An anti-interference standard data traceability fingerprint embedding and identification method and system

Abstract

This invention relates to the fields of data processing and information security technology, and discloses a method and system for embedding and identifying source fingerprints of standard data with anti-interference capabilities. The method first models the layout boundaries, partition boundaries, field areas, table cell areas, or predefined anchor areas of the standard data to obtain multiple standardized target areas and corresponding area indices. Then, it generates region-level synchronous index reference information based on the spatial distribution relationship of each standardized target area, and embeds it into the corresponding areas after binding it with the corresponding fingerprint fragments. During identification, interference analysis is performed on the standard data to be identified. Based on the region-level synchronous index reference information, the correspondence between the area index and spatial location of each standardized target area is restored. After completing the area reconstruction, the corresponding fingerprint fragments are extracted, and index consistency verification and source matching are performed, thereby improving the stability and accuracy of source identification under conditions of cropping, rotation, scaling, resampling, compression, screenshotting, or printing / scanning.

Description

Technical Field

[0001] This invention relates to the fields of data processing and information security technology, and in particular to an anti-interference standard data traceability fingerprint embedding and identification method and system. Background Technology

[0002] With the widespread application of standard data in scenarios such as invoice circulation, certificate management, form transmission, report archiving, formatted document distribution, and electronic document exchange, embedding traceability information into standard data for source identification and accountability has become a crucial technical requirement in data security management and data flow control. To distinguish the propagation path, source of use, or distribution recipient of standard data, existing technologies typically employ digital watermarking, digital fingerprinting, or implicit identification to embed specific identification information into the data carrier. This identification information is then extracted and judged in subsequent identification stages to achieve data traceability.

[0003] Most existing traceability information embedding and recognition schemes revolve around images, watermarked documents, or security documents. Some schemes, by setting synchronization markers, synchronization blocks, calibration patterns, or reference signals in the carrier, allow the recognition end to correct for rotation, scaling, cropping, or other geometric distortions on the carrier before detecting traceability information, thereby restoring the embedding position of the watermark or fingerprint before completing the recognition. This type of scheme improves the detection capability of the carrier after geometric transformation to a certain extent and has become a common technical approach in watermark recognition against geometric interference.

[0004] However, existing technologies still have significant limitations in standard data scenarios. Standard data typically has predefined layout boundaries, field areas, table cell areas, text and image column areas, or other regularized structural areas. These areas not only have spatial relationships but also relatively stable structural correspondences. Existing synchronization markers or calibration signals mostly focus on restoring the overall geometric orientation of the image or page, but are insufficient in restoring the correspondences between multiple structural areas within standard data. Even after standard data has undergone transmission, compression, resampling, screenshotting, printing, scanning, or partial cropping, and overall geometric correction is achieved, local target area misalignment, area boundary shifts, or inaccurate embedded area recognition may still occur. This can lead to failed fingerprint extraction, increased false negative rates, or misjudgments when local areas are damaged.

[0005] Furthermore, most existing traceability solutions embed the identification information as a whole, extracting it only during the identification process. This approach is prone to errors when standard data is partially obscured, partially replaced, cropped, or processed multiple times. Distortion in local areas can negatively impact the overall identification result, leading to insufficient stability. This is especially true when standard data has multiple regions, regular distributions, fixed fields, or fixed formats. Without combining the regional structural characteristics of the standard data itself to establish a more stable correspondence between the embedding and identification positions of the traceability information, it becomes difficult to balance anti-interference capabilities and identification accuracy.

[0006] Therefore, how to construct an anti-interference fingerprint embedding and recognition scheme that can still recover the correspondence between the fingerprint embedding position and the recognition position under interference conditions such as cropping, rotation, scaling, resampling, compression, screenshotting or printing scanning, based on the structured regional features of standard data, has become a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0007] The purpose of this invention is to provide a standard data source fingerprint embedding and identification method and system with anti-interference capabilities. This method and system address the technical problems in the prior art where standard data, after being subjected to interference such as cropping, rotation, scaling, resampling, compression, screenshotting, or printing and scanning, is prone to loss of synchronization between the source fingerprint embedding position and the identification position, difficulty in stably reconstructing the target area, decreased accuracy of fingerprint extraction, and insufficient stability of source identification. This invention achieves stable embedding of source fingerprints in standard data, restoration of regional correspondence, and accurate identification, thereby improving the source traceability reliability and identification robustness of standard data under complex interference conditions.

[0008] To achieve the above-mentioned technical objectives, the present invention adopts the following technical solution:

[0009] An interference-resistant standard data traceability fingerprint embedding and recognition system, comprising:

[0010] The standard data modeling module is used to acquire the standard data to be processed, and to establish a regionalized structure model of the standard data based on the layout boundaries, partition boundaries, field areas, table cell areas or predefined anchor areas, so as to obtain multiple standardized target areas and regional indexes of each standardized target area.

[0011] The regional-level synchronization index reference generation module is used to generate regional-level synchronization index reference information based on the spatial distribution relationship of the multiple standardized target regions. The regional-level synchronization index reference information is used to characterize the correspondence between the regional index and spatial location of each standardized target region.

[0012] A fingerprint fragment construction module is used to generate traceability fingerprint information based on the source identifier and to split the traceability fingerprint information into multiple fingerprint fragments corresponding to different region indices.

[0013] The binding and embedding module is used to bind the regional-level synchronization index reference information with the corresponding fingerprint fragments according to the regional index and embed them into the corresponding standardized target regions to obtain the embedded standard data.

[0014] The interference analysis and region recovery module is used to analyze the interference type of the standard data to be identified, and to recover the correspondence between the regional index and spatial location of each standardized target region in the standard data to be identified based on the regional-level synchronous index reference information, so as to complete the reconstruction of the standardized target region.

[0015] The fragment extraction module is used to extract the corresponding fingerprint fragments from the corresponding standardized target region according to the recovered region index after the standardized target region reconstruction is completed.

[0016] The consistency identification module is used to perform index consistency verification and source matching on multiple extracted fingerprint fragments to output the source identification results;

[0017] The regional-level synchronization index reference information and the fingerprint fragment are bound together based on the same regional index, so that after the standard data to be identified is subjected to cropping, rotation, scaling, resampling, compression, screenshotting or printing scanning interference, the regional index of the standardized target area can be recovered first, and then the corresponding fingerprint fragment can be extracted for source identification.

[0018] Optionally, the regionalized structure model established by the standard data modeling module includes at least one of the following regions: page boundary region, header and footer region, field carrying region, table cell region, text and image column region, or predefined anchor region.

[0019] Optionally, the regional-level synchronization index reference information generated by the regional-level synchronization index reference generation module includes at least one of the following: regional boundary reference information, regional center reference information, regional adjacency relationship reference information, regional scale reference information, or regional hierarchical reference information.

[0020] Optionally, the binding embedding module is used to set different embedding strengths, embedding densities, redundancy levels or priorities for the standardized target regions corresponding to different region indices, so that when some standardized target regions of the standard data to be identified are damaged, the fingerprint fragments in the remaining standardized target regions can still be used for source identification.

[0021] Optionally, the consistency identification module is used to perform regional index consistency verification, fragment integrity assessment and cross-regional matching result fusion on the extracted multiple fingerprint fragments before source matching, and output the final source identification result based on the fusion result.

[0022] To achieve the above-mentioned technical objectives, the present invention also adopts the following technical solution:

[0023] An interference-resistant standard data tracing fingerprint embedding and recognition method, comprising:

[0024] S1: Obtain the standard data to be processed, and establish a regional structure model based on the layout boundary, partition boundary, field area, table cell area or predefined anchor area of ​​the standard data to obtain multiple standardized target areas and the regional index of each standardized target area;

[0025] S2: Based on the spatial distribution relationship of the multiple standardized target regions, generate regional-level synchronous index reference information to characterize the correspondence between the regional index and spatial location of each standardized target region;

[0026] S3: Generate traceability fingerprint information based on the source identifier, and split the traceability fingerprint information into multiple fingerprint fragments corresponding to different region indices;

[0027] S4: According to the regional index, the regional-level synchronization index reference information is bound to the corresponding fingerprint fragment and then embedded into the corresponding standardized target region to obtain the embedded standard data;

[0028] S5: Obtain the standard data to be identified, and perform interference type analysis on the standard data to be identified;

[0029] S6: Based on the regional-level synchronous index reference information, restore the correspondence between the regional index and spatial location of each standardized target region in the standard data to be identified, so as to complete the reconstruction of the standardized target region;

[0030] S7: Extract the corresponding fingerprint fragment from the corresponding standardized target region according to the recovered region index;

[0031] S8: Perform index consistency verification and source matching on the extracted multiple fingerprint fragments, and output the source identification results;

[0032] Specifically, by establishing a binding relationship between regional-level synchronization index reference information and fingerprint fragments based on the same regional index, the standard data to be identified can first recover the regional index of the standardized target area after being interfered with, and then extract the corresponding fingerprint fragments for source identification.

[0033] Optionally, the step S1 of establishing a regionalized structural model based on the standard data includes:

[0034] Identify page boundaries, partition boundaries, field boundaries, table cell boundaries, text / image column boundaries, or predefined anchor area boundaries in the standard data, and generate standardized target areas and corresponding area indexes based on the positional relationships between each boundary.

[0035] Optionally, the generation of region-level synchronization index reference information based on the spatial distribution relationship of the multiple standardized target regions in step S2 includes:

[0036] Based on the center location, boundary location, adjacency relationship, scale relationship or hierarchical relationship of multiple standardized target areas, corresponding regional boundary reference information, regional center reference information, regional adjacency relationship reference information, regional scale reference information or regional hierarchical reference information are generated.

[0037] Optionally, the interference type analysis of the standard data to be identified in S5 includes:

[0038] Identify at least one of the following interferences affecting the standard data to be identified: cropping, rotation, scaling, resampling, compression, screenshotting, or printing / scanning; and determine the corresponding region recovery processing path based on the identification results.

[0039] Optionally, in step S8, the index consistency check and source matching of the extracted multiple fingerprint fragments are performed, and the source tracing and identification results are output, including:

[0040] Each fingerprint fragment is checked to see if its corresponding region index is consistent with the recovered region index. The results of consistency, fragment integrity and fragment matching are then used to make a fusion judgment to output the final source identification result.

[0041] The main advantages of this invention compared to existing technologies are as follows:

[0042] This invention establishes a regionalized structural model of standard data, generates regional-level synchronization index reference information corresponding to each standardized target region, and embeds the regional-level synchronization index reference information with the corresponding fingerprint fragments into the corresponding standardized target regions. This allows the standard data to be identified to still prioritize the recovery of the correspondence between the regional index and spatial location of each standardized target region after being subjected to interference such as cropping, rotation, scaling, resampling, compression, screenshotting, or printing scanning. Then, the corresponding fingerprint fragments are extracted for source identification. This effectively solves the problem in the prior art that it can only perform overall geometric correction and is difficult to stably recover the correspondence between the regions within the standard data. It improves the relocation capability and recognition stability of traceable fingerprints under complex interference conditions.

[0043] This invention splits the traceability fingerprint information corresponding to the source identifier into multiple fingerprint segments corresponding to different regional indices, and processes them using a regional binding embedding method. This makes the traceability information in the standard data no longer dependent on a single region or a whole continuous region for carrying. Therefore, when some standardized target regions are partially cut off, occluded, replaced or damaged, the fingerprint segments in the remaining undamaged regions can still participate in the identification, which helps to improve the resistance to local damage in traceability identification.

[0044] In the identification stage, this invention does not directly perform a single match on the extracted fingerprint fragments. Instead, it first performs a region index consistency check on multiple fingerprint fragments, and then combines the fragment integrity and matching results for fusion judgment. This can reduce the risk of misidentification caused by local region misalignment, fragment distortion or interference residue, and improve the accuracy and reliability of the final source determination result.

[0045] The regionalized structural modeling method adopted in this invention can adapt to various standard data objects with layout boundaries, field areas, table cell areas, text and image column areas, or predefined anchor areas. Therefore, this invention is not only applicable to a single type of data carrier, but can also be applied to invoices, certificates, forms, reports, layout documents, and other regularized page data scenarios, and has good universality and engineering application value.

[0046] This invention modularizes the construction of regional-level synchronous index reference information, fingerprint fragment generation and binding embedding, interference analysis, regional recovery, fragment extraction and consistency recognition, making the processing links between the embedding end and the recognition end clear and the functional boundaries well-defined. This facilitates flexible configuration of embedding strength, redundancy level, recovery strategy and recognition strategy according to different application scenarios, which is beneficial to improving the scalability and practicality of system deployment. Attached Figure Description

[0047] Figure 1 This is a structural diagram of the fingerprint embedding and recognition system of the present invention;

[0048] Figure 2 This is a flowchart illustrating the fingerprint embedding and recognition method of the present invention. Detailed Implementation

[0049] The specific embodiments of the present invention will be further described in detail below with reference to the technical solution of the present invention. It should be understood that the following specific embodiments are only used to explain the present invention and are not intended to limit the scope of protection of the present invention. Equivalent substitutions, simple modifications, or conventional adjustments made by those skilled in the art based on the disclosure of the present invention without departing from the concept of the present invention should all fall within the scope of protection of the present invention.

[0050] In this embodiment, "standard data" refers to data objects with a regular layout structure, predefined partitioning relationships, or stable region boundaries. The standard data can be invoice page data, certificate page data, form page data, report page data, formatted document page data, electronic document page data, or other data objects with fixed layout characteristics. This type of standard data typically has one or more stable regions such as page boundaries, header / footer areas, field-carrying areas, table cell areas, text / image column areas, and predefined anchor areas. Therefore, it can be modeled with a regional structure, and traceability fingerprint embedding and recognition processing can be performed at the region level.

[0051] This invention provides an anti-interference standard data source fingerprint embedding and recognition system. The system can be broadly divided into two parts: an embedding end and a recognition end. The embedding end is responsible for completing regional modeling, constructing region-level synchronous index reference information, generating fingerprint fragments, and binding embedding based on the regional structure of the standard data. The recognition end is responsible for performing interference analysis, region restoration, fragment extraction, and consistency recognition on the standard data to be identified after it has been subjected to interference such as cropping, rotation, scaling, resampling, compression, screenshotting, or printing / scanning, in order to output the final source identification result. Therefore, this invention does not merely restore the geometric pose at the overall image level, but rather uses each standardized target region within the standard data as the processing object, restoring the correspondence between the region index and spatial position of each region, and then performing fragment-level fingerprint extraction and source determination, thereby improving the stability of source identification under complex interference scenarios.

[0052] In this embodiment, such as Figure 1 As shown, the anti-interference standard data traceability fingerprint embedding and recognition system includes a standard data modeling module, a regional-level synchronous index reference generation module, a fingerprint fragment construction module, a binding embedding module, an interference analysis and region recovery module, a fragment extraction module, and a consistency recognition module. The affiliation, function, input / output, and system location of each module are as follows.

[0053] The standard data modeling module is an embedded preprocessing module and can also be used as an auxiliary analysis module on the recognition end before reconstructing the standard data to be recognized. Its main function is to acquire the standard data to be processed and, based on the layout boundaries, partition boundaries, field areas, table cell areas, or predefined anchor areas of the standard data, establish a regionalized structure model for the standard data to obtain multiple standardized target areas and regional indexes for each standardized target area.

[0054] The input to the standard data modeling module is the original standard data page. This original standard data page can be a rendered page image of an electronic document, a structured page object, a scanned page image, or a printed scanned return page image. The output of the standard data modeling module is a regionalized structural model, which includes at least: boundary information, center location information, adjacency relationship information, hierarchical relationship information, and corresponding region indexes for multiple standardized target regions.

[0055] In a preferred embodiment, the standard data modeling module first performs layout standardization processing on the input page, such as page orientation correction, unified resolution processing, boundary trimming, and noise pre-suppression. Then, based on the regular layout features of the page, it identifies one or more regions among page boundary regions, header / footer regions, field-carrying regions, table cell regions, text / image column regions, and predefined anchor areas, and establishes a regionalized structural model based on the spatial relationships of each region on the page. For example, the entire page can be divided into a top information area, a middle content area, and a bottom signature area; the middle content area can be further divided into field row areas, table cell areas, or left / right column areas; and several anchor areas can be predefined according to the business template as priority reference areas for subsequently building regional-level synchronous index reference information.

[0056] In this module, the region index is not limited to simple sequential numbering. Preferably, the region index can be generated using a hierarchical indexing method, such as a multi-level encoding structure of "page-level index—partition-level index—sub-region-level index". This method ensures that different standardized target regions not only have unique identifiers but also clear hierarchical relationships, providing a foundation for subsequent construction of region-level synchronous index reference information and fragment-level source identification.

[0057] The regional-level synchronization index reference generation module belongs to the embedded reference construction module. Its function is to generate regional-level synchronization index reference information based on the spatial distribution relationship of the multiple standardized target regions. The regional-level synchronization index reference information is used to characterize the correspondence between the regional index and spatial location of each standardized target region.

[0058] The input to the regional-level synchronous index reference generation module is the regionalized structural model output by the standard data modeling module, and the output is regional-level synchronous index reference information. The regional-level synchronous index reference information may include at least one of the following: regional boundary reference information, regional center reference information, regional adjacency relationship reference information, regional scale reference information, or regional hierarchical reference information.

[0059] In one specific embodiment, if the standard data is a page with fixed fields and a table layout, then region-level synchronous index reference information can be generated based on the boundary coordinates, center coordinates, and adjacency relationships of each field area and table cell area. This reference information can be used to characterize the relative position, size ratio, and adjacency topology of each target area with adjacent areas. After the data is affected by rotation, scaling, or cropping, the recognition end can prioritize restoring the region index of each target area based on this region-level synchronous index reference information, rather than simply restoring the overall page orientation.

[0060] In another embodiment, if predefined anchor zones are set in the standard data, the boundary and center positions of the anchor zones can be used to generate region boundary reference information and region center reference information. Since the anchor zones are usually located in relatively stable structural positions, they can serve as the priority alignment basis when the identification end performs region restoration.

[0061] It should be noted that the region-level synchronization index reference information in this invention is not a simple global synchronization pattern, nor is it ordinary calibration information used only to recover the overall geometric attitude. Instead, it is structured index reference information that is associated with and corresponds to multiple standardized target regions within the standard data. It directly serves the subsequent index recovery of standardized target regions and fragment-level fingerprint extraction.

[0062] The fingerprint fragment construction module belongs to the embedded fingerprint generation module. Its function is to generate traceability fingerprint information based on the source identifier and split the traceability fingerprint information into multiple fingerprint fragments corresponding to different region indices.

[0063] The fingerprint fragment construction module takes a source identifier as its input, which can be a user identifier, distribution object identifier, version identifier, terminal identifier, circulation batch identifier, or other identification information that can be used for traceability. The output consists of multiple fingerprint fragments corresponding to different region indices.

[0064] In a preferred embodiment, the fingerprint fragment construction module first generates an original traceability fingerprint string based on the source identifier, and then splits the original traceability fingerprint string into multiple fingerprint fragments according to the number of region indexes. Each fingerprint fragment can be formed using an equal-division method, or its length can be set according to region importance, region stability, or business scenario requirements. For target regions with stable structures and easier recovery after interference, a higher information carrying weight can be assigned, so that the corresponding fingerprint fragment contains more recognition bits; for regions that are easily affected by local cropping or occlusion, a lower information carrying weight can be assigned.

[0065] Furthermore, to improve the tolerance of the recognition end to local damage, the fingerprint fragment construction module can also add redundant check bits, consistency check bits, or fragment sequence identification information to some or all fingerprint fragments, so that the subsequent recognition stage can still complete source matching and consistency judgment based on the remaining fragments when some fragments are missing or distorted.

[0066] The binding and embedding module belongs to the embedding execution module. Its function is to bind the regional-level synchronization index reference information and the corresponding fingerprint fragments according to the regional index and embed them into the corresponding standardized target regions to obtain the embedded standard data.

[0067] The module takes as input regional-level synchronization index reference information, multiple fingerprint fragments, and standardized target regions corresponding one-to-one with each regional index; the output is the embedded standard data. The binding embedding module is located at the end of the embedding execution chain and is the key module for implementing the aforementioned reference construction and fingerprint generation results into specific page regions.

[0068] In this invention, "binding" refers to establishing a correspondence between region-level synchronization index reference information and corresponding fingerprint fragments based on the same region index. For example, for a standardized target region with region index R1, region-level synchronization index reference information associated with R1 and a fingerprint fragment corresponding to R1 are embedded in this region; for a target region with region index R2, another set of reference information and another fingerprint fragment corresponding to R2 are embedded. In this way, each region forms a ternary correspondence of "region index - reference information - fingerprint fragment" during the embedding stage.

[0069] Regarding embedding strategies, different embedding strengths, embedding densities, redundancy levels, or priorities can be set according to the standardized target regions corresponding to different regional indexes. Taking the stable anchor region near the page boundary as an example, the embedding priority of the regional-level synchronous index reference information can be appropriately increased, making it easier for the recognition end to recover the index of that region first. For the multiple field-carrying regions in the middle content area, a more uniform fingerprint fragment embedding density can be adopted to ensure the overall recognition integrity. Through this regionally differentiated embedding method, the impact on the overall visual effect or layout quality of the standard data can be reduced while ensuring recognition robustness.

[0070] The interference analysis and region recovery module belongs to the identification end recovery module. Its function is to perform interference type analysis on the standard data to be identified, and restore the correspondence between the regional index and spatial location of each standardized target region in the standard data to be identified based on the regional-level synchronous index reference information, so as to complete the reconstruction of the standardized target region.

[0071] This module takes as input the standard data to be identified and pre-stored or known regional-level synchronization index reference information, and outputs the restored regionalized structural model and the reconstructed standardized target region set. It is located at the front end of the identification processing chain and is a prerequisite for the identification end to stably complete segment extraction.

[0072] In one embodiment, the interference analysis and region recovery module first performs interference type analysis on the standard data to be identified to identify at least one of the following interferences: cropping, rotation, scaling, resampling, compression, screenshotting, or printing / scanning interference. Specifically, it can determine whether cropping interference exists by analyzing page boundary integrity; whether rotation interference exists by analyzing text line direction, page main axis direction, or anchor area offset angle; whether scaling interference exists by comparing the relative size of standard areas; whether resampling interference exists by analyzing edge jaggedness, changes in sampling interval, or changes in pixel distribution; whether compression interference exists by analyzing block effects, quantization traces, or loss of detail; and whether screenshotting or printing / scanning interference exists by analyzing changes in background texture, changes in edge shadows, or local reflection traces.

[0073] After identifying the interference type, the module performs region restoration processing based on the region-level synchronization index reference information. The restoration object is not only the overall pose of the page, but also the correspondence between the region indices and spatial locations of each standardized target region. That is, the system first determines "which region should correspond to which region index," and then restores the boundary position, center position, and adjacency relationship of that region accordingly. If some target regions are missing due to cropping or occlusion, the correct location of the missing regions can be inferred based on the relationships between adjacent regions, hierarchical relationships, and scale relationships, resulting in a compensatory region reconstruction result.

[0074] The fragment extraction module belongs to the recognition end extraction module. Its function is to extract the corresponding fingerprint fragments from the corresponding standardized target area according to the recovered area index after the standardized target area reconstruction is completed.

[0075] The module takes as input the reconstructed standardized target region set and its corresponding region index, and outputs multiple fingerprint fragments to be identified. Since the preceding interference analysis and region recovery module has restored the correspondence between each target region and its region index, the fragment extraction module can extract fingerprint fragments that match the region index in the correct region, avoiding erroneous extraction caused by region misalignment.

[0076] In actual execution, the fragment extraction module can extract fingerprint fragments from each region sequentially according to the region index order, or it can prioritize extracting fragments from target regions with high stability and integrity, and then gradually supplement fragments from other regions. For regions that are not fully recovered, the system can estimate whether they meet the subsequent identification requirements based on the fragment integrity; if they do not meet the requirements, the fragments in that region are marked as low-confidence fragments or missing fragments for reference by the consistency identification module during subsequent fusion judgment.

[0077] The consistency recognition module is a recognition end judgment module. Its function is to perform index consistency verification and source matching on multiple extracted fingerprint fragments to output the source tracing recognition result.

[0078] The consistency identification module takes multiple extracted fingerprint fragments and their corresponding region indices as input and outputs the final source identification result. This module is located at the end of the identification processing chain and is responsible for realizing the final determination from "fragment extraction" to "source determination".

[0079] In a preferred embodiment, the consistency identification module performs a region index consistency check before source matching. The region index consistency check determines whether the region index corresponding to each fingerprint segment is consistent with the previously recovered region index, and whether multiple segments match each other in terms of sequence, hierarchy, and combination. If a segment is extracted but its corresponding region index is inconsistent with the recovered structure model, it can be determined that the segment is at risk of mis-extraction, false matching, or partial tampering.

[0080] After completing the index consistency verification, the consistency identification module further performs fragment integrity assessment and cross-region matching result fusion. For example, fragments with high integrity, consistent regional indexes, and high matching strength can be assigned higher identification weights; fragments with low integrity or insufficient boundary recovery are assigned lower weights. Finally, the source identification result is output by combining the fusion results of multiple fragments. Thus, this invention not only avoids the decisive impact of a single region failure on the overall identification, but also reduces the risk of misidentification caused by local region misalignment, local tampering, and fragment distortion.

[0081] Based on the above system, such as Figure 2 As shown, this invention also provides an interference-resistant standard data traceability fingerprint embedding and identification method. This method can be executed by the aforementioned system, and its steps correspond one-to-one with the functions of each module. The implementation method will be described in detail below in conjunction with the above content.

[0082] Step S1: Establish a regionalized structural model

[0083] First, the standard data to be processed is obtained, and a regionalized structure model is established based on the layout boundaries, partition boundaries, field areas, table cell areas or predefined anchor areas of the standard data to obtain multiple standardized target areas and regional indexes for each standardized target area.

[0084] In this step, page boundaries, partition boundaries, field boundaries, table cell boundaries, image / text column boundaries, or predefined anchor area boundaries in the standard data are preferably identified, and standardized target regions and corresponding region indexes are generated based on the positional relationships between these boundaries. This approach allows for the templated and hierarchical representation of the inherent structure of the standard data, providing a regional foundation for subsequent synchronous index reference information construction and fingerprint fragment embedding.

[0085] Step S2: Generate region-level synchronization index reference information

[0086] Based on the spatial distribution relationship of the multiple standardized target regions, a regional-level synchronous index reference information is generated to characterize the correspondence between the regional index and the spatial location of each standardized target region.

[0087] In this step, preferably, corresponding regional boundary reference information, regional center reference information, regional adjacency reference information, regional scale reference information, or regional hierarchical reference information are generated based on the center location, boundary location, adjacency relationship, scale relationship, or hierarchical relationship of multiple standardized target regions. Therefore, when performing subsequent region restoration, the recognition end can first use this reference information to recover "which spatial region the region index corresponds to" before extracting segments, instead of relying solely on overall geometric correction.

[0088] Step S3: Construct fingerprint fragment

[0089] The source fingerprint information is generated based on the source identifier, and the source fingerprint information is split into multiple fingerprint fragments corresponding to different region indices.

[0090] In this step, the source identifier can originate from the user, terminal, version, distribution batch, or other responsible entity. By performing region-level segmentation on the original source fingerprint information, each region index corresponds to at least one fingerprint fragment. Preferably, fingerprint fragments of different lengths and redundancy levels can be allocated according to the stability of different target regions to improve recognition capabilities under complex interference conditions.

[0091] Step S4: Perform binding embedding

[0092] According to the regional index, the regional-level synchronization index reference information is bound to the corresponding fingerprint fragment and then embedded into the corresponding standardized target region to obtain the embedded standard data.

[0093] This step is crucial to the entire embedding process. Since the region-level synchronization index reference information and the fingerprint fragment are bound together based on the same region index, once the recognition device recovers a certain region index from the interfered data, it can extract the corresponding fingerprint fragment from the corresponding target region. Preferably, differentiated embedding parameters can be set for target regions corresponding to different region indices to balance regional stability and overall layout quality.

[0094] Step S5: Perform interference type analysis

[0095] Obtain the standard data to be identified, and perform interference type analysis on the standard data to be identified.

[0096] In this step, it is preferable to identify at least one of the following interferences affecting the standard data to be identified: cropping, rotation, scaling, resampling, compression, screenshotting, or printing / scanning interference. Based on the identification results, the corresponding region recovery processing path is determined. Different interferences correspond to different region recovery priorities. For example, if the interference is cropping, the focus is on determining which standardized target regions are still completely preserved and which regions are partially missing; if the interference is rotation and scaling, the focus is on restoring the relative position and scale of each target region; if the interference is printing / scanning, the focus is on suppressing the impact of background texture changes and page edge distortion on region recognition.

[0097] Step S6: Restore the standardized target region

[0098] Based on the regional-level synchronous index reference information, the correspondence between the regional index and spatial location of each standardized target region in the standard data to be identified is restored to complete the reconstruction of the standardized target region.

[0099] This step does not only restore the overall orientation of the page, but also restores the matching relationship between each standardized target region and its corresponding region index. Even if some regions are missing due to cropping, occlusion, or partial replacement, the system can infer the location range of the missing regions based on the adjacency, scale, and hierarchical relationships of the remaining regions, thereby establishing a complete or near-complete set of reconstructed regions.

[0100] Step S7: Extract the corresponding fingerprint fragment

[0101] Based on the recovered region index, extract the corresponding fingerprint fragments from the corresponding standardized target region.

[0102] In this step, the system can first extract fingerprint fragments from the target area with high integrity, and then extract fragments from other areas. For areas with obvious damage or low recovery confidence, their corresponding fragments can be assigned a low-confidence label for subsequent fusion and determination.

[0103] Step S8: Perform index consistency check and source matching.

[0104] Perform index consistency verification and source matching on multiple extracted fingerprint fragments, and output the source identification results.

[0105] Specifically, the system can determine whether the region index corresponding to each fingerprint fragment is consistent with the recovered region index, and then perform a fusion judgment based on the consistency result, fragment completeness, and fragment matching result to output the final source identification result. If some fragments have high consistency and clear matching results, a stable source identification conclusion can still be formed even if there are a few low-confidence fragments or missing fragments; conversely, if the region index relationship between fragments conflicts, the system can determine the identification result as a low-confidence or abnormal result to prevent misjudgment.

[0106] For ease of understanding, the following uses a standard report page with fixed field areas and table areas as an example to illustrate the execution process of this invention.

[0107] First, the system receives the standard report page and, through the standard data modeling module, divides the page into four standardized target areas: header / title area, basic information field area, table data area, and footer / signature area, while generating a region index for each area. Next, the region-level synchronous index reference generation module generates region-level synchronous index reference information based on the boundary position, center position, and vertical adjacency of each area. Then, the fingerprint fragment construction module generates traceability fingerprint information based on the distribution object identifier and splits it into multiple fingerprint fragments, corresponding to the header / title area, field area, table area, and footer area respectively. Finally, the binding and embedding module embeds the reference information and fingerprint fragments corresponding to each area into the respective areas, forming the embedded standard report page.

[0108] When the report page is compressed during transmission and then screenshotted or scanned after printing, the recognition terminal obtains the page to be recognized. The system first uses the interference analysis and region recovery module to determine that it has been simultaneously affected by compression, screenshotting, and slight rotation interference. Then, based on the region-level synchronous index reference information, it restores the relationship between the header / title area, field area, table area, and footer area and the corresponding region index. Next, the fragment extraction module extracts the corresponding fingerprint fragments from each of the above areas. The consistency recognition module first performs a region index consistency check on the extracted fragments, and then combines the fragment completeness and source matching results to output the final distribution object recognition result. Thus, even if the table area is partially cropped or the footer area is obscured, as long as the remaining standardized target areas retain sufficient fragments, this invention can still complete effective source recognition.

[0109] It is worth further explanation that:

[0110] The core of this invention does not lie in limiting itself to a specific fingerprint embedding algorithm, image processing algorithm, or interference detection algorithm. Rather, it lies in: leveraging the stable regional structure of standard data, first establishing a regionalized structural model, then constructing regional-level synchronous index reference information, and binding this regional-level synchronous index reference information with fingerprint fragments based on the same regional index. This allows the recognition end to prioritize the recovery of the regional index of the standardized target region under complex interference conditions, before extracting the corresponding fragment and identifying its source. This technical approach differs from traditional methods that directly extract the overall watermark after overall page geometric correction, and is better suited to scenarios involving local damage, local misalignment, and structural interference in standard data.

[0111] Furthermore, the standardized target area in this invention is not limited to the aforementioned page boundary areas, header / footer areas, field-carrying areas, table cell areas, image / text column areas, or predefined anchor areas. It can also be extended to other page areas with fixed boundaries and stable spatial relationships according to specific business templates. Correspondingly, the regional-level synchronous index reference information can also be adjusted according to the regional structural characteristics. Any technical solution that constructs reference information based on the correspondence between standard data region indexes and spatial locations, and binds it with regional-level fingerprint fragments for anti-interference embedding and identification, can be considered to fall within the scope of this invention.

[0112] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, or combinations made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A standard data traceability fingerprint embedding and recognition system with anti-interference capabilities, characterized in that, include: The standard data modeling module is used to acquire the standard data to be processed, and to establish a regionalized structure model of the standard data based on the layout boundaries, partition boundaries, field areas, table cell areas or predefined anchor areas, so as to obtain multiple standardized target areas and regional indexes of each standardized target area. The regional-level synchronization index reference generation module is used to generate regional-level synchronization index reference information based on the spatial distribution relationship of the multiple standardized target regions. The regional-level synchronization index reference information is used to characterize the correspondence between the regional index and spatial location of each standardized target region. A fingerprint fragment construction module is used to generate traceability fingerprint information based on the source identifier and to split the traceability fingerprint information into multiple fingerprint fragments corresponding to different region indices. The binding and embedding module is used to bind the regional-level synchronization index reference information with the corresponding fingerprint fragments according to the regional index and embed them into the corresponding standardized target regions to obtain the embedded standard data. The interference analysis and region recovery module is used to analyze the interference type of the standard data to be identified, and to recover the correspondence between the regional index and spatial location of each standardized target region in the standard data to be identified based on the regional-level synchronous index reference information, so as to complete the reconstruction of the standardized target region. The fragment extraction module is used to extract the corresponding fingerprint fragments from the corresponding standardized target region according to the recovered region index after the standardized target region reconstruction is completed. The consistency identification module is used to perform index consistency verification and source matching on multiple extracted fingerprint fragments to output the source identification results; The regional-level synchronization index reference information and the fingerprint fragment are bound together based on the same regional index, so that after the standard data to be identified is subjected to cropping, rotation, scaling, resampling, compression, screenshotting or printing scanning interference, the regional index of the standardized target area can be recovered first, and then the corresponding fingerprint fragment can be extracted for source identification.

2. The system according to claim 1, characterized in that, The regionalized structure model established by the standard data modeling module includes at least one of the following regions: page boundary region, header and footer region, field carrying region, table cell region, text and image column region, or predefined anchor area.

3. The system according to claim 1, characterized in that, The regional-level synchronization index reference information generated by the regional-level synchronization index reference generation module includes at least one of the following: regional boundary reference information, regional center reference information, regional adjacency relationship reference information, regional scale reference information, or regional hierarchical reference information.

4. The system according to claim 1, characterized in that, The binding embedding module is used to set different embedding strengths, embedding densities, redundancy levels or priorities for the standardized target regions corresponding to different region indices, so that when some standardized target regions of the standard data to be identified are damaged, the fingerprint fragments in the remaining standardized target regions can still be used for source identification.

5. The system according to claim 1, characterized in that, The consistency identification module is used to perform regional index consistency verification, fragment integrity assessment and cross-regional matching result fusion on multiple extracted fingerprint fragments before source matching, and output the final source identification result based on the fusion result.

6. A standard data traceability fingerprint embedding and recognition method with anti-interference capabilities, characterized in that, include: S1: Obtain the standard data to be processed, and establish a regional structure model based on the layout boundary, partition boundary, field area, table cell area or predefined anchor area of ​​the standard data to obtain multiple standardized target areas and the regional index of each standardized target area; S2: Based on the spatial distribution relationship of the multiple standardized target regions, generate regional-level synchronous index reference information to characterize the correspondence between the regional index and spatial location of each standardized target region; S3: Generate traceability fingerprint information based on the source identifier, and split the traceability fingerprint information into multiple fingerprint fragments corresponding to different region indices; S4: According to the regional index, the regional-level synchronization index reference information is bound to the corresponding fingerprint fragment and then embedded into the corresponding standardized target region to obtain the embedded standard data; S5: Obtain the standard data to be identified, and perform interference type analysis on the standard data to be identified; S6: Based on the regional-level synchronous index reference information, restore the correspondence between the regional index and spatial location of each standardized target region in the standard data to be identified, so as to complete the reconstruction of the standardized target region; S7: Extract the corresponding fingerprint fragment from the corresponding standardized target region according to the recovered region index; S8: Perform index consistency verification and source matching on the extracted multiple fingerprint fragments, and output the source identification results; Specifically, by establishing a binding relationship between regional-level synchronization index reference information and fingerprint fragments based on the same regional index, the standard data to be identified can first recover the regional index of the standardized target area after being interfered with, and then extract the corresponding fingerprint fragments for source identification.

7. The method according to claim 6, characterized in that, The process of establishing a regionalized structural model based on the standard data in S1 includes: Identify page boundaries, partition boundaries, field boundaries, table cell boundaries, text / image column boundaries, or predefined anchor area boundaries in the standard data, and generate standardized target areas and corresponding area indexes based on the positional relationships between each boundary.

8. The method according to claim 6, characterized in that, The generation of regional-level synchronization index reference information based on the spatial distribution relationship of the multiple standardized target regions in S2 includes: Based on the center location, boundary location, adjacency relationship, scale relationship or hierarchical relationship of multiple standardized target areas, corresponding regional boundary reference information, regional center reference information, regional adjacency relationship reference information, regional scale reference information or regional hierarchical reference information are generated.

9. The method according to claim 6, characterized in that, The interference type analysis of the standard data to be identified in S5 includes: Identify at least one of the following interferences affecting the standard data to be identified: cropping, rotation, scaling, resampling, compression, screenshotting, or printing / scanning; and determine the corresponding region recovery processing path based on the identification results.

10. The method according to claim 6, characterized in that, In step S8, the index consistency check and source matching of the extracted multiple fingerprint fragments are performed, and the source tracing and identification results are output, including: Each fingerprint fragment is checked to see if its corresponding region index is consistent with the recovered region index. The results of consistency, fragment integrity and fragment matching are then used to make a fusion judgment to output the final source identification result.

Images