A method and system for generating a digital watermark image of a business

Abstract

The application discloses a kind of enterprise digital watermark image generation methods, the method includes: in response to enterprise business image acquisition request, original business image is collected and associated business association information is obtained;Watermark load data is generated based on business association information, and the watermark load data includes business watermark information and anti-fake verification information;According to the image content features of original business image, determine watermark embedding parameter;According to watermark embedding parameter, watermark load data is embedded in original business image, and generates watermarked business image. Thus, business image is formed in generation stage with business association information It is trusted to bind, and provide data basis for subsequent authenticity check and anomaly identification.

Description

Technical Field

[0001] This application relates to the fields of image processing and data security technology, and in particular to a method and system for generating digital watermarked images for enterprises. Background Technology

[0002] With the widespread adoption of digital management and mobile office systems in enterprises, image data has been extensively used in business scenarios such as on-site inspections, equipment maintenance, work order processing, security record keeping, and quality verification. Relevant personnel typically collect on-site images using mobile terminals or dedicated imaging equipment and upload them to the enterprise's business system to record the work process, confirm processing results, and support subsequent accountability. Because these images are often associated with specific times, locations, personnel, tasks, or equipment status, their authenticity and completeness directly impact the reliability of internal management, business audits, and evidence retention.

[0003] In related technologies, to improve the credibility of image records, time, location, and other identifying information are usually added to the images, or uploaded images are archived and preserved through business systems. However, existing methods focus more on the storage and display of image files, with limited correlation between image acquisition behavior, on-site business information, and image content. Furthermore, after images undergo transmission, compression, forwarding, or post-processing, the system often struggles to promptly determine whether the image content has changed, and it is difficult to distinguish whether image anomalies are caused by normal processing or human modification. For frequently used scenarios such as inspections, maintenance, and work orders, relying solely on manual review or ordinary file storage methods easily leads to problems such as difficulty in verifying image sources, insufficient credibility of business records, and low efficiency in identifying abnormal images.

[0004] Therefore, in enterprise digital image recording and business verification, the lack of assurance of the authenticity of image content, the weak reliability of the correlation between images and on-site business information, and the difficulty in timely and accurate identification of image anomalies have become urgent problems that need to be solved. Summary of the Invention

[0005] This application provides a method and system for generating digital watermarked images for enterprises, aiming to solve the problems of insufficient guarantee of image content authenticity, weak reliability of image-to-on-site business information correlation, and difficulty in timely and accurate identification of image anomalies in existing technologies for enterprise digital image recording and business verification.

[0006] In a first aspect, a method for generating digital watermarked images for enterprises, the method comprising: In response to an enterprise business image acquisition request, the system acquires original business images and obtains business association information associated with the enterprise business image acquisition request. Watermark payload data is generated based on the business association information. The watermark payload data includes business watermark information used to characterize the business association information and anti-counterfeiting verification information used to verify the authenticity of the business watermark information. The watermark embedding parameters are determined based on the image content features of the original business image; According to the watermark embedding parameters, the watermark payload data is embedded into the original service image to generate a watermarked service image.

[0007] Optionally, in the above scheme, the step of responding to an enterprise business image acquisition request, acquiring an original business image, and obtaining business association information associated with the enterprise business image acquisition request includes: Analyze the enterprise business image acquisition request to determine the corresponding business scenario type and business object identifier; A business information collection template is determined based on the business scenario type. The business information collection template includes at least two types of fields from the following categories: time field, location field, personnel field, equipment field, and business object field. Based on the aforementioned business information collection template, candidate business information is obtained from terminal collection information, logged-in user information, and enterprise business system information; The candidate business information is subjected to field integrity verification and field format standardization processing to obtain the business association information.

[0008] Optionally, in the above scheme, generating watermark payload data based on the business association information includes: The business-related information is structurally encapsulated according to a preset field order to obtain structured business data; The structured business data is compressed and encoded to obtain business-encoded data; The business encoding data is subjected to error correction encoding processing to obtain the business watermark information; The anti-counterfeiting verification information is generated based on the structured business data, and the business watermark information and the anti-counterfeiting verification information are combined into the watermark payload data.

[0009] Optionally, in the above scheme, generating the anti-counterfeiting verification information based on the structured business data and combining the business watermark information and the anti-counterfeiting verification information into the watermark payload data includes: The structured business data is used to perform a summary calculation to obtain business summary data; The business digest data is digitally signed based on a preset signature key to obtain signed data. The anti-counterfeiting verification information is generated based on the signature data, digest algorithm identifier, and signature key identifier; The business watermark information, the anti-counterfeiting verification information, and the payload version information are combined to obtain the watermark payload data.

[0010] Optionally, in the above scheme, determining the watermark embedding parameters based on the image content features of the original business image includes: Image content analysis is performed on the original business image to obtain image content features including at least one of texture complexity, brightness distribution, and edge intensity; Based on the image content features, a set of candidate embedding regions is determined, and the embedding adaptation evaluation value corresponding to each candidate embedding region is determined. Based on the embedding adaptation evaluation value, the target embedding region is determined from the candidate embedding region set; The watermark embedding strength is determined based on the image content features of the target embedding region and the preset invisibility constraint. The target embedding region and the watermark embedding intensity are determined as the watermark embedding parameters.

[0011] Optionally, in the above scheme, embedding the watermark payload data into the original service image according to the watermark embedding parameters to generate a watermarked service image includes: The original service image is subjected to frequency domain transformation to obtain the original frequency domain data; Based on the target embedding region, determine the set of frequency domain coefficients to be embedded from the original frequency domain data; Based on the watermark embedding strength, the watermark sequence corresponding to the watermark payload data is written into the set of frequency domain coefficients to be embedded, thereby obtaining watermarked frequency domain data. The watermarked frequency domain data is subjected to inverse frequency domain transformation to obtain the watermarked service image.

[0012] Optionally, in the above scheme, after generating the watermarked business image, the method further includes: Based on the image identifier of the watermarked service image and the service association information, generate image service index information; A baseline watermark record is generated based on the watermark payload data or the summary information corresponding to the watermark payload data. The image service index information, the baseline watermark record, and the storage address of the watermarked service image are associated to obtain watermark verification baseline data. The watermark verification baseline data is stored in the enterprise business system or the watermark verification server.

[0013] Optionally, in the above scheme, after generating the watermarked business image, the method further includes: Obtain the image of the service to be verified, and perform watermark extraction processing on the image of the service to be verified to obtain the watermark data to be verified. Based on the image service index information corresponding to the service image to be verified, obtain the corresponding watermark verification benchmark data; Based on the watermark data to be verified and the watermark verification benchmark data, the watermark consistency evaluation result is determined; Based on the anti-counterfeiting verification information in the watermark data to be verified, the authenticity of the business watermark information in the watermark data to be verified is verified to obtain the anti-counterfeiting verification result. Based on the watermark consistency evaluation result and the anti-counterfeiting verification result, the authenticity verification result of the business image to be verified is determined.

[0014] Optionally, in the above scheme, after determining the authenticity verification result of the business image to be verified based on the watermark consistency evaluation result and the anti-counterfeiting verification result, the method further includes: If the authenticity verification result indicates that the image to be verified is abnormal, the image to be verified is divided into blocks to obtain multiple image blocks to be detected. Local watermark extraction and local consistency detection are performed on each image block to be detected to obtain the local anomaly evaluation value corresponding to each image block to be detected. The set of abnormal image blocks is determined based on the local anomaly evaluation value corresponding to each image block to be detected; Based on the location distribution of the abnormal image block set in the service image to be verified, abnormal area identification information is generated.

[0015] Secondly, a digital watermark image generation system for enterprises, the system comprising: The image and business information acquisition module is used to respond to an enterprise business image acquisition request, acquire original business images, and obtain business association information associated with the enterprise business image acquisition request; A watermark payload generation module is used to generate watermark payload data based on the business association information. The watermark payload data includes business watermark information used to characterize the business association information and anti-counterfeiting verification information used to verify the authenticity of the business watermark information. The embedding parameter determination module is used to determine the watermark embedding parameters based on the image content features of the original business image. The watermark image generation module is used to embed the watermark payload data into the original business image according to the watermark embedding parameters to generate a watermarked business image. The image verification module is used to extract watermarks from the business image to be verified, and to determine the authenticity verification result of the business image to be verified based on the extracted watermark data and the corresponding watermark verification benchmark data. The anomaly localization module is used to perform block watermark consistency detection on the service image to be verified when the authenticity verification result indicates that there is an anomaly in the service image to be verified, and determine the anomaly area identification information based on the detection result.

[0016] Compared with the prior art, this application has at least the following beneficial effects: Based on further analysis and research of existing technical problems, this application recognizes that existing technologies in enterprise digital image recording and business verification suffer from insufficient assurance of image content authenticity, weak reliability of image-to-on-site business information association, and difficulty in timely and accurate identification of image anomalies. By responding to enterprise business image acquisition requests, this application simultaneously acquires the original business image and related business information associated with the acquisition request, establishing a connection between the image content and the corresponding business scenario, business object, or on-site acquisition behavior at the data source. Furthermore, based on the business association information, it generates watermark payload data including business watermark information and anti-counterfeiting verification information, enabling the business association information not only to be embedded in the image but also to provide a basis for subsequent authenticity verification. Then, it determines watermark embedding parameters according to the image content characteristics of the original business image and embeds the watermark payload data into the original business image according to these parameters to generate a watermarked business image. This allows the watermark embedding process to adapt to the image content characteristics, reducing the impact on normal image use and viewing. Based on the above processing, watermarked business images carry watermark payload data corresponding to business-related information and usable for authenticity verification during the generation stage. This enhances the reliability of the association between image content and on-site business information and provides data basis for subsequent identification of whether the image is abnormal. This solves the problems of insufficient guarantee of image content authenticity, weak reliability of the association between images and on-site business information, and difficulty in timely and accurate identification of image anomalies in enterprise digital image records. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating a method for generating digital watermarked images for enterprises, provided in one embodiment of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0019] In one embodiment, such as Figure 1As shown, a method for generating digital watermarked images for enterprises is provided, including the following steps: In response to an enterprise business image acquisition request, the system acquires original business images and obtains business association information associated with the enterprise business image acquisition request. Watermark payload data is generated based on the business association information. The watermark payload data includes business watermark information used to characterize the business association information and anti-counterfeiting verification information used to verify the authenticity of the business watermark information. The watermark embedding parameters are determined based on the image content features of the original business image; According to the watermark embedding parameters, the watermark payload data is embedded into the original service image to generate a watermarked service image.

[0020] In one embodiment, a method for generating enterprise digital watermarked images is provided. This method can be executed by a mobile terminal, an enterprise business system, a watermark processing server, or a combination of both. The mobile terminal can be a smartphone, tablet, inspection terminal, law enforcement recording device, or other electronic device with image acquisition capabilities, all equipped with an enterprise digital watermarked image acquisition program. The enterprise business system can be a property inspection system, work order management system, equipment maintenance system, security management system, or other systems that require the recording, verification, and archiving of business images.

[0021] Specifically, in response to an enterprise business image acquisition request, the system acquires original business images and obtains business association information associated with the request. The enterprise business image acquisition request can be triggered by a user clicking a camera control, entering an inspection point shooting page, entering a work order initiation page, entering a work order completion page, entering an equipment maintenance record page, or entering a security evidence collection page within the enterprise business system. The original business image is image data acquired by the image acquisition device before watermark embedding. The business association information describes the on-site business context corresponding to this image acquisition action and may include at least two of the following: shooting time information, shooting location information, personnel identity information, acquisition device information, and business object information. For example, the shooting time information may include system time or trusted time source time accurate to milliseconds; the shooting location information may include GPS latitude and longitude, address resolution results, point of interest information, building, floor, unit or inspection point location; personnel identity information may include user ID, name, department, role; data collection device information may include device model, unique device identifier, device IP address; business object information may include work order ID, work order type, work order title, device ID, inspection point number, camera ID or event number.

[0022] After acquiring business-related information, watermark payload data is generated based on this information. The watermark payload data includes business watermark information and anti-counterfeiting verification information. The business watermark information characterizes the business-related information, enabling subsequent recovery or verification of the business context corresponding to the image based on the watermark content. The anti-counterfeiting verification information verifies the authenticity of the business watermark information, making it difficult to forge or replace. Specifically, the business-related information can first be encapsulated into structured business data according to a preset field structure. Then, the structured business data undergoes digest calculation, signature processing, encoding processing, or error correction processing to form watermark payload data suitable for embedding in the image. The watermark payload data can be a binary data sequence, the length of which can be determined based on the amount of information in the business-related information, the image resolution, and the embedding capacity. For example, it can be set to 512 bits to 1024 bits, or configured to other lengths according to the actual business system.

[0023] Furthermore, watermark embedding parameters are determined based on the image content features of the original business image. Image content features may include at least one of texture complexity, brightness distribution, edge strength, local flatness, color channel features, and compression-sensitive region features. Watermark embedding parameters may include at least one of watermark embedding region, watermark embedding position, watermark embedding intensity, frequency domain coefficient selection rules, number of repeated embedding times, or error correction redundancy ratio. By determining watermark embedding parameters based on image content features, the watermark embedding process can adapt to the texture, brightness, and edge distribution of different images, reducing the impact of the watermark on the image's visual quality and improving the extractability of the watermark after compression, scaling, and transmission processing.

[0024] Finally, according to the watermark embedding parameters, the watermark payload data is embedded into the original business image to generate a watermarked business image. Specifically, the watermark payload data can be converted into a watermark sequence, and this watermark sequence can be embedded into the pixel domain, transform domain, or frequency domain data of the original business image. Preferably, the watermark payload data can be embedded into the frequency domain data of the original business image, so that the watermark information does not directly cover the image surface, thereby reducing the impact on the image viewing effect. The generated watermarked business image can be uploaded to the enterprise business system for archiving, or it can be saved locally on the mobile terminal and then synchronized to the server.

[0025] Through the above implementation method, this application simultaneously acquires business-related information during the business image acquisition stage, and combines the business watermark information used to characterize the business-related information with the anti-counterfeiting verification information used to verify its authenticity to form watermark payload data. Then, based on the image content characteristics of the original business image, the watermark embedding parameters are determined and a watermarked business image is generated, so that the image content and the on-site business information are bound together during the acquisition stage. At the same time, it provides a data foundation for subsequent authenticity verification, thereby improving the problems of insufficient guarantee of image content authenticity and weak reliability of image-on-site business information association in enterprise digital image records.

[0026] In this embodiment, the step of responding to an enterprise business image acquisition request, acquiring an original business image, and obtaining business association information associated with the enterprise business image acquisition request includes: Analyze the enterprise business image acquisition request to determine the corresponding business scenario type and business object identifier; A business information collection template is determined based on the business scenario type. The business information collection template includes at least two types of fields from the following categories: time field, location field, personnel field, equipment field, and business object field. Based on the aforementioned business information collection template, candidate business information is obtained from terminal collection information, logged-in user information, and enterprise business system information; The candidate business information is subjected to field integrity verification and field format standardization processing to obtain the business association information.

[0027] In one embodiment, in response to an enterprise business image acquisition request, original business images are acquired, and business association information associated with the enterprise business image acquisition request is obtained, which can be achieved in the following way.

[0028] First, the enterprise business image acquisition request is parsed to determine the corresponding business scenario type and business object identifier. The business scenario type characterizes the enterprise business category to which this image acquisition belongs, such as property inspection, work order recording, equipment maintenance, security monitoring, and evidence collection from objects thrown from heights. The business object identifier identifies the specific business object corresponding to this image acquisition, such as inspection point number, work order ID, equipment ID, camera ID, building identifier, floor identifier, or event identifier. In practice, the enterprise business image acquisition request may carry a page source identifier, business process node identifier, work order number, equipment QR code scan result, or inspection point location result, which the system can use to determine the business scenario type and business object identifier.

[0029] Secondly, a business information collection template is determined based on the business scenario type. This template includes at least two types of fields: time, location, personnel, equipment, and business object. Different business scenario types can correspond to different business information collection templates. For example, the business information collection template for a property inspection scenario may include fields such as shooting time, inspection location, inspection point, and inspection personnel; the business information collection template for a work order record scenario may include fields such as shooting time, shooting location, work order ID, work order type, work order title, and processing personnel; the business information collection template for an equipment maintenance scenario may include fields such as shooting time, equipment ID, maintenance personnel, maintenance location, and equipment status; the business information collection template for a security monitoring scenario may include shooting time, location, camera ID, or evidence collection personnel; and the business information collection template for a high-altitude object throwing scenario may include time, location, building, floor, or event number. Through template-based collection, business-related information can be matched with specific enterprise scenarios.

[0030] Then, based on the business information collection template, candidate business information is obtained from terminal-collected information, logged-in user information, and enterprise business system information. Terminal-collected information may include GPS latitude and longitude, network location information, device model, unique device identifier, device IP address, and system time output by the mobile terminal positioning module; logged-in user information may include the user ID, name, department, role, and permissions of the currently logged-in user; enterprise business system information may include work order ID, work order type, work order title, inspection point information, equipment ledger information, camera information, or task information. The system can extract corresponding fields from the above information sources according to the business information collection template to form candidate business information.

[0031] Finally, the candidate business information is subjected to field integrity verification and field format standardization to obtain the business association information. Field integrity verification may include verifying whether required fields are empty, whether the work order ID exists, whether the device ID matches, whether the location information is valid, and whether the personnel information is consistent with the currently logged-in user. Field format standardization may include unifying the time to a preset time format, unifying the latitude and longitude to a preset number of decimal places, parsing the address into standard address text, converting user information and device information into unified field names, and outputting business object information according to a preset encoding method. The business association information obtained after verification and standardization can be used for subsequent watermark payload data generation.

[0032] Through the above implementation methods, this application can determine the corresponding business information collection template according to different enterprise business scenarios, and collect candidate business information from terminals, logged-in users and enterprise business systems. Then, through integrity verification and standardization processing, business association information is obtained, so that watermarked business images can be reliably associated with specific business objects, personnel, locations and times, thereby improving the standardization and verifiability of enterprise business image records.

[0033] In this embodiment, generating watermark payload data based on the business association information includes: The business-related information is structurally encapsulated according to a preset field order to obtain structured business data; The structured business data is compressed and encoded to obtain business-encoded data; The business encoding data is subjected to error correction encoding processing to obtain the business watermark information; The anti-counterfeiting verification information is generated based on the structured business data, and the business watermark information and the anti-counterfeiting verification information are combined into the watermark payload data.

[0034] In one embodiment, watermark payload data is generated based on the business association information, which can be achieved in the following way.

[0035] First, the business-related information is structurally encapsulated according to a preset field order to obtain structured business data. This structured encapsulation can use JSON, XML, key-value pairs, TLV, or other parsable data formats. The preset field order may include a time field, a location field, a personnel field, a device field, and a business object field. For example, the time field may include the shooting time accurate to milliseconds; the location field may include GPS latitude, GPS longitude, address resolution results, and point-of-interest information; the personnel field may include user ID, name, department, and role; the device field may include device model, unique device identifier, and device IP address; and the business object field may include work order ID, work order type, work order title, device ID, or inspection point number. By using a unified field order for structured encapsulation, subsequent digest calculations, signature verification, and watermark parsing can have a consistent data foundation.

[0036] Secondly, the structured business data undergoes field compression and encoding conversion to obtain business-encoded data. Field compression may include deleting redundant field names, replacing field names with field numbers, performing dictionary compression on duplicate fields, limiting the length of text fields, or using binary serialization. Encoding conversion may include converting the structured business data to UTF-8 encoded data, Base64 encoded data, binary encoded data, or other data formats suitable for watermark embedding. Through field compression and encoding conversion, the length of the watermark payload can be reduced, enabling multi-dimensional business-related information to adapt to the image watermark embedding capacity.

[0037] Then, error correction coding is performed on the business coding data to obtain the business watermark information. Error correction coding can employ BCH codes, Reed-Solomon codes, convolutional codes, LDPC codes, or other error correction coding methods, or a combination of repetitive coding and check bits. The business watermark information after error correction coding has a certain degree of noise resistance, and even after the watermarked business image has undergone compression, transmission, scaling, or slight processing, it can still improve the reliability of watermark extraction and business information recovery.

[0038] Finally, the anti-counterfeiting verification information is generated based on the structured business data, and the business watermark information and the anti-counterfeiting verification information are combined into the watermark payload data. The anti-counterfeiting verification information can consist of at least one of the following: digest data, signature data, signature algorithm identifier, key identifier, timestamp identifier, or payload version information. During combination, the payload header, business watermark information, anti-counterfeiting verification information, and verification information can be set sequentially according to a preset payload format to obtain the final watermark payload data. This watermark payload data can be further converted into a 512-bit to 1024-bit binary watermark sequence, or configured to other lengths according to the image size and business field length.

[0039] Through the above implementation methods, this application converts business-related information into structured business data, which is then compressed, encoded, and error-corrected to form business watermark information. Combined with anti-counterfeiting verification information, watermark payload data is generated, enabling the watermark payload data to carry multi-dimensional enterprise business information and possess subsequent authenticity verification capabilities, thereby improving the integrity and reliability of the binding between business images and business information.

[0040] In this embodiment, the step of generating the anti-counterfeiting verification information based on the structured business data, and combining the business watermark information and the anti-counterfeiting verification information into the watermark payload data, includes: The structured business data is used to perform a summary calculation to obtain business summary data; The business digest data is digitally signed based on a preset signature key to obtain signed data. The anti-counterfeiting verification information is generated based on the signature data, digest algorithm identifier, and signature key identifier; The business watermark information, the anti-counterfeiting verification information, and the payload version information are combined to obtain the watermark payload data.

[0041] In one embodiment, the anti-counterfeiting verification information is generated based on the structured business data, and the business watermark information and the anti-counterfeiting verification information are combined into the watermark payload data, which can be achieved in the following way.

[0042] First, a digest calculation is performed on the structured business data to obtain business digest data. The digest calculation can use the SHA-256 algorithm, or other hash algorithms such as SM3, SHA-512, or others. Before the digest calculation, the structured business data can be normalized according to a preset field order to ensure that the same business-related information yields consistent digest data across different terminals or systems. The business digest data represents the content characteristics of the structured business data; when any field in the structured business data is modified, the recalculated digest data will change.

[0043] Secondly, the business digest data is digitally signed based on a preset signature key to obtain signed data. The preset signature key can be a private key allocated to the trusted watermarking service by the enterprise business system, or a private key in the terminal security module. The digital signature processing can use the RSA-SHA256 signature algorithm, with an RSA key length of 2048 bits; alternatively, SM2, ECDSA, or other asymmetric signature algorithms can be used. To improve key security, the preset signature key can be stored in a server key management system, a terminal security chip, a trusted execution environment, or an encryption module, preventing ordinary application layers from directly reading the private key.

[0044] Then, the anti-counterfeiting verification information is generated based on the signature data, the digest algorithm identifier, and the signature key identifier. The digest algorithm identifier indicates the algorithm type used to generate the business digest data, and the signature key identifier indicates the public key or certificate that the verification end should use. The anti-counterfeiting verification information may also include signature time, certificate serial number, signature version number, or signature validity period information. The verification end can recalculate the business digest data based on the digest algorithm identifier and obtain the corresponding public key based on the signature key identifier, thereby verifying the legitimacy of the signature data.

[0045] Finally, the business watermark information, the anti-counterfeiting verification information, and the payload version information are combined to obtain the watermark payload data. The payload version information identifies the data format version of the watermark payload data, enabling the system to parse different versions of watermark payload data during subsequent upgrades. The combined watermark payload data may include a payload version field, a business watermark field, an anti-counterfeiting verification field, an error correction field, and a verification field. After this watermark payload data is embedded in the original business image, business association information and anti-counterfeiting verification information can be simultaneously bound to the image content.

[0046] Through the above implementation methods, this application utilizes digest calculation and digital signature processing to generate anti-counterfeiting verification information, enabling business watermark information to not only represent business-related information, but also to confirm whether it has been forged or tampered with through signature verification; even if an attacker attempts to modify fields such as time, location, personnel, equipment, or work order, it will be difficult to generate legitimate anti-counterfeiting verification information, thereby improving the anti-counterfeiting capability and reliable evidence storage capability of enterprise business images.

[0047] In this embodiment, determining the watermark embedding parameters based on the image content features of the original business image includes: Image content analysis is performed on the original business image to obtain image content features including at least one of texture complexity, brightness distribution, and edge intensity; Based on the image content features, a set of candidate embedding regions is determined, and the embedding adaptation evaluation value corresponding to each candidate embedding region is determined. Based on the embedding adaptation evaluation value, the target embedding region is determined from the candidate embedding region set; The watermark embedding strength is determined based on the image content features of the target embedding region and the preset invisibility constraint. The target embedding region and the watermark embedding intensity are determined as the watermark embedding parameters.

[0048] In one embodiment, the watermark embedding parameters are determined based on the image content features of the original business image, which can be achieved in the following way.

[0049] First, image content analysis is performed on the original business image to obtain image content features including at least one of texture complexity, brightness distribution, and edge intensity. Specifically, the original business image can be converted to a grayscale space, a YCbCr color space, or another color space suitable for image analysis, and the image can be divided into blocks. For each image block, indices such as grayscale variance, gradient magnitude, edge density, mean brightness, brightness contrast, texture directionality, or local frequency distribution can be calculated to obtain image content features. Regions with more complex textures and richer edges are generally less sensitive to slight watermark perturbations and are more suitable as candidate embedding regions; regions with excessively low brightness, overexposure, or excessive flatness are usually prone to visible distortion and can have their embedding priority reduced.

[0050] Secondly, a set of candidate embedding regions is determined based on the image content features, and an embedding adaptation evaluation value is determined for each candidate embedding region. Candidate embedding regions can be image blocks, frequency domain sub-regions, color channel regions, or local regions divided according to semantic content. The embedding adaptation evaluation value can be determined by comprehensively considering texture complexity, brightness distribution, edge strength, local visual sensitivity, embedding capacity, and expected extraction stability. For images from property inspections, work order records, or equipment maintenance, the system can also reduce the embedding intensity of key visible areas to avoid affecting the viewing effect of equipment status, fault locations, work order sites, and other business content.

[0051] Then, based on the embedding fit evaluation value, a target embedding region is determined from the set of candidate embedding regions. The target embedding region can be multiple dispersed regions to improve the redundancy embedding capability and resistance to local tampering of the watermark payload data. Specifically, a preset number of candidate embedding regions can be selected according to the embedding fit evaluation value from high to low, or the number and distribution of the target embedding regions can be dynamically determined by combining the watermark payload length and error correction redundancy requirements.

[0052] Further, the watermark embedding strength is determined based on the image content features of the target embedding region and a preset invisibility constraint. The preset invisibility constraint may include peak signal-to-noise ratio (PSNR) constraints, structural similarity constraints, brightness variation constraints, color deviation constraints, or human visual perception constraints. For example, the PSNR between the watermarked image and the original image can be set to be greater than a preset value, such as 40 dB, or the structural similarity can be set to be greater than a preset threshold. In one implementation, a generative adversarial network (GAN) can be used to optimize the watermark embedding position and watermark embedding strength. The input to the GAN includes the original image, candidate embedding regions, image content features, and invisibility constraints, and the output includes the embedding weights or embedding strength adjustment coefficients for each candidate embedding region. This method allows the embedding strength to adaptively change with the image content.

[0053] Finally, the target embedding region and the watermark embedding strength are determined as the watermark embedding parameters. The watermark embedding parameters may also include embedding order, embedding channels, frequency domain coefficient index, number of repeated embeddings, and redundancy allocation rules. The watermark image generation process can write the watermark payload data into the original service image according to these parameters.

[0054] Through the above implementation methods, this application does not use fixed position or fixed strength to embed watermarks. Instead, it determines the watermark embedding parameters based on the image content features such as texture, brightness and edge of the original business image. It can also combine generative adversarial networks to achieve adaptive optimization of the embedding position and embedding strength, thereby reducing the impact on the visual quality of the image while ensuring the subsequent extractability of the watermark and improving the applicability of watermarked business images in enterprise business viewing and archiving.

[0055] In this embodiment, the step of embedding the watermark payload data into the original service image according to the watermark embedding parameters to generate a watermarked service image includes: The original service image is subjected to frequency domain transformation to obtain the original frequency domain data; Based on the target embedding region, determine the set of frequency domain coefficients to be embedded from the original frequency domain data; Based on the watermark embedding strength, the watermark sequence corresponding to the watermark payload data is written into the set of frequency domain coefficients to be embedded, thereby obtaining watermarked frequency domain data. The watermarked frequency domain data is subjected to inverse frequency domain transformation to obtain the watermarked service image.

[0056] In one embodiment, the watermark payload data is embedded into the original service image according to the watermark embedding parameters to generate a watermarked service image, which can be achieved in the following way.

[0057] First, the original service image is subjected to frequency domain transformation processing to obtain the original frequency domain data. Frequency domain transformation processing can employ Discrete Cosine Transform (DCT), Discrete Wavelet Transform (DWT), Fourier Transform, or other transformation methods suitable for digital watermark embedding. Preferably, the original service image can be subjected to block-based DCT transformation, for example, processed into 8×8 pixel blocks or other image blocks of different sizes, to obtain the frequency domain coefficients corresponding to each image block. DCT transformation can distribute image energy across different frequency components, facilitating the embedding of watermark payload data in frequency regions with lower visual sensitivity and better robustness.

[0058] Secondly, based on the target embedding region, a set of frequency domain coefficients to be embedded is determined from the original frequency domain data. This set can be selected from the mid-frequency region of the frequency domain data. Low-frequency coefficients typically correspond to the brightness of the main image subject and large-scale structures; excessive modification of low-frequency coefficients can easily affect the visual quality of the image. High-frequency coefficients are easily affected by compression, filtering, and noise processing. Mid-frequency coefficients can achieve a better balance between invisibility and robustness. Therefore, the frequency domain coefficients in the mid-frequency region can be preferentially selected as the set of frequency domain coefficients to be embedded. In specific implementations, the set of frequency domain coefficients to be embedded can be determined based on a preset mid-frequency coefficient index table, image block texture complexity, and watermark embedding strength.

[0059] Then, based on the watermark embedding strength, the watermark sequence corresponding to the watermark payload data is written into the set of frequency domain coefficients to be embedded, resulting in watermarked frequency domain data. The watermark sequence can be obtained from the watermark payload data through binary encoding, interleaving, scrambling, and error correction encoding. The writing method can include coefficient quantization modulation, coefficient difference modulation, coefficient sign modulation, or coefficient amplitude fine-tuning. For example, the binary watermark bit can be represented by adjusting the magnitude relationship between two intermediate frequency coefficients, or the watermark bit can be represented by adjusting the frequency domain coefficients to a preset quantization range. The watermark embedding strength is used to control the adjustment range of the frequency domain coefficients. If the adjustment range is too small, the watermark extraction may be unstable; if the adjustment range is too large, it may affect the image quality. Therefore, the writing can be performed according to the watermark embedding strength determined in the above embodiments.

[0060] Finally, the watermarked frequency domain data is subjected to inverse frequency domain transformation to obtain the watermarked business image. When using DCT transformation, an inverse DCT transformation can be performed on the watermarked frequency domain data to recover the spatial domain image. After recovery, the image pixel values ​​can be cropped, color space restored, image format encoded, and quality checked to ensure that the output image meets the upload and storage requirements of the enterprise business system. The watermark information in the watermarked business image is invisible or low-perceptibility, and does not cover the image surface as visible text, overlays, or textures, therefore it does not affect business personnel's ability to view the main content of the image.

[0061] Through the above implementation methods, this application embeds watermark payload data into the frequency domain data of the original service image, especially into the frequency domain coefficients in the mid-frequency region, so that the watermark information is not directly displayed on the image surface, while having good robustness. This method can reduce the problem of traditional visible watermarks obscuring image content or being cropped and removed, and provides a reliable foundation for extracting watermark payload data from the service image to be verified.

[0062] In this embodiment, after generating the watermarked business image, the method further includes: Based on the image identifier of the watermarked service image and the service association information, generate image service index information; A baseline watermark record is generated based on the watermark payload data or the summary information corresponding to the watermark payload data. The image service index information, the baseline watermark record, and the storage address of the watermarked service image are associated to obtain watermark verification baseline data. The watermark verification baseline data is stored in the enterprise business system or the watermark verification server.

[0063] In one embodiment, after generating the watermarked business image, watermark verification benchmark data can be further generated and stored to support subsequent image authenticity verification and business traceability.

[0064] First, image business index information is generated based on the image identifier of the watermarked business image and the business association information. The image identifier can be an image file ID, image hash value, image storage path, image upload serial number, or a unique record number assigned by the enterprise business system. The image business index information may include the image identifier, business scenario type, business object identifier, shooting time, photographer, work order ID, device ID, inspection point number, or other index fields used to retrieve the image. Using the image business index information, the business record corresponding to the watermarked business image can be quickly located within the enterprise business system.

[0065] Secondly, a baseline watermark record is generated based on the watermark payload data or the digest information corresponding to the watermark payload data. The baseline watermark record can directly store the watermark payload data, or it can store the hash digest, business digest data, signature data, public key identifier, payload version information, or error correction coding parameters of the watermark payload data. To reduce storage requirements and improve security, only the digest information and necessary verification parameters corresponding to the watermark payload data can be stored, without storing the complete business watermark information. During subsequent verification, the watermark data to be verified extracted from the business image to be verified can be compared with the baseline watermark record for consistency.

[0066] Then, a correlation is established between the image service index information, the baseline watermark record, and the storage address of the watermarked service image to obtain watermark verification baseline data. The storage address of the watermarked service image can be a file storage path, object storage address, database record address, or archive system address in the enterprise business system. The correlation can be stored in the enterprise business database, watermark verification database, or trusted evidence storage system. Through this correlation, the verification end can quickly retrieve the corresponding baseline watermark record based on the image identifier, service object identifier, or service field when it receives a service image to be verified.

[0067] Finally, the watermark verification baseline data is stored in the enterprise business system or the watermark verification server. The enterprise business system can be used to complete business process archiving, such as work order completion archiving, inspection record archiving, and equipment maintenance record archiving; the watermark verification server can be used to uniformly manage the watermark verification baseline data, public key information, algorithm version, and verification logs. To improve credibility, the watermark verification baseline data or its digest can also be synchronized to the enterprise audit system, log system, or trusted evidence storage system.

[0068] Through the above implementation method, this application simultaneously establishes watermark verification benchmark data after generating watermarked business images, so that a searchable association relationship is formed between watermarked business images, business-related information and benchmark watermark records; thus, after subsequent image circulation, compression, forwarding or archiving, authenticity verification and business traceability can still be performed based on watermark verification benchmark data, thereby improving the verification efficiency of enterprise business images and the reliability of evidence retention.

[0069] In this embodiment, after generating the watermarked business image, the method further includes: Obtain the image of the service to be verified, and perform watermark extraction processing on the image of the service to be verified to obtain the watermark data to be verified. Based on the image service index information corresponding to the service image to be verified, obtain the corresponding watermark verification benchmark data; Based on the watermark data to be verified and the watermark verification benchmark data, the watermark consistency evaluation result is determined; Based on the anti-counterfeiting verification information in the watermark data to be verified, the authenticity of the business watermark information in the watermark data to be verified is verified to obtain the anti-counterfeiting verification result. Based on the watermark consistency evaluation result and the anti-counterfeiting verification result, the authenticity verification result of the business image to be verified is determined.

[0070] In one embodiment, after generating the watermarked business image, the authenticity of the business image to be verified can also be verified.

[0071] First, the business image to be verified is acquired, and watermark extraction processing is performed on the image to obtain the watermark data to be verified. The business image to be verified can be an image downloaded from the enterprise business system, an image uploaded by a user for verification, an image obtained from the message flow link, or an image retrieved from the archive system. The watermark extraction processing can correspond to the watermark embedding processing. For example, when using DCT frequency domain embedding, the verification end can perform block-wise DCT transform on the business image to be verified, extract the watermark bits from the corresponding intermediate frequency coefficients, and then obtain the watermark data to be verified through descrambling, deinterleaving, error correction decoding, and data parsing. The watermark data to be verified can include the business watermark information to be verified and the anti-counterfeiting verification information to be verified.

[0072] Secondly, based on the image service index information corresponding to the image to be verified, the corresponding watermark verification benchmark data is obtained. The image service index information can come from the service object identifier parsed from the watermark data to be verified, or from the work order ID, device ID, inspection point number, or image file ID input by the user, or from the storage record of the image in the enterprise business system. The watermark verification server can query the corresponding watermark verification benchmark data based on the image service index information. This watermark verification benchmark data includes the benchmark watermark record, public key identifier, payload version information, or algorithm version information.

[0073] Then, based on the watermark data to be verified and the watermark verification benchmark data, the watermark consistency evaluation result is determined. Watermark consistency evaluation may include bit-by-bit comparison, hash comparison, post-error correction comparison, or normalized correlation calculation. For example, the normalized correlation score between the watermark data to be verified and the benchmark watermark record can be calculated. When the normalized correlation score is greater than a preset consistency threshold, the watermark consistency is determined to be high; when the normalized correlation score is less than or equal to the preset consistency threshold, the watermark consistency is determined to be low. The preset consistency threshold can be configured according to enterprise business requirements and image processing tolerance; for example, it can be set to 0.85, or it can be dynamically adjusted according to different image types or compression quality.

[0074] Furthermore, based on the anti-counterfeiting verification information in the watermark data to be verified, the authenticity of the business watermark information in the watermark data to be verified is verified to obtain the anti-counterfeiting verification result. Specifically, the digest of the business watermark information in the watermark data to be verified can be recalculated to obtain the digest data to be verified; then, the corresponding signature verification algorithm and public key can be obtained based on the digest algorithm identifier and signature key identifier in the anti-counterfeiting verification information; then, the signature data can be verified using the public key. If the signature verification passes, it means that the business watermark information in the watermark data to be verified has not been forged; if the signature verification fails, it means that the business watermark information may have been tampered with or forged.

[0075] Finally, based on the watermark consistency evaluation result and the anti-counterfeiting verification result, the authenticity verification result of the business image to be verified is determined. The authenticity verification result may include states such as no anomaly found, suspected anomaly, confirmed anomaly, and unable to verify. For example, when the watermark consistency evaluation result meets the preset requirements and the anti-counterfeiting verification result is a signature verification pass, it can be determined that no anomaly was found in the business image to be verified; when the watermark consistency evaluation result does not meet the preset requirements or the anti-counterfeiting verification result is a signature verification failure, it can be determined that the business image to be verified has an anomaly; when watermark extraction fails or watermark verification benchmark data is missing, it can be determined that the business image to be verified cannot be verified or requires manual review.

[0076] Through the above implementation methods, this application not only judges whether the watermark in the business image to be verified is consistent with the benchmark watermark record through watermark consistency evaluation, but also judges whether the business watermark information has been forged through anti-counterfeiting verification information, so as to combine image content verification and business information verification, thereby enabling rapid judgment of the authenticity and integrity of enterprise business images and improving the efficiency of abnormal image recognition.

[0077] In this embodiment, after determining the authenticity verification result of the business image to be verified based on the watermark consistency evaluation result and the anti-counterfeiting verification result, the method further includes: If the authenticity verification result indicates that the image to be verified is abnormal, the image to be verified is divided into blocks to obtain multiple image blocks to be detected. Local watermark extraction and local consistency detection are performed on each image block to be detected to obtain the local anomaly evaluation value corresponding to each image block to be detected. The set of abnormal image blocks is determined based on the local anomaly evaluation value corresponding to each image block to be detected; Based on the location distribution of the abnormal image block set in the service image to be verified, abnormal area identification information is generated.

[0078] In one embodiment, after determining the authenticity verification result of the service image to be verified, abnormal area identification information can be further determined.

[0079] First, if the authenticity verification result indicates that the image to be verified is abnormal, the image to be verified is divided into blocks to obtain multiple image blocks to be detected. The block division can be performed according to a fixed size, such as 8×8 pixel blocks, 16×16 pixel blocks, or 32×32 pixel blocks; alternatively, the image block size can be dynamically determined based on image resolution, compression method, and embedding parameters. Preferably, a block division rule consistent with the image block division method used in the watermark embedding process can be adopted to extract the corresponding local watermark responses in each image block.

[0080] Secondly, local watermark extraction and local consistency detection are performed on each image block to be detected to obtain the local anomaly evaluation value corresponding to each image block. Local watermark extraction can employ a local frequency domain extraction method corresponding to the watermark extraction process described in the above embodiments. For example, a DCT transform can be performed on each image block to be detected, and the local watermark bit can be extracted from the intermediate frequency coefficients of that image block, and compared with the corresponding reference watermark bit or redundant watermark information. Local consistency detection may include local normalized correlation calculation, local bit error rate calculation, local watermark extraction confidence calculation, or local error correction failure ratio calculation. The local anomaly evaluation value can be used to represent the probability that the corresponding image block has been tampered with or the watermark is damaged.

[0081] Then, based on the local anomaly evaluation values ​​corresponding to each image block to be detected, a set of abnormal image blocks is determined. Specifically, image blocks to be detected with local anomaly evaluation values ​​greater than a preset anomaly threshold can be identified as abnormal image blocks. Alternatively, connected component analysis can be performed by combining the anomaly evaluation values ​​of adjacent image blocks to filter out isolated noise points and retain continuous abnormal regions. Image blocks with local anomaly evaluation values ​​within the critical range can be marked as suspected abnormal image blocks for manual review or further verification.

[0082] Finally, based on the location distribution of the abnormal image patch set within the business image to be verified, abnormal region identification information is generated. This information may include the coordinates of the abnormal image patches, the bounding box of the abnormal region, the outline of the abnormal region, an anomaly probability map, or an anomaly heatmap. The anomaly heatmap can use different colors or grayscale levels to represent the probability of tampering; for example, red indicates a high-probability abnormal region, green indicates an area without detected anomalies, and yellow or orange indicates a suspected abnormal region. The abnormal region identification information can be overlaid on the business image to be verified or output as a separate verification result file to the enterprise business system.

[0083] Through the above implementation method, after detecting an anomaly in the business image to be verified, this application not only outputs the overall authenticity verification result, but also obtains the local anomaly evaluation value corresponding to each image block through block watermark consistency detection, and further generates anomaly area identification information, so that the enterprise business system can locate the image area suspected of being modified, thereby improving the efficiency of anomaly investigation and the ability to trace responsibility.

[0084] In one embodiment, an enterprise digital watermark image generation system is provided, comprising: The image and business information acquisition module is used to respond to an enterprise business image acquisition request, acquire original business images, and obtain business association information associated with the enterprise business image acquisition request; A watermark payload generation module is used to generate watermark payload data based on the business association information. The watermark payload data includes business watermark information used to characterize the business association information and anti-counterfeiting verification information used to verify the authenticity of the business watermark information. The embedding parameter determination module is used to determine the watermark embedding parameters based on the image content features of the original business image. The watermark image generation module is used to embed the watermark payload data into the original business image according to the watermark embedding parameters to generate a watermarked business image. The image verification module is used to extract watermarks from the business image to be verified, and to determine the authenticity verification result of the business image to be verified based on the extracted watermark data and the corresponding watermark verification benchmark data. The anomaly localization module is used to perform block watermark consistency detection on the service image to be verified when the authenticity verification result indicates that there is an anomaly in the service image to be verified, and determine the anomaly area identification information based on the detection result.

[0085] The specific implementation details of each module can be found in the above description of the limitations of the enterprise digital watermark image generation method, and will not be repeated here.

[0086] In one embodiment, an enterprise digital watermark image generation and verification system is provided. This system can be deployed on a mobile terminal, an enterprise business server, a watermark verification server, or a combination of a mobile terminal and a server. The system includes an image and business information acquisition module, a watermark payload generation module, an embedding parameter determination module, a watermark image generation module, an image verification module, and an anomaly location module.

[0087] The image and business information acquisition module is used to respond to enterprise business image acquisition requests, acquire raw business images, and obtain business-related information associated with the enterprise business image acquisition request. This module can call the mobile terminal's camera, positioning module, system time module, and device information interface, and interact with the enterprise business system's user login module, work order module, inspection module, or equipment ledger module to obtain information such as shooting time, shooting location, personnel identity, acquisition device, and business object. For property inspection scenarios, this module can obtain inspection points, inspection routes, inspection personnel, and inspection time; for work order recording scenarios, this module can obtain work order ID, work order type, work order title, and processing personnel; for equipment maintenance scenarios, this module can obtain equipment ID, maintenance personnel, and maintenance location.

[0088] The watermark payload generation module generates watermark payload data based on the business association information. The watermark payload data includes business watermark information characterizing the business association information and anti-counterfeiting verification information verifying the authenticity of the business watermark information. This module can perform structured encapsulation, field compression, encoding conversion, and error correction encoding on the business association information to generate business watermark information; it can also perform digest calculation and digital signature processing on the structured business data to generate anti-counterfeiting verification information. The digital signature can use the RSA-SHA256 algorithm with a key length of 2048 bits, or other signature algorithms that meet enterprise security requirements. The signature private key can be stored in a server key management system, a terminal security module, or a trusted execution environment, while the signature verification public key can be configured in an image verification module or a watermark verification server.

[0089] The embedding parameter determination module is used to determine watermark embedding parameters based on the image content features of the original business image. This module can perform texture complexity analysis, brightness distribution analysis, and edge intensity analysis on the original business image to determine the target embedding region and watermark embedding intensity. In one implementation, this module may include an embedding optimization unit based on a generative adversarial network. This embedding optimization unit outputs embedding position weights or embedding intensity adjustment coefficients based on image content features, candidate embedding regions, and invisibility constraints, ensuring that the watermarked business image maintains good watermark extraction stability while meeting preset invisibility requirements.

[0090] The watermarked image generation module is used to embed the watermark payload data into the original service image according to the watermark embedding parameters, thereby generating a watermarked service image. This module can encode the watermark payload data into a watermark sequence, perform a DCT frequency domain transform on the original service image, embed the watermark sequence into the frequency domain coefficients in the mid-frequency region, and then generate the watermarked service image through an inverse DCT transform. The watermarked image generation module can also perform image quality detection on the watermarked service image, and readjust the watermark embedding strength or embedding area if the image quality does not meet preset requirements.

[0091] The image verification module extracts watermarks from the image to be verified and determines the authenticity verification result of the image based on the extracted watermark data and the corresponding watermark verification benchmark data. This module can perform frequency domain transformation on the image to be verified, extract the watermark data from the corresponding frequency domain coefficients, and obtain the watermark verification benchmark data based on the image service index information. The image verification module can calculate the watermark consistency evaluation result between the watermark data to be verified and the benchmark watermark record, such as the normalized correlation score; it can also verify the signature based on the signature data, public key identifier, and digest algorithm identifier in the anti-counterfeiting verification information to obtain the anti-counterfeiting verification result. Subsequently, the image verification module can output the authenticity verification result by combining the watermark consistency evaluation result and the anti-counterfeiting verification result.

[0092] The anomaly localization module is used to perform block-based watermark consistency detection on the business image to be verified when the authenticity verification result indicates that the image to be verified is abnormal, and to determine the anomaly region identification information based on the detection results. This module can divide the business image to be verified into multiple image blocks to be detected, such as 8×8 pixel blocks, and perform local watermark extraction and local consistency detection on each image block to obtain a local anomaly evaluation value. The anomaly localization module can determine the set of abnormal image blocks based on the local anomaly evaluation values ​​and generate anomaly region coordinates, anomaly region contours, or anomaly heatmaps. Anomaly heatmaps can be used to visually display potentially tampered areas in an image within an enterprise business system.

[0093] Through the above implementation methods, the enterprise digital watermark image generation and verification system of this application can complete the acquisition of business-related information, watermark payload generation, adaptive watermark embedding, and generation of watermarked business images during the image acquisition stage, and complete watermark extraction, anti-counterfeiting verification of business watermark information, watermark consistency evaluation, and abnormal area location during the verification stage. As a result, the system can simultaneously improve the reliability of the association between enterprise business images and on-site business information, the efficiency of image authenticity verification, and the ability to identify abnormal areas, and is suitable for enterprise digital business scenarios such as property inspection, work order recording, equipment maintenance, and security evidence collection.

[0094] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

Claims

1. A method for generating digital watermarked images for enterprises, characterized in that, The method includes: In response to an enterprise business image acquisition request, the system acquires original business images and obtains business association information associated with the enterprise business image acquisition request. Watermark payload data is generated based on the business association information. The watermark payload data includes business watermark information used to characterize the business association information and anti-counterfeiting verification information used to verify the authenticity of the business watermark information. The watermark embedding parameters are determined based on the image content features of the original business image; According to the watermark embedding parameters, the watermark payload data is embedded into the original service image to generate a watermarked service image; The step of generating watermark payload data based on the business association information includes: The business-related information is structurally encapsulated according to a preset field order to obtain structured business data; The structured business data is compressed and encoded to obtain business-encoded data; The business encoding data is subjected to error correction encoding processing to obtain the business watermark information; The structured business data is used to perform a summary calculation to obtain business summary data; The business digest data is digitally signed based on a preset signature key to obtain signed data. The anti-counterfeiting verification information is generated based on the signature data, digest algorithm identifier, and signature key identifier; The business watermark information, the anti-counterfeiting verification information, and the payload version information are combined to obtain the watermark payload data; The step of determining the watermark embedding parameters based on the image content features of the original business image includes: Image content analysis is performed on the original business image to obtain image content features including at least one of texture complexity, brightness distribution, and edge intensity; Based on the image content features, a set of candidate embedding regions is determined, and the embedding adaptation evaluation value corresponding to each candidate embedding region is determined. Based on the embedding adaptation evaluation value, the target embedding region is determined from the candidate embedding region set; The watermark embedding strength is determined based on the image content features of the target embedding region and the preset invisibility constraint. The target embedding region and the watermark embedding intensity are determined as the watermark embedding parameters; The step of embedding the watermark payload data into the original service image according to the watermark embedding parameters to generate a watermarked service image includes: The original service image is subjected to frequency domain transformation to obtain the original frequency domain data; Based on the target embedding region, determine the set of frequency domain coefficients to be embedded from the original frequency domain data; Based on the watermark embedding strength, the watermark sequence corresponding to the watermark payload data is written into the set of frequency domain coefficients to be embedded, thereby obtaining watermarked frequency domain data. The watermarked frequency domain data is subjected to inverse frequency domain transformation to obtain the watermarked service image.

2. The method for generating digital watermarked images for enterprises according to claim 1, characterized in that, The step of responding to an enterprise business image acquisition request by acquiring original business images and obtaining business association information associated with the enterprise business image acquisition request includes: Analyze the enterprise business image acquisition request to determine the corresponding business scenario type and business object identifier; A business information collection template is determined based on the business scenario type. The business information collection template includes at least two types of fields from the following categories: time field, location field, personnel field, equipment field, and business object field. Based on the aforementioned business information collection template, candidate business information is obtained from terminal collection information, logged-in user information, and enterprise business system information; The candidate business information is subjected to field integrity verification and field format standardization processing to obtain the business association information.

3. The method for generating digital watermarked images for enterprises according to claim 1, characterized in that, After generating the watermarked business image, the method further includes: Based on the image identifier of the watermarked service image and the service association information, generate image service index information; A baseline watermark record is generated based on the watermark payload data or the summary information corresponding to the watermark payload data. The image service index information, the baseline watermark record, and the storage address of the watermarked service image are associated to obtain watermark verification baseline data. The watermark verification baseline data is stored in the enterprise business system or the watermark verification server.

4. The method for generating digital watermarked images for enterprises according to claim 3, characterized in that, After generating the watermarked business image, the method further includes: Obtain the image of the service to be verified, and perform watermark extraction processing on the image of the service to be verified to obtain the watermark data to be verified. Based on the image service index information corresponding to the service image to be verified, obtain the corresponding watermark verification benchmark data; Based on the watermark data to be verified and the watermark verification benchmark data, the watermark consistency evaluation result is determined; Based on the anti-counterfeiting verification information in the watermark data to be verified, the authenticity of the business watermark information in the watermark data to be verified is verified to obtain the anti-counterfeiting verification result. Based on the watermark consistency evaluation result and the anti-counterfeiting verification result, the authenticity verification result of the business image to be verified is determined.

5. The method for generating a digital watermark image for an enterprise according to claim 4, characterized in that, After determining the authenticity verification result of the business image to be verified based on the watermark consistency evaluation result and the anti-counterfeiting verification result, the method further includes: If the authenticity verification result indicates that the image to be verified is abnormal, the image to be verified is divided into blocks to obtain multiple image blocks to be detected. Local watermark extraction and local consistency detection are performed on each image block to be detected to obtain the local anomaly evaluation value corresponding to each image block to be detected. The set of abnormal image blocks is determined based on the local anomaly evaluation value corresponding to each image block to be detected; Based on the location distribution of the abnormal image block set in the service image to be verified, abnormal area identification information is generated.

6. A digital watermark image generation and verification system for enterprises, used to perform the method according to any one of claims 1 to 5, characterized in that, include: The image and business information acquisition module is used to respond to an enterprise business image acquisition request, acquire original business images, and obtain business association information associated with the enterprise business image acquisition request; A watermark payload generation module is used to generate watermark payload data based on the business association information. The watermark payload data includes business watermark information used to characterize the business association information and anti-counterfeiting verification information used to verify the authenticity of the business watermark information. The embedding parameter determination module is used to determine the watermark embedding parameters based on the image content features of the original business image. The watermark image generation module is used to embed the watermark payload data into the original business image according to the watermark embedding parameters to generate a watermarked business image. The image verification module is used to extract watermarks from the business image to be verified, and to determine the authenticity verification result of the business image to be verified based on the extracted watermark data and the corresponding watermark verification benchmark data. An anomaly localization module is used to perform block watermark consistency detection on the image to be verified when the authenticity verification result indicates that the image to be verified is abnormal, and to determine the abnormal area identification information based on the detection result. Specifically, the watermark payload generation module is used for: structurally encapsulating the business-related information according to a preset field order to obtain structured business data; compressing and encoding the structured business data to obtain business encoded data; performing error correction encoding on the business encoded data to obtain the business watermark information; performing digest calculation on the structured business data to obtain business digest data; performing digital signature processing on the business digest data based on a preset signature key to obtain signature data; generating the anti-counterfeiting verification information according to the signature data, digest algorithm identifier, and signature key identifier; and combining the business watermark information, the anti-counterfeiting verification information, and payload version information to obtain the watermark payload data. The embedding parameter determination module is specifically used for: performing image content analysis on the original business image to obtain image content features including at least one of texture complexity, brightness distribution, and edge intensity; determining a set of candidate embedding regions based on the image content features, and determining the embedding adaptation evaluation value corresponding to each candidate embedding region; determining a target embedding region from the set of candidate embedding regions based on the embedding adaptation evaluation value; determining the watermark embedding intensity based on the image content features of the target embedding region and a preset invisibility constraint; and determining the target embedding region and the watermark embedding intensity as the watermark embedding parameters. The watermarked image generation module is specifically used for: performing frequency domain transformation processing on the original service image to obtain original frequency domain data; determining the set of frequency domain coefficients to be embedded from the original frequency domain data according to the target embedding region; writing the watermark sequence corresponding to the watermark payload data into the set of frequency domain coefficients to be embedded according to the watermark embedding strength to obtain watermarked frequency domain data; and performing inverse frequency domain transformation processing on the watermarked frequency domain data to obtain the watermarked service image.

Images