooting at L;

[0114] Use y final This indicates a comprehensive judgment result, combining both content and spatial factors;

[0115]

[0116] allowed_areas represents areas where shooting is permitted; restricted_areas represents areas where shooting is not permitted; λ(u)≥λ(L) means that the user's job level is greater than or equal to the minimum job level at which shooting is permitted in L; λ(u)<λ(L) means that the user's job level is less than the minimum job level at which shooting is permitted in L.

[0117] If the overall judgment result is valid, image I will be stored in the database; otherwise, a warning will be issued according to the preset warning mechanism.

[0118] For example, when y final If it is illegal, the user's information for the captured image I will be sent to the company's information security management personnel.

[0119] Furthermore, considering that some areas are only allowed to shoot during specific time periods (such as shooting is prohibited late at night), a regularization term R that links time and space conditions is introduced. adj (t,L) allows for fine-tuning of the comprehensive judgment result based on the current time t and the regional time strategy.

[0120] The formula is as follows:

[0121] yb final =R(y final )+α·R adj (t,L)

[0122] If y final If it is "illegal", then R(y) final ) = 1; if y final If it is "legal", then R(y) final ) = 0.

[0123] R adj (t,L) is the regularization term, which combines time t and location L to determine special scenarios (such as nighttime or temporarily enclosed areas); yb final This is the result obtained after fine-tuning;

[0124] If yb final If the result is greater than the set comparison threshold (e.g., 0.5), the result is considered invalid; otherwise, the result is considered valid.

[0125] The specific calculation of the regularization term:

[0126] If the shooting time falls within a prohibited shooting period (such as from 10 PM the previous day to 9 AM the next day), then the behavior is considered very high-risk and given a high value, such as 1.

[0127] If the shooting time is within the permitted shooting period, such as from 9 am to 9 pm, the risk is lower, so we'll give it a moderate value, such as 0.5.

[0128] α is the regularization term adjustment strength coefficient, the value of which is configured by the system strategy, for example, it can be 0.6, controlling its effect on y. final The impact.

[0129] For example:

[0130] When a user requests to take photos of "meeting minutes," the image content is deemed "legitimate" through step 2.

[0131] Location L is in "Meeting Room A", an area where photography is permitted, and the photography time is 10:30, which complies with the area's time policy;

[0132] L requires the job level to be "ordinary employee", while u requires the job level to be "supervisor", i.e., λ(u)≥λ(L);

[0133] Final judgment yb final =Legal.

[0134] This step introduces spatial recognition, job level comparison, and spatiotemporal linkage rule judgment to form the "environmental compliance modeling layer" in the system, ensuring that the equipment is not abused for shooting in illegal areas.

[0135] The core task of step 4 is to generate a unique identity fingerprint for each image and embed it into the image's metadata. This fingerprint, combined with the image's usage log, ensures comprehensive recording and guarantees the traceability of the image throughout its lifecycle. Through image identity fingerprints and detailed log records, the system can trace the generation, processing, transmission, and deletion of each image, ensuring the integrity, security, and legality of the image data.

[0136] Image fingerprint definition: An image fingerprint is a unique identifier that combines the hash value of the image content with the image's metadata (such as shooting time, location, user information, etc.). This fingerprint ensures the authenticity of the image content and provides traceability throughout the image's lifecycle.

[0137] Furthermore, the identity fingerprint of the user's captured image is obtained, including:

[0138] Perform content hashing on image I to obtain the hash value H of the content of image I. I ;

[0139] Content hashing of image I can be performed using algorithms such as MD5 and SHA256;

[0140] The time t for capturing image I r The device's spatial location L, device information d, and user information u are concatenated into a string, and a hash calculation is performed to generate a metadata hash value H. M ;

[0141] H M =hash(t) r ||L||u||d)

[0142] For H I and H M By stitching the images together, we can obtain the identity fingerprint F of image I;

[0143] F = hash(H) I ||H M )

[0144] Embed the identity fingerprint F into the metadata of image I.

[0145] Images and their identity fingerprints will be stored in an encrypted manner to prevent data leakage or tampering. Every image upload, download, access, or deletion operation will be logged to ensure that image use and management always comply with regulations.

[0146] This method combines the image's content hash with the metadata from the time of capture to generate a comprehensive identity fingerprint. This approach prevents the potential for forgery or tampering with traditional hash values, ensuring the image's immutability.

[0147] Furthermore, the operation log includes the operation type when operating on image I in the database, the identity fingerprint of image I, the user performing the operation, and the start and end times of the operation.

[0148] Operation types include transfer, storage, access, and deletion.

[0149] The users who perform the operation include their job title, ID, etc.

[0150] Each time an image passes the system's validity check, the system generates a new record of the storage operation in the log database.

[0151] The operation logs are categorized according to the operation type and correspond one-to-one with the identity fingerprint F of the image.

[0152] For example, if a user accesses an image or performs a download operation, the system will record information such as the access time, operation type, and user performing the operation to ensure that every image operation is recorded in detail.

[0153] Operation logs are stored in a dedicated log management system to ensure efficient retrieval and secure storage. In the event of any unauthorized activity, administrators can accurately trace the entire operation history of the image based on the log information.

[0154] Furthermore, operation logs can be used for periodic auditing. By combining the image's identity fingerprint with the operation logs, the system can generate detailed audit reports, recording all operations performed on each image.

[0155] The introduction of image fingerprinting and logging enables comprehensive management and tracking of each image throughout its entire lifecycle, from generation and transmission to deletion. Any unauthorized operation on an image will be recorded and an alarm will be triggered immediately, ensuring image security.

[0156] The above solution comprehensively designs image identity fingerprint generation, operation logs, and image lifecycle management, ensuring the complete traceability of images throughout the system and enhancing data security and compliance.

[0157] The core task of step 5 is to build a risk model based on user access behavior and image usage history data to predict potential risks and misuse, and to formulate corresponding strategies for image data cleanup, storage duration, and access control management. By modeling and analyzing user access behavior, the system can trigger automatic cleanup strategies or permission changes based on risk assessments throughout the image data storage lifecycle, thereby reducing the risks associated with data misuse, leakage, or long-term storage.

[0158] Furthermore, a risk score is calculated based on the operation logs, including:

[0159] The operation logs are used to statistically analyze the frequency of access, time of access, duration of access, and job level of the users accessing the images within a preset period (e.g., one week).

[0160] Based on these factors, the system will calculate a risk score R. u This is used to measure whether each user's access to images poses a potential risk. Unlike traditional data cleaning strategies, this step introduces a risk prediction model based on access behavior. It not only relies on users' direct behavioral data but also considers multiple factors such as time and frequency, enabling it to more accurately predict potential risks.

[0161] To more accurately assess the risk of images, this invention weights multiple factors related to user access behavior to generate a comprehensive risk score R. u Assume each factor f i (Such as access frequency, access duration, etc.) all have corresponding weights. i The present invention can use the following formula to calculate the risk score:

[0162] R u = w1·f1 + w2·f2 + w3·f3

[0163] f1 represents the score for access frequency. For example, the access frequency can be divided into 10 intervals, and the total score is 100 points. The higher the access frequency, the higher the score. For example, when the access frequency is 95%, it belongs to the interval (90,100], so the score is 100 points; when the access frequency is 5%, it belongs to the interval [0,10), so the score is 10 points.

[0164] f2 represents the score for the access period, which is based on whether the access period falls within working hours. For example, if the working hours are from 9:00 to 18:00 and the access period is from 10:00 to 11:00, then the access period falls within working hours and the score is 50 points; if the access period is from 19:00 to 21:00, then the access period falls within working hours and the score is 100 points.

[0165] f3 represents the score based on visit duration. There is a direct proportional relationship between visit duration and score. For example, if the maximum score is 100 points, a visit duration of 12 hours will result in a score of [score missing].

[0166] w1, w2, and w3 are the weights for each rating, which can be adjusted based on historical data from risk analysis.

[0167] R u The higher the value, the greater the risk.

[0168] Based on risk score R u It can classify users into different risk levels, such as low-risk users, medium-risk users, and high-risk users, and design corresponding risk management strategies for each risk level.

[0169] For example, when the sum of the three weights is 1, and the maximum values ​​of f1, f2, and f3 are all 100, then when the risk score is in the range [0, 30), the user is a low-risk user; when the risk score is in the range [30, 60), the user is a medium-risk user; and when the risk score is in the range (60, 100), the user is a high-risk user.

[0170] Furthermore, risk-driven cleanup strategies are implemented based on risk scores, including:

[0171] Use R threshold This represents the threshold for risk scoring; if R... u More than R threshold If so, a risk-driven cleanup strategy will be implemented, which includes:

[0172] High-risk users: Encrypt, remove, or mark their related images as pending review;

[0173] Medium-risk users: restrict their access or require additional authentication;

[0174] Low-risk users: Continue to maintain normal access permissions.

[0175] The cleanup strategy in this step is not only based on the image's storage time and usage frequency, but also incorporates user behavior analysis. For example, if a user frequently accesses high-privilege images at night and for extended periods, the system will determine this behavior as potential abuse and trigger cleanup measures.

[0176] Based on risk scoring, the system can dynamically adjust users' access permissions to images. For example, if the system determines that a user's access behavior poses a high risk, it can adjust permissions in the following ways:

[0177] Lower the user's job level (e.g., from "Manager" to "Employee");

[0178] Restrict access to certain high-risk images;

[0179] Users are required to undergo two-factor authentication to access images.

[0180] Step 5 provides a dynamic and intelligent management framework for the entire image management system through user access behavior modeling and risk-driven cleanup strategies. This framework can assess potential risks based on actual user behavior and respond promptly.

[0181] The core of this invention lies in constructing a multi-dimensional security control mechanism centered on compliance review of shooting behavior, traceability of shooting content, and secure area perception and control. The innovations of this invention are reflected in the following aspects: First, the system can determine in real time whether the image content matches the intended use when a user uses the device to shoot, thereby identifying potential abuse and solving the problem of "disconnect between application content and actual use" in existing systems. Second, the system automatically embeds responsible person identification information when the image is generated, achieving a strong binding between the image and the individual's behavior. Even if the image is exported, renamed, or forwarded, it can still be traced back to its source, overcoming the core defect of "unbinding the image from the responsible person" in current systems. Third, the system has environmental perception capabilities, able to identify whether the user is currently in a high-sensitivity area where shooting is prohibited based on visual and sensor data, and dynamically adjust device permissions accordingly, filling the loophole of insufficient spatial restrictions in existing systems. Fourth, this invention provides a data cleaning strategy that combines image usage behavior and access trajectory, differentiating itself from simple "periodic deletion," and realizing a dynamic assessment of the value of image data and a risk-driven cleaning mechanism. Through the aforementioned multi-layered and collaborative security control technologies, this invention significantly enhances the compliance control and data security of shooting behavior while ensuring the normal and convenient use of imaging equipment. It is applicable to various enterprise, scientific research, and government scenarios with strict requirements for information shooting security.

[0182] The present invention also provides a product image data security management system, including a first acquisition module, a second acquisition module, a third acquisition module, a fourth acquisition module and a risk control module;

[0183] The first acquisition module is used to obtain usage tags based on the borrowing application information and usage description submitted by the user when borrowing the equipment, and to generate structured data record tuples based on the usage tags;

[0184] The second acquisition module is used to obtain the result of the legality judgment of the purpose of the image taken by the user based on the purpose tag;

[0185] The third acquisition module is used to acquire the spatial location of the device used when capturing images, and to obtain a comprehensive judgment result based on the spatial location;

[0186] The fourth acquisition module is used to acquire the identity fingerprint of the image captured by the user, and record operation logs about the image captured by the user based on the identity fingerprint;

[0187] The risk control module is used to calculate risk scores based on operation logs and execute risk-driven cleanup strategies based on the risk scores.

[0188] Specifically, structured usage tags are obtained based on the borrowing application and usage description submitted by the user when borrowing the device, including:

[0189] Retrieve the usage tag corresponding to the usage description;

[0190] The system determines whether to approve a user's borrowing application based on the purpose tag and the borrowing application information. If so, it generates a structured data record tuple based on the purpose tag.

[0191] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for secure management of product photography image data, characterized in that, include: Step 1: Obtain usage tags based on the borrowing application information and usage description submitted by the user when borrowing the device, and generate structured data record tuples based on the usage tags; Step 2: Obtain the legality judgment result of the purpose of the image taken by the user based on the purpose tag; Step 3: Obtain the spatial location of the device used when capturing the image, and obtain a comprehensive judgment result based on the spatial location; Step 4: Obtain the user's identity fingerprint from the image taken by the user, and record an operation log for the image taken by the user based on the identity fingerprint; Step 5: Calculate the risk score based on the operation log, and execute the risk-driven cleanup strategy based on the risk score; Specifically, structured usage tags are obtained based on the borrowing application and usage description submitted by the user when borrowing the device, including: Retrieve the usage tag corresponding to the usage description; The system determines whether to approve a user's borrowing application based on the purpose tag and the borrowing application information. If so, it generates a structured data record tuple based on the purpose tag. Step 2 includes: Obtain the feature vector of image I taken by the user. ; Obtain the feature vector corresponding to the purpose label T ; calculate and The similarity S between them; Let y represent the result of the legality judgment of the purpose of the image taken by the user; ； This indicates the preset similarity threshold; Step 3 includes: Let L represent the spatial location of the device used by the user when capturing image I; This indicates the lowest rank capable of taking photos at L; using This indicates the result of a comprehensive judgment; ； Indicates the areas where photography is permitted; This indicates areas where photography is not permitted. This indicates that the user's job level is greater than or equal to the minimum job level at which they can take photos at L; This indicates that the user's job level is lower than the minimum job level required to take photos at L; If the overall judgment result is valid, then image I is stored in the database; otherwise, an early warning is issued according to the preset warning mechanism. Risk scores are calculated based on operation logs, including: The operation log is used to statistically analyze the frequency of access to images, the time period of access, the duration of access, and the job level of the users who accessed the images within a preset period. Calculate the risk score using the following formula: ； A score indicating the frequency of visits. Indicates the score for the period of visit. The rating indicates the duration of the visit. These are the weights for each type of rating; Based on risk score Users are categorized into different risk levels; Risk-driven cleanup strategies are implemented based on risk scores, including: use This represents the threshold for risk scoring. Exceed If so, a risk-driven cleanup strategy will be implemented, which includes: High-risk users: Encrypt, remove, or mark their related images as pending review; Medium-risk users: restrict their access or require additional authentication; Low-risk users: Continue to maintain normal access permissions.

2. The product image data security management method according to claim 1, characterized in that, The borrowing application information includes user information u, device information d, and the borrowing period. The purpose is described as D.

3. The product image data security management method according to claim 2, characterized in that, Retrieve the usage tags corresponding to the usage description, including: Calculate the similarity between D and each tag in the tag library; The tag with the highest similarity in the tag library is taken as the usage tag T corresponding to D.

4. The product image data security management method according to claim 3, characterized in that, The decision to approve a user's borrowing request is based on the purpose tag and borrowing application information, including: Retrieve the job title contained in user information u ,judge Does the minimum job level requirement corresponding to the usage label T meet? If yes, the user's borrowing application will be approved; otherwise, the user's borrowing application will be rejected.

5. The product image data security management method according to claim 3, characterized in that, Structured data record tuples are generated based on usage tags, including: ； R represents a structured data record tuple.

6. The product image data security management method according to claim 1, characterized in that, Obtain the user's identity fingerprint from the image captured, including: Perform content hashing on image I to obtain the hash value of the content of image I. ; The time of image I was captured Spatial location of equipment Equipment Information and user information Concatenate the data into a string, perform a hash calculation, and generate a metadata hash value. ; right and By stitching the images together, we can obtain the identity fingerprint F of image I; Embed the identity fingerprint F into the metadata of image I.

7. The product image data security management method according to claim 6, characterized in that, The operation log includes the operation type when operating on image I in the database, the identity fingerprint of image I, the user performing the operation, and the start and end times of the operation.

8. A product image data security management system, characterized in that, It includes a first acquisition module, a second acquisition module, a third acquisition module, a fourth acquisition module, and a risk control module; The first acquisition module is used to obtain usage tags based on the borrowing application information and usage description submitted by the user when borrowing the equipment, and to generate structured data record tuples based on the usage tags; The second acquisition module is used to obtain the result of the legality judgment of the purpose of the image taken by the user based on the purpose tag; The third acquisition module is used to acquire the spatial location of the device used when capturing images, and to obtain a comprehensive judgment result based on the spatial location; The fourth acquisition module is used to acquire the identity fingerprint of the image captured by the user, and record operation logs about the image captured by the user based on the identity fingerprint; The risk control module is used to calculate risk scores based on operation logs and execute risk-driven cleanup strategies based on the risk scores. Specifically, structured usage tags are obtained based on the borrowing application and usage description submitted by the user when borrowing the device, including: Retrieve the usage tag corresponding to the usage description; The system determines whether to approve a user's borrowing application based on the purpose tag and the borrowing application information. If so, it generates a structured data record tuple based on the purpose tag. The results of determining the legality of the intended use of user-captured images based on usage tags include: Obtain the feature vector of image I taken by the user. ; Obtain the feature vector corresponding to the purpose label T ; calculate and The similarity S between them; Let y represent the result of the legality judgment of the purpose of the image taken by the user; ； This indicates the preset similarity threshold; Obtain the spatial location of the device used when capturing the image, and obtain a comprehensive judgment result based on the spatial location, including: Let L represent the spatial location of the device used by the user when capturing image I; This indicates the lowest rank capable of taking photos at L; using This indicates the result of a comprehensive judgment; ； Indicates the areas where photography is permitted; This indicates areas where photography is not permitted. This indicates that the user's job level is greater than or equal to the minimum job level at which they can take photos at L; This indicates that the user's job level is lower than the minimum job level required to take photos at L; If the overall judgment result is valid, then image I is stored in the database; otherwise, an early warning is issued according to the preset warning mechanism. Risk scores are calculated based on operation logs, including: The operation log is used to statistically analyze the frequency of access to images, the time period of access, the duration of access, and the job level of the users who accessed the images within a preset period. Calculate the risk score using the following formula: ； A score indicating the frequency of visits. Indicates the score for the period of visit. The rating indicates the duration of the visit. These are the weights for each type of rating; Based on risk score Users are categorized into different risk levels; Risk-driven cleanup strategies are implemented based on risk scores, including: use This represents the threshold for risk scoring. Exceed If so, a risk-driven cleanup strategy will be implemented, which includes: High-risk users: Encrypt, remove, or mark their related images as pending review; Medium-risk users: restrict their access or require additional authentication; Low-risk users: Continue to maintain normal access permissions.

Images