A delivery document generation system for a delivery service
By using partitioned pixel scattering and cross-domain visual fragment fusion technology, the core business data and anti-counterfeiting features of the delivery document are deeply integrated at the pixel level, which solves the problem of weak anti-counterfeiting capabilities in the existing delivery document generation technology and achieves non-replicability and efficient traceability verification.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- 张家港保税数据科技有限公司
- Filing Date
- 2026-04-09
- Publication Date
- 2026-07-07
AI Technical Summary
Existing delivery document generation technologies suffer from weak anti-counterfeiting capabilities, high security risks, and the ease with which anti-counterfeiting features can be separated from the document itself, making it difficult to effectively verify authenticity and integrity in high-security scenarios.
By using techniques such as partitioned pixel scattering, cross-domain visual fragment fusion, and fragmented character embedding, core business data and non-core anti-counterfeiting features are deeply integrated at the pixel level to generate an uncopyable visual form. Anti-counterfeiting information is also deeply bound to business data, supporting efficient traceability and verification.
It achieves uncopyable visual form of documents and deep binding of anti-counterfeiting information with business data, enhancing anti-counterfeiting strength and uncopyability, and ensuring the overall visual consistency of documents and the traceability and verification capabilities of the backend.
Smart Images
Figure CN121997890B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of delivery document generation technology, specifically a delivery document generation system for delivery services. Background Technology
[0002] In service sectors such as finance and logistics that involve physical delivery, delivery documents serve as crucial evidence for recording the rights and obligations of both parties, clarifying the information of the subject matter, and completing asset transfer. The efficiency and security of their generation, transmission, and verification directly affect the reliability and compliance of business processes. Currently, common delivery document generation methods mainly rely on two technical paths: one is an electronic document auto-fill system based on fixed templates and structured data, which quickly generates uniformly formatted electronic documents by mapping business data to corresponding fields in preset templates; the other is to scan or photograph paper documents to form electronic images, and then extract key information through optical character recognition (OCR) technology and store it in the system to achieve digital archiving of paper documents. These existing technologies have improved the automation level of document processing to a certain extent, reduced manual input errors, and realized the online transformation of some business processes.
[0003] However, existing technologies still have significant limitations in practical applications. First, while template-generated electronic documents have a regular format, their fixed layout and regular element arrangement lack effective visual anti-counterfeiting features, making them easily copied or tampered with and difficult to resist targeted counterfeiting attacks. In delivery scenarios requiring high security, there are risks to verifying authenticity and integrity. Second, while digitization through image acquisition and OCR recognition preserves the visual form of the original document, its focus is on information extraction rather than anti-counterfeiting enhancement. The generated electronic files are merely simple copies or compressions of the original images, lacking anti-counterfeiting mechanisms deeply integrated with business data. They lack uniqueness and non-repudiation, making it impossible to effectively trace the document's origin and circulation process. Finally, existing solutions focus primarily on the machine readability and format standardization of document content, failing to deeply and unstructure the integration of business data with the document's visual carrier. This results in anti-counterfeiting features often existing as visible watermarks, digital signatures, or external encrypted attachments, separated from the main visual content of the document. This not only increases processing complexity but also makes them easy to separate, remove, or destroy. Summary of the Invention
[0004] (a) Technical problems to be solved
[0005] To address the shortcomings of existing technologies, this invention provides a delivery document generation system for delivery services. By employing techniques such as partitioned pixel scattering, cross-domain visual fragment fusion, and fragmented character embedding, core business data and non-core anti-counterfeiting features are deeply integrated at the pixel level. This solves the problems of weak anti-counterfeiting capabilities, high security risks, and easy separation of anti-counterfeiting features from the document body in existing delivery document generation technologies. It achieves the beneficial effects of making the document's visual form uncopyable, deeply binding anti-counterfeiting information with business data, and supporting efficient traceability and verification.
[0006] (II) Technical Solution
[0007] To achieve the above objectives, the present invention provides the following technical solution: a delivery document generation system for delivery services, comprising:
[0008] The data acquisition module acquires raw images of physical paper delivery documents, generates a template base canvas based on the standard delivery document template, performs partitioned pixel scattering on the raw images and template base canvas, generates physical pixel units and template pixel units, and constructs a hybrid visual pixel pool.
[0009] The visual substrate processing module extracts physical pixel units from the hybrid visual pixel pool and performs grayscale inverse conversion. It then performs secondary fragmentation of the template pixel units to generate template micro-pixel fragments. Based on the grayscale difference, it locates the delivery node and randomly delivers the fragments to form a cross-domain visual fragment fusion substrate.
[0010] The anti-counterfeiting character embedding module retrieves core delivery business data and generates anti-counterfeiting filler placeholder characters. It then splits the anti-counterfeiting filler placeholder characters into fragmented character units. Based on connected component analysis, it extracts the irregular embedding carrier of the cross-domain visual fragment fusion substrate and embeds the fragmented character units into the corresponding irregular embedding carrier to obtain the visual fusion image.
[0011] The output module performs secondary grayscale inverse restoration on the non-core anti-counterfeiting areas of the visual fusion image, conducts dual-layer partition verification, and generates an electronic delivery document with anti-counterfeiting traceability mark after the verification is passed.
[0012] Furthermore, the template base canvas generation includes retrieving a standard blank electronic template of a delivery document, dividing the core business area into a non-core anti-counterfeiting area, stripping field positioning lines, text format marks, layout coordinate parameters, business field annotations, and structural association instruction information from the non-core anti-counterfeiting area, and retaining only the basic pixels of the printed font, table lines, and document outer border, while the core business area retains the original field positioning and semantic structure.
[0013] Furthermore, for the original physical image, background pixels exceeding the main body of the document are cropped, and the image area corresponding to the non-core anti-counterfeiting area is decomposed into independent physical pixel units, while the image area corresponding to the core business area retains the continuous pixel arrangement structure; the non-core anti-counterfeiting area of the template base canvas is decomposed into independent template pixel units, while the core business area maintains the original pixel structure unchanged; the original spatial position and arrangement order of pixels in the non-core anti-counterfeiting area are scrambled, while the original pixel arrangement and spatial positioning in the core business area are maintained, thus constructing a hybrid visual pixel pool.
[0014] Furthermore, physical pixel units are extracted to form a pure physical document base pixel group. Grayscale pixel inverse conversion is performed on this pixel group, and the original grayscale value is subtracted from 255 for inverse mapping to generate an inverse visual base image. Secondary fragmentation processing is performed on the template pixel units, and neighborhood aggregation is performed with a threshold of 5 to 10 pixels to form template micro-pixel fragments.
[0015] Furthermore, the generated inverse visual base image is traversed row by row and column by column. The grayscale difference between each group of horizontal and vertical adjacent pixels is calculated. Pixels with grayscale differences greater than a preset fixed threshold are selected and marked as fragment placement and positioning nodes. With each fragment placement and positioning node as the center, a circular placement range with a radius of 10 to 20 pixels is defined. Single template micro-pixel fragments are randomly placed into the corresponding circular range. During the placement process, the fragments are randomly assigned a rotation angle of 0° to 360° to complete the initial adsorption and positioning.
[0016] Furthermore, the minimum spacing between fragment edges is calculated, fragments with a spacing of less than 3 pixels are separated by equal-distance outward expansion, and fragments with a spacing of more than 20 pixels are moved closer together by equal-distance inward convergence. The spacing between adjacent fragment edges is uniformly controlled within the range of 3 to 20 pixels to generate a cross-domain visual fragment fusion substrate.
[0017] Furthermore, the core settlement business data includes the main information of both parties to the settlement, the details of the underlying asset, the transaction amount, the settlement time, the filing number, and the serial number; the anti-counterfeiting filler placeholder characters are random placeholder character streams without actual business meaning, the internal format marks and separators of the characters are deleted, the structured attributes are stripped, and only plain text filler characters are retained.
[0018] Furthermore, a splitting length range of 1 to 5 characters is preset. Within the splitting length range, a single value is randomly selected as the threshold for a single splitting. The anti-counterfeiting filling placeholder characters are randomly split segment by segment, decomposed into several independent fragmented character units. The cross-domain visual fragment fusion substrate is traversed, and a connected component analysis algorithm is used to extract the gap region and independent irregular blank pixel region in the cross-domain visual fragment fusion substrate. Blank regions with an area of less than 5 pixels are removed, the boundary of the remaining blank region is locked, and the connected component is determined to form an independent and closed irregular embedded carrier.
[0019] Furthermore, the fragmented character unit embedding includes analyzing the grayscale mean, pixel area, and boundary perimeter of the irregularly shaped embedding carrier, selecting carriers that are compatible with the fragmented character units, randomly binding the fragmented character units to the compatible carriers, and adaptively adjusting the character size, rotation angle, and character spacing according to the carrier contour direction, boundary curvature, and internal pixel trajectory, so that the characters are non-linearly arranged along the internal trajectory of the carrier and completely fall into the interior of the irregularly shaped embedding carrier; and performing edge grayscale transition fusion on the pixels of the fragmented character units and the base pixels of the cross-domain visual fragment fusion, unifying the grayscale difference at the edge connection to within ±10.
[0020] Furthermore, the secondary grayscale inverse restoration involves performing a secondary grayscale inverse transformation on the non-core anti-counterfeiting areas of the visual fusion image by subtracting the current pixel grayscale value from 255; the dual-layer partition verification includes a first layer verifying the data integrity, font clarity, and layout regularity of the core business area, and a second layer verifying the uniformity of pixel fragment distribution and the validity of anti-counterfeiting filling characters in the non-core anti-counterfeiting area; and embedding an invisible watermark-style partition processing identifier and a unique encrypted character sequence anti-counterfeiting traceability mark.
[0021] (III) Beneficial Effects
[0022] This invention provides a delivery document generation system for delivery services, which has the following advantages:
[0023] (1) By finely partitioning and breaking down pixels and constructing a hybrid visual pixel pool, the deep integration of physical images and standard templates at the pixel level is achieved. This not only preserves the visual details of the original document, but also effectively distinguishes the core business area from the non-core anti-counterfeiting area. This provides a clear and controllable pixel foundation for subsequent visual base processing and anti-counterfeiting embedding, thereby enhancing the overall anti-counterfeiting non-replicability while ensuring the readability of the core information of the document.
[0024] (2) By performing grayscale inverse transformation on physical pixel units and secondary fragmentation on template pixel units, irregular and semantically meaningless template micro-pixel fragments are generated. The placement node is located based on the pixel grayscale difference, and the fragments are randomly and rotatably placed into the inverse substrate. The spacing between fragments is controlled to form a nonlinear and disordered cross-domain visual fragment fusion substrate, which completely strips away the original structure and regularity of the template. A highly random and visually mixed anti-counterfeiting bottom layer is constructed at the pixel level, providing a highly unpredictable and difficult-to-replicate visual carrier for subsequent anti-counterfeiting character embedding.
[0025] (3) By embedding fragmented random placeholder characters irregularly into an irregularly shaped carrier composed of fragmented substrates, the anti-counterfeiting information and the randomized visual substrate are deeply integrated at the pixel level, ensuring that the distribution and form of the anti-counterfeiting mark have a high degree of nonlinearity, randomness and uniqueness, thereby greatly enhancing the anti-counterfeiting strength and non-replicability of the document, while not interfering with the readability of core business data.
[0026] (4) By performing secondary grayscale inverse restoration on the non-core anti-counterfeiting areas of the visual fusion image, the anti-counterfeiting features are visible under normal display. At the same time, double-layer partition verification is carried out to ensure that the core business data is clear and readable and the anti-counterfeiting layout is complete and effective. Finally, an electronic delivery document integrating an invisible watermark and a unique encrypted sequence is generated, realizing the deep binding of anti-counterfeiting information and the document subject, and ensuring the overall visual coordination of the document and the traceability and verification capability of the background. Attached Figure Description
[0027] Figure 1 This is a schematic diagram of the delivery document generation system for delivery services according to the present invention. Detailed Implementation
[0028] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0029] Please see Figure 1 This invention provides a delivery document generation system for delivery services, comprising: a data acquisition module, a visual substrate processing module, an anti-counterfeiting character embedding module, and an output module, wherein:
[0030] The data acquisition module acquires raw images of physical paper delivery documents, generates a template base canvas based on the standard delivery document template, performs partitioned pixel scattering on the raw images and template base canvas, generates physical pixel units and template pixel units, and constructs a hybrid visual pixel pool.
[0031] Specifically, static images of physical paper delivery documents are acquired using industrial high-definition vision acquisition equipment with a fixed focal length. During the acquisition process, the automatic angle correction, automatic exposure adjustment, automatic noise removal, and automatic sharpening optimization functions of the equipment are actively turned off to obtain the original physical document images without any post-processing optimization. The background redundant pixels, ambient light and shadow deviation pixels, physical wrinkle and deformation pixels, and image edge distortion pixels in the original image are completely preserved.
[0032] The system retrieves pre-stored blank electronic templates of standard delivery documents from the business system. These templates are divided into core business areas and non-core anti-counterfeiting areas. The core business areas include mandatory read / write fields such as document amount, transaction party information, target number, and signature area. The non-core anti-counterfeiting areas include background borders, blank gaps, decorative lines, and redundant edge areas. The electronic template is then decomposed into a single-layer bitmap format pure pixel canvas. For the non-core anti-counterfeiting areas, field positioning lines, text format markers, layout coordinate parameters, business field annotations, and structural association instructions are stripped away line by line, retaining only the basic pixels of the printed fonts, table lines, and document outer borders. For the core business areas, the original field positioning and semantic structure are preserved, ultimately generating a layered, controllable, and partition-differentiated attribute-free template base canvas.
[0033] The system performs unified resolution and pixel size normalization on multiple frames of original physical document images, adjusting all images to a uniform 1920×1080 pixel specification, and cropping out redundant background pixels that exceed the main body of the document. For the normalized physical document images, a partitioned pixel scattering operation is performed. For non-core business image areas, the continuous pixel arrangement of text, lines, and folds is broken, and the images are decomposed into independent physical pixel units that are not related to each other. For core business image areas, the continuous pixel arrangement structure is retained, and no scattering processing is performed. Simultaneously, partitioned pixel scattering is performed on the attribute-free template base canvas. In non-core areas, the original continuity of text, lines, and border pixel arrangement is broken, and the images are decomposed into independent discrete pixel units. In core areas, the original pixel structure remains unchanged, ensuring normal data embedding and retrieval in subsequent processes.
[0034] The normalized physical image and the broken and split template discrete pixel units are uniformly calibrated for grayscale range and color channel parameters of the two types of materials to eliminate pixel parameter deviations between heterogeneous materials; all materials are traversed to remove blurry, distorted, and duplicate invalid and redundant pixels, retain clear and effective pixel units, and no initial spatial relationship is established between various types of pixel units.
[0035] For non-core anti-counterfeiting areas, the original spatial position and arrangement order of pixels are completely disrupted, and the preset layout traces of the standard template are completely removed; for core business areas, the original pixel arrangement and spatial positioning are maintained; finally, a hybrid visual pixel pool that combines partitioned control, anti-counterfeiting and readability is constructed.
[0036] By finely partitioning and breaking down pixels and constructing a hybrid visual pixel pool, deep fusion of physical images and standard templates at the pixel level is achieved. This not only preserves the visual details of the original document but also effectively distinguishes between core business areas and non-core anti-counterfeiting areas. This provides a clear and controllable pixel foundation for subsequent visual base processing and anti-counterfeiting embedding, thereby enhancing the overall anti-counterfeiting non-replicability while ensuring the readability of the core information of the document.
[0037] The visual substrate processing module extracts physical pixel units from the hybrid visual pixel pool and performs grayscale inverse conversion. It then performs secondary fragmentation of the template pixel units to generate template micro-pixel fragments. Based on the grayscale difference, it locates the delivery node and randomly delivers the fragments to form a cross-domain visual fragment fusion substrate.
[0038] Specifically, from the hybrid visual pixel pool, physical pixel units are selected and extracted to form a pure physical document base pixel group; grayscale pixel inverse transformation is performed on the pure physical document base pixel group in the original image, and the grayscale value is reversed according to the fixed rule of subtracting the original grayscale value from 255, without changing the total number of pixels, the relative position between pixels and the overall distribution density, to generate an inverse visual base image.
[0039] From the same mixed visual pixel pool, the scattered and split template discrete pixel units are retrieved, and secondary fragmentation processing is performed on the discrete pixel units. According to the randomly generated and non-fixed pixel number threshold, the pixel number threshold range is preset to 5~10 pixels. Using this number threshold as the unit, the template discrete single pixel is aggregated into irregular neighborhoods. The aggregation forms template micro-pixel fragments with irregular size, irregular outline shape, no complete semantics, and no structural continuity. The original template structure, text semantics, table associations and other conventional information are completely stripped away and transformed into pure visual pixel carriers.
[0040] The generated inverse visual base image is traversed row by row and column by column. The grayscale difference between each group of horizontal and vertical adjacent pixels is calculated. Pixels with grayscale differences greater than a preset fixed threshold are selected and marked as fragment placement and positioning nodes. With each fragment placement and positioning node as the center, a circular placement range with a radius of 10 to 20 pixels is defined. Single pixel fragments are randomly placed into the corresponding circular range. During the placement process, the fragments are randomly assigned a rotation angle of 0° to 360°. The fragments are not scaled or aligned proportionally. The only constraint is that the fragments do not exceed the corresponding placement range. This allows various pixel fragments to autonomously conform to the pixel distribution of the inverse visual base image and complete the initial adsorption and positioning.
[0041] Spatial overlap detection is performed on all pixel fragments after deployment. Redundant fragments that completely overlap are removed, while locally overlapping pixel fragments are retained. For non-overlapping adjacent pixel fragments, the minimum distance between fragment edges is calculated. Fragments with a distance of less than 3 pixels are separated by equal-distance outward expansion, while fragments with a distance of more than 20 pixels are clustered together by equal-distance inward expansion. The distance between adjacent fragment edges is uniformly controlled within the range of 3 to 20 pixels to avoid local over-density or over-sparseness and to maintain the overall nonlinear disordered distribution state. Finally, the spatial coordinates, rotation angle, and scaling ratio of all effective fragments are locked, and subsequent autonomous displacement and deformation of fragments are stopped to generate a fixed disordered cross-domain visual fragment fusion substrate.
[0042] By performing grayscale inverse transformation on physical pixel units and secondary fragmenting of template pixel units, irregular and semantically meaningless template micro-pixel fragments are generated. The placement nodes are located based on pixel grayscale differences, and the fragments are randomly and rotatably placed into the inverse substrate. The spacing between fragments is controlled to form a non-linear and disordered cross-domain visual fragment fusion substrate. The original structure and regularity of the template are completely stripped away, and a highly random and visually mixed anti-counterfeiting bottom layer is constructed at the pixel level, providing a highly unpredictable and difficult-to-replicate visual carrier for subsequent anti-counterfeiting character embedding.
[0043] The anti-counterfeiting character embedding module retrieves core delivery business data and generates anti-counterfeiting filler placeholder characters. It then splits the anti-counterfeiting filler placeholder characters into fragmented character units. Based on connected component analysis, it extracts the irregular embedding carrier of the cross-domain visual fragment fusion substrate and embeds the fragmented character units into the corresponding irregular embedding carrier to obtain the visual fusion image.
[0044] Specifically, the system retrieves the complete set of original structured business data corresponding to this transaction from the business database, including the main information of both parties involved in the transaction, details of the underlying assets, transaction amount, transaction time, filing number, and serial number. This data is essential core business data for the transaction process, and the original field structure and semantic integrity are preserved throughout the process without any fragmentation or format conversion. Simultaneously, a random placeholder character stream with no actual business meaning is generated and is used only for anti-counterfeiting purposes. The character content does not contain any valid transaction information and is only used as visual fill material. The internal format marks and separators of the characters are deleted, all structured attributes are stripped away, and only plain text fill characters are retained.
[0045] The preset character splitting length range is 1 to 5 characters. Within this length range, a single value is randomly selected as the threshold for a single splitting. Only the anti-counterfeiting filler placeholder character stream without business meaning is subjected to segmented random splitting. The splitting process does not follow semantic or statement constraints and breaks down the filler characters into several independent fragmented character units. All core delivery business data maintains complete field units and does not perform any splitting or disassembling operations.
[0046] By traversing the cross-domain visual fragment fusion substrate, a connected component analysis algorithm is used to extract the gap regions formed by the intersection of each pixel fragment and the independent irregular blank pixel regions inside the substrate. The extremely small blank regions with an area of less than 5 pixels are removed and merged into adjacent fragments. The remaining blank regions are bounded and connected component determined to form several independent, closed, irregularly shaped embedding carriers.
[0047] The three visual parameters of grayscale mean, pixel area, and boundary perimeter of each irregularly shaped embedded carrier are analyzed one by one to establish a pure visual adaptation screening rule: the carrier area range that matches the total pixel occupancy of the fragmented character units is selected, and carriers with a grayscale mean difference greater than ±15 are also selected; within the selected adaptation carrier range, each fragmented character unit is randomly bound to the corresponding adaptation irregularly shaped embedded carrier one by one. Core business data does not participate in random filling and binding, and the fixed field embedding path is used throughout the process without changing the original data layout.
[0048] For the small character units that have been bound and matched, the format adaptive reverse adjustment is performed. Based on the outline, boundary curvature, and internal pixel trajectory of the corresponding irregularly shaped embedding carrier, the character size, rotation angle, and character spacing are adaptively adjusted so that the characters are arranged non-linearly along the internal trajectory of the carrier throughout the process. The arrangement process ensures that the characters fall completely inside the carrier and do not exceed the carrier boundary, thus completing the non-standard reverse embedding of characters. For the complete set of core delivery business data, the standard document field embedding rules are followed to perform centering and alignment to ensure that the data semantics are clear and the format is regular, meeting the needs of normal reading, manual verification, and machine recognition.
[0049] The embedded fragmented character unit pixels and the cross-domain visual fragment fusion base pixels are subjected to edge grayscale transition fusion processing to unify the grayscale difference at the edge connection of the two to within ±10, eliminating pixel breaks, hard boundaries and obvious color differences; redundant character pixels that exceed the carrier boundary are cropped and removed, and blank pixel gaps inside the carrier that are not covered are automatically filled, so that the character pixels and visual base pixels fit seamlessly and without boundary breaks, resulting in a visual fusion image with complete character embedding and edge fusion.
[0050] By embedding fragmented, random placeholder characters irregularly into an irregularly shaped carrier composed of fragmented substrates, the anti-counterfeiting information and the randomized visual substrate are deeply integrated at the pixel level. This ensures that the distribution and form of the anti-counterfeiting mark have a high degree of non-linearity, randomness, and uniqueness, thereby greatly enhancing the anti-counterfeiting strength and non-replicability of the document, while not interfering with the readability of core business data.
[0051] The output module performs secondary grayscale inverse restoration on the non-core anti-counterfeiting areas of the visual fusion image, conducts dual-layer partition verification, and generates an electronic delivery document with anti-counterfeiting traceability mark after the verification is passed.
[0052] Specifically, the process retrieves the visually fused image after character embedding and edge blending are completed, and initiates the partitioned pixel reverse restoration process. Only non-core anti-counterfeiting areas undergo a secondary pixel grayscale inverse conversion. The grayscale value of the non-core area is reduced by 255, and the reversed visual pixels of the non-core area are restored to the normal visible light display state. The original layout of the disordered arrangement of pixel fragments and the embedding of anti-counterfeiting characters in this area is preserved throughout the process without any positional adjustments. Since the core business area does not undergo pixel inverse operation throughout the process, the original normal visual parameters remain unchanged, and it does not participate in this reverse restoration, generating a partitioned visually unified image of the document to be verified.
[0053] The restored image of the document to be verified is optimized for full-area pixel smoothing. This is done to specifically weaken the harsh boundaries and visual abruptness caused by pixel fragment splicing and anti-counterfeiting character embedding in non-core areas. The grayscale and color transition values of local pixels are adjusted to eliminate image discontinuity, noise and visual fragmentation without changing the disordered anti-counterfeiting layout or touching the core field pixels, thereby improving the overall visual coordination of the document.
[0054] The optimized images of documents to be verified undergo a two-layer partitioned verification process. The first layer targets the core business area, verifying data integrity, font clarity, and layout regularity field by field. Specifically, it verifies that the core fields have no missing pixels, the font is clearly visible, and there is no offset or misalignment, ensuring that both machine recognition and manual verification can read them normally. The second layer targets non-core anti-counterfeiting areas, verifying the uniformity of pixel fragment spatial distribution: determining that the difference in fragment density within a single frame does not exceed a preset range of 3 to 20 pixels, with no local dense stacking or large areas of blank gaps; simultaneously verifying the effectiveness of anti-counterfeiting fill characters: determining that the fill characters are not missing, do not overlap or blur, and do not abnormally cross the boundaries into the core business area, removing locally distorted redundant pixels, and checking for any core area boundary crossings or anti-counterfeiting layout failures.
[0055] After verification, the compliant images are rendered and parameters are encapsulated according to the preset PDF and PNG high-definition lossless dual-format standards of the business system. The parameter encapsulation involves solidifying and packaging four basic visual parameters: document image resolution, grayscale range, color channels, and pixel size. Simultaneously, partition processing identifiers and anti-counterfeiting traceability marks are embedded. The partition processing identifier is a background mark in the form of an invisible watermark, used to distinguish core business areas from non-core anti-counterfeiting areas. The anti-counterfeiting traceability mark is a unique encrypted character sequence, corresponding to the parameters of the entire document generation process. The mark is only used for background verification and does not interfere with the visual display and data reading of the document. Finally, an electronic delivery document with non-standard anti-counterfeiting, data stability, and scenario adaptability is generated.
[0056] By performing secondary grayscale inverse restoration on the non-core anti-counterfeiting areas of the visual fusion image, the anti-counterfeiting features are made visible under normal display. At the same time, dual-layer partition verification is carried out to ensure that the core business data is clear and readable and the anti-counterfeiting layout is complete and effective. Finally, an electronic delivery document integrating an invisible watermark and a unique encrypted sequence is generated, realizing the deep binding of anti-counterfeiting information and the document subject, and ensuring the overall visual consistency of the document and the traceability and verification capabilities of the backend.
[0057] In the application, the various formulas mentioned are all calculated by removing dimensions and taking their numerical values. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The coefficients in the formulas are set by those skilled in the art according to the actual situation.
[0058] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, and combinations thereof. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.
[0059] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.
[0060] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.
Claims
1. A delivery document generation system for delivery services, characterized in that: include: The data acquisition module acquires raw images of physical paper delivery documents, generates a template base canvas based on the standard delivery document template, performs partitioned pixel scattering on the raw images and template base canvas, generates physical pixel units and template pixel units, and constructs a hybrid visual pixel pool. The visual substrate processing module extracts physical pixel units from the hybrid visual pixel pool to form a pure physical document substrate pixel group. It then performs an inverse grayscale conversion operation on this pixel group, subtracting the original grayscale value from 255 for reverse mapping to generate an inverse visual substrate image. The module performs secondary fragmentation processing on the template pixel units, aggregating neighborhoods with a threshold of 5-10 pixels to form template micro-pixel fragments. It then iterates through the generated inverse visual substrate image row by row and column by column, calculating the grayscale difference between each group of horizontally and vertically adjacent pixels. Pixels with grayscale differences greater than a preset fixed threshold are selected and marked. Each fragment placement node is designated as a fragment placement location node. A circular placement range with a radius of 10 to 20 pixels is defined with each fragment placement location node as the center. Single template micro-pixel fragments are randomly placed into the corresponding circular range. During the placement process, the fragments are randomly assigned a rotation angle of 0° to 360° to complete the initial adsorption and positioning. The minimum distance between fragment edges is calculated. Fragments with a distance of less than 3 pixels are separated by equal-distance outward expansion, and fragments with a distance of more than 20 pixels are moved closer together by equal-distance inward convergence. The distance between adjacent fragment edges is uniformly controlled within the range of 3 to 20 pixels to form a cross-domain visual fragment fusion base. The anti-counterfeiting character embedding module retrieves core delivery business data and generates anti-counterfeiting filler placeholder characters. It then splits the anti-counterfeiting filler placeholder characters into fragmented character units. Based on connected component analysis, it extracts the irregular embedding carrier of the cross-domain visual fragment fusion substrate and embeds the fragmented character units into the corresponding irregular embedding carrier to obtain the visual fusion image. The output module performs secondary grayscale inverse restoration on the non-core anti-counterfeiting areas of the visual fusion image, conducts dual-layer partition verification, and generates an electronic delivery document with anti-counterfeiting traceability mark after the verification is passed.
2. The delivery document generation system for delivery services according to claim 1, characterized in that: The template base canvas generation includes retrieving a standard blank electronic template of a delivery document, dividing the core business area into a non-core anti-counterfeiting area, stripping field positioning lines, text format marks, layout coordinate parameters, business field annotations, and structural association instruction information from the non-core anti-counterfeiting area, and retaining only the basic pixels of the printed font, table lines, and document outer border, while the core business area retains the original field positioning and semantic structure.
3. A delivery document generation system for delivery services according to claim 2, characterized in that: For the original physical image, background pixels exceeding the main body of the document are cropped, and the image area corresponding to the non-core anti-counterfeiting area is decomposed into independent physical pixel units, while the image area corresponding to the core business area retains the continuous pixel arrangement structure; the non-core anti-counterfeiting area of the template base canvas is decomposed into independent template pixel units, while the core business area maintains the original pixel structure unchanged; the original spatial position and arrangement order of pixels in the non-core anti-counterfeiting area are scrambled, while the original pixel arrangement and spatial positioning in the core business area are maintained, thus constructing a hybrid visual pixel pool.
4. A delivery document generation system for delivery services according to claim 1, characterized in that: Core settlement data includes information on both parties, details of the underlying asset, transaction amount, settlement time, filing number, and serial number. The anti-counterfeiting filler characters are random placeholder character streams with no actual business meaning. Internal format marks and separators are removed, structured attributes are stripped, and only plain text filler characters are retained.
5. A delivery document generation system for delivery services according to claim 4, characterized in that: A pre-defined splitting length range of 1 to 5 characters is used. Within this range, a single value is randomly selected as the threshold for a single splitting operation. The anti-counterfeiting filler placeholder characters are randomly split segment by segment into several independent and fragmented character units. The cross-domain visual fragment fusion substrate is traversed, and a connected component analysis algorithm is used to extract the gap region and independent irregular blank pixel region in the cross-domain visual fragment fusion substrate. Blank regions with an area of less than 5 pixels are removed, the boundaries of the remaining blank regions are locked, and the connected components are determined to form an independent and closed irregular embedded carrier.
6. A delivery document generation system for delivery services according to claim 5, characterized in that: The embedding of fragmented character units includes analyzing the grayscale mean, pixel area, and boundary perimeter of the irregularly shaped embedding carrier, selecting carriers that are compatible with the fragmented character units, randomly binding the fragmented character units to the compatible carriers, and adaptively adjusting the character size, rotation angle, and character spacing according to the carrier contour direction, boundary curvature, and internal pixel trajectory, so that the characters are arranged non-linearly along the internal trajectory of the carrier and completely fall into the interior of the irregularly shaped embedding carrier; and performing edge grayscale transition fusion on the pixels of the fragmented character units and the base pixels of the cross-domain visual fragment fusion, unifying the grayscale difference at the edge connection to within ±10.
7. A delivery document generation system for delivery services according to claim 1, characterized in that: The secondary grayscale inverse restoration involves performing a secondary grayscale inverse transformation on the non-core anti-counterfeiting area of the visual fusion image by subtracting the current pixel grayscale value from 255. The dual-layer partition verification includes a first layer verifying the data integrity, font clarity, and layout regularity of the core business area, and a second layer verifying the uniformity of pixel fragment distribution and the validity of anti-counterfeiting filling characters in the non-core anti-counterfeiting area. An invisible watermark-style partition processing identifier and a unique encrypted character sequence anti-counterfeiting traceability mark are also embedded.