A dark watermark encryption system and its traceability method and storage medium

Abstract

The present invention belongs to the technical field of encryption tracking and tracing, and specifically relates to a dark watermark encryption system, a tracing method and a storage medium thereof. The system comprises a dark watermark generation module, a dark watermark embedding module and a dark watermark decoding module; the dark watermark generation module generates a dark watermark according to a preset configuration, and sends its encoding to the embedding module; the embedding module requests the dark watermark from the dark watermark generation module, and uses a configurable editing tool and a virtual printing tool to embed the dark watermark; the dark watermark decoding module determines the source and the transmission path according to the dark watermark carried in the sample, and comprises a sample identification component, an element identification component and a dark watermark extraction component; the dark watermark encryption system and the tracing method and the storage medium thereof of the present invention are mainly used to create a tracking and tracing system for confidential files in a complex environment network, improve the security of confidential files in the system, and provide a basis for realizing tracking and tracing of nodes in the entire transmission chain.

Description

Technical Field

[0001] The present invention belongs to the technical field of encryption tracking and tracing, and in particular, relates to a dark watermark encryption system and a tracing method and storage medium thereof. Background Art

[0002] Dark watermark technology is an information encryption technology in the fields of message authentication, tracking and tracing, encryption and protection. It embeds hidden information into the text and then extracts the information by tracing the source. The most commonly used encoding schemes include word offset and line offset. Word offset uses the difference in text spacing to achieve hidden information encoding and embedding; line offset uses the text line spacing for encoding and embedding. Compared with the two, line offset moves a smaller and more concealed distance, and has a lower encoding difficulty. However, in general, the overall encoding capacity of the dark watermark under the two schemes is still small. At the same time, the decryption process after encryption is heavily dependent on the integrity of the sample. During the copying or transmission process, its recognizability will be directly affected by surface painting, damage, wrinkles, etc. In order to achieve full tracking and tracing of confidential information, it is necessary to effectively track the entire path of information transmission and add the corresponding information of each node to the dark watermark. As a result, the application scope of traditional solutions is limited. Generally, they can only be applied to occasions with a small number of nodes and a short information transmission chain. They are difficult to be effectively used in complex transmission paths and large-scale systems. Summary of the invention

[0003] The purpose of the present invention is to provide a dark watermark encryption system and its tracing method and storage medium suitable for realizing traceability tracking in a complex environment with multiple nodes based on actual needs.

[0004] To achieve the above purpose, the present invention adopts the following technical solution.

[0005] A dark watermark encryption system includes a dark watermark generation module, a dark watermark embedding module, and a dark watermark decoding module;

[0006] The dark watermark generation module generates a dark watermark according to a preset configuration and sends its encoding to the embedding module;

[0007] The embedding module requests the dark watermark from the dark watermark generating module, and implements the embedding of the dark watermark by using the configurable editing tool and the virtual printing tool;

[0008] The dark watermark decoding module determines the source and transmission path according to the dark watermark carried in the sample; it includes a sample identification component, an element identification component, and a dark watermark extraction component.

[0009] As a further improvement or preferred implementation of the aforementioned dark watermark encryption system, an editing tool may be configured to control text elements during information generation to achieve dark watermark embedding; a virtual printing tool may form a dark watermark on a paper text or electronic text during copying and faxing;

[0010] The sample recognition component extracts and recognizes samples to generate recognizable text; the feature recognition component extracts the feature information carried by the recognizable text; the dark watermark extraction component obtains the recognized feature information and decodes it according to the encoding scheme of the dark watermark generation module, generates the corresponding code, and uses the code to reconstruct the dark watermark.

[0011] A further improvement or preferred implementation of the aforementioned dark watermark encryption system, using a configurable editing tool to implement dark watermark embedding includes one or more of the following steps:

[0012] A0. Convert the file to be marked into a graphic format file P H×W ={x(h,w)}, where x(h,w) represents the grayscale value at the coordinate (h,w), w∈(0,W-1), h∈(0,H-1), W represents the text image width, and H represents the text image height;

[0013] A1. Word spacing recognition

[0014] For image P H×W Project in the vertical direction to determine the image P H×W Horizontal profile and vertical profile The horizontal profile l1(h) and the vertical profile l2(w) are used to determine the row and column division information of the text:

[0015] Right now

[0016] A2. Spacing standardization

[0017] Determine whether there is a special format based on the word spacing range, that is, for k1AVG i,j ≤d i,j or i,j ≤k2AVG i,j , then it is considered that the jth word spacing of the i-th line of text has a special format, where k1>k2 is the preset judgment threshold, d i,j Indicates the spacing value of the jth character in the i-th line of text; is the average word spacing in the ith row, where N i represents the number of word spacings between the i-th line of text, J represents the number of words in the text, and d j Indicates the spacing value of the j-th word spacing;

[0018] Keep the character spacing of special formats unchanged, count the spacing values ​​of the remaining character spacing and calculate the average spacing, adjust the spacing value of the remaining character spacing to the average spacing value, and count whether the pixel width of the total character spacing of each row after adjustment changes. If it changes, the pixel difference corresponding to the width change needs to be evenly distributed to each character spacing to obtain the standardized character spacing value

[0019] A3. Embedding feasibility test

[0020] According to the dark watermark code given by the dark watermark generation module and the preset encoding displacement Δ, the displacement sequence {Δ k}, Δ k ={Δ k,1 ,Δ k,2 ...Δ k,t ...Δ k,t} represents the displacement sequence corresponding to the k-th row watermark code value, Δ k,t Indicates the displacement corresponding to the t-th watermark code value in the k-th row;

[0021] According to the sequence order, determine the line of watermark that is not embedded k , in text order, select all non-special format word spacing {d i,j};

[0022] According to the sequence order, select the spacing d of the unembedded displacement i,j , the watermark Δ k The displacement Δ not embedded in k,t Pre-embed and calculate the word spacing d' after embedding i,j =d i,j +Δ k,t , whether the following conditions are met:

[0023] Where d0 refers to the character spacing that can be recognized by the naked eye;

[0024] If the condition is met, the spacing d of the displacement is not embedded. i,j It can be used as an embedding location and marked, otherwise it cannot be used as an embedding location;

[0025] Check whether the marked character spacing of the line of text can ensure the watermark Δ k Can be fully embedded in N k If it is possible, the embedding step is performed. If not, the watermark Δ k The displacement that can be embedded in the watermark is incompletely embedded, and the next line is changed to repeat the test until the watermark Δ k Can be fully embedded in N k Repeat the above steps until all watermark rows can be embedded, where Nk >1 is the error tolerance factor to ensure that the complete watermark line can be extracted as much as possible when the text line is not saved completely;

[0026] Select the next watermark line and the next unembedded text line, and repeat the above steps until all watermark lines can be embedded;

[0027] A4, watermark embedding

[0028] For the watermark Δ that can be embedded k and the corresponding word spacing {d i,j}; Position each character spacing d i,j and its corresponding watermark displacement. After each complete embedding of a line of watermark, it will be repeatedly embedded from the next group spacing until the remaining character spacing is insufficient to embed a line of watermark. Then, part or all of the watermark of the next line of watermark will be embedded into the remaining character spacing. Regardless of whether it is completely embedded or not, this embedding does not affect the subsequent embedding plan of the next line of watermark.

[0029] A further improvement or preferred implementation of the aforementioned dark watermark encryption system, wherein the dark watermark extraction component obtains the identified element information and comprises the steps of extracting chart element information:

[0030] Acquire or synthesize chart data used in the system, and perform partial cropping and rotation on chart samples to expand the diversity of sample data;

[0031] A graph feature extraction model is established based on the variable convolution residual network ResNet, and feature fusion is performed using the feature pyramid network FPN. In each fusion process, the multi-level target detection model Cascade RCNN is used to check and adjust the feature extraction results, and graph samples are input into the model for graph area detection training.

[0032] A multi-dimensional feature extraction model for graphs is established based on the deep residual ResNet network. The backbone network output is used as input, and the feature pyramid network FPN is used to perform convolution operations to fuse features of different dimensions. The prediction graph for optimizing the model is generated through the Head detection network, and the graph samples are input into the model for text block detection training.

[0033] The model is trained using samples. Once the training results meet the requirements, the chart samples to be analyzed are input into the model for chart area detection and text block detection.

[0034] A further improvement or preferred implementation of the aforementioned dark watermark encryption system, implementing dark watermark embedding based on a virtual printing tool also includes: using a dynamic link library to build a printer hardware port monitor, establishing a communication link between the virtual printer and the hardware port driver: using the port monitor to obtain printing tasks, calling the StartDocPort function to use the Windows API to establish a data link, and realizing data exchange with the hardware through read and write functions; customizing the virtual printing handler, establishing a Botify registry key value to determine the printer type, establishing a Printers registry key value to determine the monitoring status, and establishing a Creat path definition function to determine the output path.

[0035] A further improvement or preferred implementation of the aforementioned dark watermark encryption system, controlling text elements to achieve dark watermark embedding means utilizing character spacing configuration, text format configuration, and text information configuration related elements to achieve dark watermark embedding;

[0036] Character spacing configuration refers to controlling the character spacing and line spacing based on the parameters generated by the dark watermark generation module to achieve dark watermark encoding embedding;

[0037] Text format configuration refers to implementing embedding of a specific dark watermark by adjusting and controlling one or more format configuration parameters based on parameters generated by a dark watermark generation module through a dedicated text format;

[0038] Virtual printing parameters refer to virtual printing parameters that are not directly related to information content;

[0039] The format configuration parameters include table row and column spacing parameters; the dark watermark generation module converts the code into row and column spacing parameter values ​​of one or more rows in the table, generates format configuration parameters and applies them to the table style in the text. The format configuration parameters refer to the parameters used to control the text style, layout and other information styles during the text editing process.

[0040] A further improvement or preferred implementation of the aforementioned dark watermark encryption system, the virtual printing parameters include pixel values, the dark watermark generation module obtains the pixel value of the character or string in the standard text virtual printing parameters, compares it with the pixel value of the current row, or current column, or the row and column fragment containing the character or string as a standard value, converts the code into a pixel value ratio, generates a new pixel value ratio, and generates printing parameters to send to the virtual printing tool, and the virtual printing tool controls the corresponding printing and copying device to adjust the pixel value of the corresponding character according to the virtual printing parameters:

[0041] The specific steps of generating dark watermark based on pixel value include:

[0042] Establish a standardized text database including all kinds of texts used in the system, and determine the standard pixel value database of text blocks in various formats based on unified virtual printing parameters;

[0043] According to the dark watermark code given by the dark watermark generation module and the preset pixel variable Δ, the pixel variable sequence {Δ k}, Δ k ={Δ k,1 ,Δ k,2 ...Δ k,t ...Δ k,T} represents the pixel variable sequence corresponding to the watermark code value of the kth row;

[0044] Get the text line sequence {R i} and the standard pixel value of each line of text;

[0045] Select the text row R that is not embedded with pixel variables in the order of sequence. i , R i Represents the i-th text line in the text sequence;

[0046] The watermark Δ k The corresponding pixel variable is embedded in the i-th text block in the text line, and the pixel variable Δ is embedded k,t The pixel value of the text block can be expressed as q i represents the pixel value of the i-th text block in the text sequence, and n>1 indicates the number of text blocks that are limited to single-pixel variable embedding;

[0047] Check whether the line of text can guarantee the watermark Δ k Can be fully embedded in N k If it is possible, the embedding step is performed. If not, the next line is changed and the test is repeated until the watermark Δ k Can be fully embedded in N k Repeat the above steps until all watermark rows can be embedded, where N k >1 is the error tolerance factor to ensure that the complete watermark line can be extracted as much as possible when the text line is not saved completely; after the end, select the next line of watermark and the next line of text that is not embedded, and repeat the above steps until all watermark lines can be embedded.

[0048] The present invention also provides a source tracing method for a dark watermark encryption system, comprising the following steps:

[0049] Step 1: Establish a dark watermark traceability network, including establishing the software and hardware platform and transmission network of the dark watermark generation module, encoding module, and printing and decoding module, and publishing the corresponding configurable editing tools and virtual printing tools in the embedded modules at the information generation and transmission nodes in the system;

[0050] Step 2: Create dark watermarks for the corresponding nodes according to the rules of the dark watermark generation component, use the encoding component to encode the dark watermarks and store them in the encoding module;

[0051] Step 3: For editable information, during the information generation and transmission process, the embedding module is used to request the dark watermark code of the corresponding node from the encoding module to the encoding component, and the dark watermark is generated in the information file through the configurable editing tool and the virtual printing tool; for non-editable information, the information data is obtained through the graphic recognition component and the element recognition component in the decoding system, and the embedding module is used to request the dark watermark code of the corresponding node from the encoding module to the encoding component, and the dark watermark is generated in the information file through the virtual printing tool;

[0052] Step 4: Use the decoding module to decode the message to be analyzed, obtain the dark watermark content, and extract the dark watermark according to the dark watermark content to determine the transmission node.

[0053] The present invention also provides a dark watermark encryption system storage medium, including: a dark watermark module storage medium, a configurable editing tool storage medium, a virtual printing tool storage medium, and a dark watermark encoding storage medium;

[0054] The dark watermark module storage medium refers to a storage medium storing programs or driver data for creating or updating various functional modules of the dark watermark encryption system;

[0055] Configurable editing tool storage medium refers to a storage medium storing the body and plug-in data used to create or update the configurable editing tool software or driver;

[0056] Virtual printing tool storage medium refers to a storage medium storing a body, plug-in or driver data used to create or update virtual printing tool software or driver;

[0057] The dark watermark encoding storage medium refers to the storage medium storing the dark watermark data of each node in the dark watermark encryption system.

[0058] The storage medium of the dark watermark encryption system includes a PC, a hard disk or a mobile device.

[0059] Its beneficial effects are:

[0060] The dark watermark encryption system, traceability method and storage medium of the present invention are mainly used to support the creation of tracking and tracing of confidential files in complex environment networks, and to improve the security of confidential files in the system. The present invention fully considers the current needs of electronic office and traditional paper document office use, and uses appropriate dark watermark generation and embedding methods to embed dark watermarks in confidential files at different transmission nodes, providing a basis for tracking and tracing nodes in the entire transmission chain. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 It is a functional mechanism diagram of the dark watermark encryption system of the present invention;

[0062] Figure 2 This is a schematic diagram of the workflow of the dark watermark encryption and traceability system. DETAILED DESCRIPTION

[0063] The present invention is described in detail below in conjunction with specific embodiments.

[0064] The dark watermark tracing system of the present invention is mainly used for tracking and tracing confidential text files. In order to meet the different production and transmission methods of confidential information, the present invention adopts targeted processing measures. For editable information, dark watermark embedding can be achieved through configurable editing tools, and node transmission path marking can be achieved simply and efficiently, thereby building an efficient and easy-to-publish dark watermark tracing network on nodes such as printers and fax machines.

[0065] To meet the above requirements, the dark watermark traceability system of the present invention mainly includes a dark watermark generation module, a dark watermark generation module, and a dark watermark decoding module;

[0066] The dark watermark generation module generates one or more dark watermarks according to the preset configuration and sends them to the dark watermark generation module; wherein the dark watermark refers to a code or graphic used to characterize the node ID and / or the device ID and / or the time information and / or the node location and / or the copy parameter information.

[0067] In order to realize dark watermark marking, it is necessary to use the dark watermark generation module to generate the required dark watermark. The dark watermark is equivalent to the identity card of each node. The transmission path or generation node of confidential information is confirmed through the dark watermark to confirm the leakage stage and leakage node. Its general form is binary digital code or graphic code after binary processing.

[0068] The dark watermark generation module identifies and reads the dark watermark, encodes it, and extracts the corresponding dark watermark code and sends it to the embedding module when the embedding component requests it;

[0069] The embedding module requests the dark watermark of the corresponding node or the associated node from the dark watermark generating module, and implements the dark watermark embedding by using the configurable editing tool and the virtual printing tool;

[0070] The configurable editing tool can be embedded in or used as a separate editing tool of the information generation program to control text elements in the editable information generation process to achieve dark watermark embedding;

[0071] Controlling text elements to achieve dark watermark embedding means using character spacing configuration, text format configuration, and text information configuration related elements to achieve dark watermark embedding;

[0072] Character spacing configuration refers to controlling the character spacing and line spacing based on the parameters generated by the dark watermark generation module to achieve dark watermark encoding embedding;

[0073] As a preferred solution, implementing dark watermark coding embedding based on character spacing and line spacing includes one or more of the following steps:

[0074] A0. Convert the file to be marked into a graphic format file P H×W ={x(h,w)}, where x(h,w) represents the grayscale value at the coordinate (h,w), w∈(0,W-1), h∈(0,H-1), W represents the text image width, and H represents the text image height;

[0075] A1. Word spacing recognition

[0076] For image P H×W Project in the vertical direction to determine the image P H×W Horizontal profile and vertical profile The horizontal profile l1(h) and the vertical profile l2(w) are used to determine the row and column division information of the text:

[0077] Right now

[0078] A2. Spacing standardization

[0079] Determine whether there is a special format based on the word spacing range, that is, for k1AVG i,j ≤d i,j or i,j ≤k2AVG i,j , then it is considered that the jth word spacing of the i-th line of text has a special format, where k1>k2 is the preset judgment threshold, d i,j Indicates the spacing value of the jth character in the i-th line of text; is the average word spacing in the ith row, where N i represents the number of word spacings between the i-th line of text, J represents the number of words in the text, and d j Indicates the spacing value of the j-th word spacing;

[0080] Keep the character spacing of special formats unchanged, count the spacing values ​​of the remaining character spacing and calculate the average spacing, adjust the spacing value of the remaining character spacing to the average spacing value, and count whether the pixel width of the total character spacing of each row after adjustment changes. If it changes, the pixel difference corresponding to the width change needs to be evenly distributed to each character spacing to obtain the standardized character spacing value

[0081] A3. Embedding feasibility test

[0082] According to the dark watermark code given by the dark watermark generation module and the preset encoding displacement Δ, the displacement sequence {Δ k}, Δ k ={Δk,1 ,Δ k,2 ...Δ k,t ...Δ k,T} represents the displacement sequence corresponding to the k-th row watermark code value, Δ k,t Indicates the displacement corresponding to the t-th watermark code value in the k-th row;

[0083] According to the sequence order, determine the line of watermark that is not embedded k , in text order, select all non-special format word spacing {d i,j};

[0084] According to the sequence order, select the spacing d of the unembedded displacement i,j , the watermark Δ k The displacement Δ not embedded in k,t Pre-embed and calculate the word spacing d' after embedding i,j =d i,j +Δ k,t , whether the following conditions are met:

[0085] Where d0 refers to the character spacing that can be recognized by the naked eye;

[0086] If the condition is met, the spacing d of the displacement is not embedded. i,j It can be used as an embedding location and marked, otherwise it cannot be used as an embedding location;

[0087] Check whether the marked character spacing of the line of text can ensure the watermark Δ k Can be fully embedded in N k If it is possible, the embedding step is performed. If not, the watermark Δ k The displacement that can be embedded in the watermark is incompletely embedded, and the next line is changed to repeat the test until the watermark Δ k Can be fully embedded in N k Repeat the above steps until all watermark rows can be embedded, where N k >1 is the error tolerance factor to ensure that the complete watermark line can be extracted as much as possible when the text line is not saved completely;

[0088] Select the next watermark line and the next unembedded text line, and repeat the above steps until all watermark lines can be embedded;

[0089] A4, watermark embedding

[0090] For the watermark Δ that can be embedded k and the corresponding word spacing {d i,j}; Position each character spacing d i,jand its corresponding watermark displacement. After each complete embedding of a line of watermark, it will be repeatedly embedded from the next group spacing until the remaining character spacing is insufficient to embed a line of watermark. Then, part or all of the watermark of the next line of watermark will be embedded into the remaining character spacing. Regardless of whether it is completely embedded or not, this embedding does not affect the subsequent embedding plan of the next line of watermark.

[0091] Text format configuration refers to implementing embedding of a specific dark watermark by adjusting and controlling one or more format configuration parameters based on parameters generated by a dark watermark generation module through a dedicated text format;

[0092] The format configuration parameters include table row and column spacing parameters; the dark watermark generation module converts the code into row and column spacing parameter values ​​of one or more rows in the table, and generates format configuration parameters and applies them to the table style in the text, wherein the format configuration parameters refer to parameters used to control text style, layout and other information styles during the text editing process.

[0093] The virtual printing tool can be embedded in the carbon copy device driver control program to control the carbon copy configuration or parameters to form a dark watermark on the paper text or electronic text during the copying and faxing process; the carbon copy configuration or parameters refer to the carbon copy parameters that are not directly related to the information content.

[0094] The implementation of dark watermark embedding based on the virtual printing tool includes: using the dynamic link library to build a printer hardware port monitor, establishing a communication link between the virtual printer and the hardware port driver: using the port monitor to obtain the printing task, calling the StartDocPort function to use the Windows API to establish a data link, and realizing data exchange with the hardware through read and write functions; customizing the virtual printing handler, establishing the Botify registry key value to determine the printer type, establishing the Printers registry key value to determine the monitoring status, and establishing the Creat path definition function to determine the output path.

[0095] In order to ensure the dark watermark marking of photocopies and faxes, when the photocopied documents cannot be edited, the specific parameters or display effects of the carbon copy materials are controlled by using carbon copy parameters, so that confidential documents such as paper copies and faxes generated by carbon copy machines can be marked. It can usually be done by controlling the carbon copy effects such as line thickness, light and dark changes of the carbon copy itself, or by forming specific content on the paper document to mark it at night, including simulating the marking as traces and ink marks caused by carbon copy quality problems. However, the concealment effect of the dark watermark formed by the above method is general, and can be detected by the naked eye under certain circumstances. Taking into account the characteristics of carbon copying, the present invention considers realizing dark watermark embedding by controlling the carbon copy pixel value parameters.

[0096] As a preferred solution, a dark watermark generation module can be used to obtain the pixel value of a character or string in a standard text copy parameter, and the pixel value is compared with the pixel value of the row, column, or row and column fragment containing the character or string as a standard value, and the code is converted into a pixel value ratio to generate a new pixel value ratio, and a copy configuration or parameter is generated and sent to a virtual printing tool. The virtual printing tool controls the corresponding copying device to adjust the pixel value of the corresponding character in the copying and printing process according to the copy configuration or parameter. Since the pixel value itself is not significant to the naked eye, the pixel value allocation can be further optimized by combining the density of the font glyphs, which can effectively reduce the changes caused by the change of pixel value. At the same time, the encoding scheme using the pixel value ratio can use the ratio between one or more characters to realize encoding, which effectively expands the encoding capacity of the dark watermark and increases the amount of information that the dark watermark can carry, which is conducive to use in longer data links and more complex working environments.

[0097] The following is a specific implementation plan:

[0098] Specific steps to generate dark watermark based on pixel value:

[0099] Establish a standardized text database including all kinds of texts used in the system, and determine the standard pixel value database of text blocks in various formats based on unified virtual printing parameters;

[0100] According to the dark watermark code given by the dark watermark generation module and the preset pixel variable Δ, the pixel variable sequence {Δ k}, Δ k ={Δ k,1 ,Δ k,2 ...Δ k,t ...Δ k,T} represents the pixel variable sequence corresponding to the watermark code value of the kth row;

[0101] Get the text line sequence {R i} and the standard pixel value of each line of text;

[0102] Select the text row R that is not embedded with pixel variables in the order of sequence. i , R i Represents the i-th text line in the text sequence;

[0103] The watermark Δ k The corresponding pixel variable is embedded in the i-th text block in the text line, and the pixel variable Δ is embedded k,t The pixel value of the text block can be expressed as q j represents the pixel value of the jth text block in the text sequence, and n>1 indicates the number of text blocks that are limited to single-pixel variable embedding;

[0104] Check whether the line of text can guarantee the watermark Δ k Can be fully embedded in N k If it is possible, the embedding step is performed. If not, the next line is changed and the test is repeated until the watermark Δ k Can be fully embedded in N k Repeat the above steps until all watermark rows can be embedded, where N k >1 is the error tolerance factor to ensure that the complete watermark line can be extracted as much as possible when the text line is not saved completely; after the end, select the next line of watermark and the next line of text that is not embedded, and repeat the above steps until all watermark lines can be embedded.

[0105] The decoding module obtains the dark watermark carried in the information sample by identifying, reading and decoding, so as to determine the source of the sample and the transmission path of the sample; it includes a sample identification component, an element identification component, a replication configuration or parameter identification component, and a dark watermark extraction component;

[0106] The sample recognition component extracts and recognizes the graphic characters in the information sample, and generates recognizable text that can be read and analyzed by the feature recognition component; the feature recognition component extracts the feature information carried by the recognizable text;

[0107] Current graphic and text recognition technology is already able to achieve high-precision text and graphic extraction and recognition. By utilizing graphic pixels, glyphs and fonts, it is easy to extract information content from raw materials, and then easily obtain the original information and dark watermark information carried in the material.

[0108] 1. The steps for extracting chart element information include:

[0109] Acquire or synthesize chart data used in the system, and perform partial cropping and rotation on chart samples to expand the diversity of sample data;

[0110] A graph feature extraction model is established based on the variable convolution residual network ResNet, and feature fusion is performed using the feature pyramid network FPN. In each fusion process, the multi-level target detection model Cascade RCNN is used to check and adjust the feature extraction results, and graph samples are input into the model for graph area detection training.

[0111] A multi-dimensional feature extraction model for graphs is established based on the deep residual ResNet network. The backbone network output is used as input, and the feature pyramid network FPN is used to perform convolution operations to fuse features of different dimensions. The prediction graph for optimizing the model is generated through the Head detection network, and the graph samples are input into the model for text block detection training.

[0112] The model is trained using samples. Once the training results meet the requirements, the chart samples to be analyzed are input into the model for chart area detection and text block detection.

[0113] The replication configuration or parameter identification component extracts and identifies the replication configuration or parameters carried by the sample;

[0114] The dark watermark extraction component obtains the identified element information, replication configuration or parameters and decodes them accordingly according to the encoding scheme of the dark watermark generation module, generates corresponding codes, and uses the codes to reconstruct the dark watermark.

[0115] Through various forms of dark watermark embedding and distributed tag information embedding methods, the retention and recognizability of dark watermarks can be effectively improved in cases where confidential file information is incomplete, or carbon copies are painted or deformed.

[0116] like Figure 2 As shown, based on the aforementioned dark watermark encryption system, the present invention proposes a corresponding dark watermark encryption system tracing method, which mainly includes the following steps:

[0117] Step 1: Establish a dark watermark traceability network, including establishing the software and hardware platform and transmission network of the dark watermark generation module, encoding module, and printing and decoding module, and publishing the corresponding configurable editing tools and virtual printing tools in the embedded modules at the information generation and transmission nodes in the system;

[0118] Step 2: Create dark watermarks for the corresponding nodes according to the rules of the dark watermark generation component, use the encoding component to encode the dark watermarks and store them in the encoding module;

[0119] Step 3: For editable information, during the information generation and transmission process, the embedding module is used to request the dark watermark code of the corresponding node from the encoding module to the encoding component, and the dark watermark is generated in the information file through the configurable editing tool and the virtual printing tool; for non-editable information, the information data is obtained through the graphic recognition component and the element recognition component in the decoding system, and the embedding module is used to request the dark watermark code of the corresponding node from the encoding module to the encoding component, and the dark watermark is generated in the information file through the virtual printing tool;

[0120] Step 4: Use the decoding module to decode the message to be analyzed, obtain the dark watermark content, and extract the dark watermark according to the dark watermark content to determine the transmission node.

[0121] The hardware and software platform includes PC, key disk and encryption program.

[0122] The present invention also provides a storage medium for pointing out and serving the aforementioned dark watermark encryption system and traceability method, including a module storage medium, a configurable editing tool storage medium, a virtual printing tool storage medium, and a dark watermark encoding storage medium;

[0123] The dark watermark module storage medium refers to a storage medium storing programs or driver data for creating or updating various functional modules of the dark watermark encryption system;

[0124] The configurable editing tool storage medium refers to a storage medium storing the body and plug-in data used to create or update the configurable editing tool software or driver;

[0125] The virtual printing tool storage medium refers to a storage medium storing the body, plug-in or driver data used to create or update the virtual printing tool software or driver;

[0126] The dark watermark encoding storage medium refers to the storage medium storing the dark watermark data of each node in the dark watermark encryption system.

[0127] Traditional confidentiality procedures require managers of key nodes to add specific node watermarks (including signatures, seals, handprints, etc.), which is convenient for management on the one hand, and can realize rapid positioning of nodes on the other hand. At each transmission, the position and form of the above-mentioned node watermarks may undergo unknown changes. In order to achieve full node tracking, the present invention adds dark watermarks to the copy transmission process of each transmission node of the confidential text based on dark watermarks, which can be compatible with traditional confidentiality procedures. While tracing the dark watermarks, it can identify and extract related node watermarks, thereby realizing all-round tracking and positioning. At the same time, the fusion of dark watermarks and node watermarks can also effectively deal with counterfeit imprints, stolen imprints and other behaviors for effective identification and tracing.

[0128] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention, rather than to limit the scope of protection of the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the essence and scope of the technical solution of the present invention.

Claims

1. A dark watermark encryption system, characterized in that: It includes a dark watermark generation module, a dark watermark embedding module, and a dark watermark decoding module; The dark watermark generation module generates a dark watermark according to a preset configuration and sends its encoding to the embedding module; The embedding module requests the dark watermark from the dark watermark generating module, and implements the embedding of the dark watermark by using the configurable editing tool and the virtual printing tool; The dark watermark decoding module determines the source and transmission path according to the dark watermark carried in the sample, including a sample identification component, an element identification component, and a dark watermark extraction component; Implementing dark watermark embedding using the configurable editing tool includes one or more of the following steps: A0. Convert the file to be marked into a graphic format file P H×W ={x(h,w)}, where x(h,w) represents the grayscale value at the coordinate (w,h), w∈(0,W-1), h∈(0,H-1), W represents the text image width, and H represents the text image height; A1. Word spacing recognition For image P H×W Project in the vertical direction to determine the image P H×W Horizontal profile and vertical profile The horizontal profile l1(h) and the vertical profile l2(w) are used to determine the row and column division information of the text, which can be specifically expressed as: A2. Spacing standardization Determine whether there is a special format based on the word spacing range, that is, for k1AVG i,j ≤d i,j or i,j ≤k2AVG i,j , then it is considered that the jth character spacing of the i-th line of text has a special format, where k1>k2 is the preset judgment threshold, d i,j Indicates the spacing value of the jth character in the i-th line of text; is the average word spacing in the ith row, where N i represents the number of word spacings in the i-th line of text, J represents the number of words in the text, and d j Indicates the spacing value of the j-th word spacing; Keep the character spacing of special formats unchanged, count the spacing values ​​of the remaining character spacing and calculate the average spacing, adjust the spacing value of the remaining character spacing to the average spacing value, and count whether the pixel width of the total character spacing of each row after adjustment changes. If it changes, the pixel difference corresponding to the width change needs to be evenly distributed to each character spacing to obtain the standardized character spacing value A3. Embedding feasibility test According to the dark watermark code given by the dark watermark generation module and the preset encoding displacement Δ, the displacement sequence {Δ k }, Δ k ={Δ k,1 ,Δ k,2 ...Δ k,t ...Δ k,T } represents the displacement sequence corresponding to the watermark code value of the kth row, Δ k,t Indicates the displacement corresponding to the t-th watermark code value in the k-th row; According to the sequence order, determine the line of watermark that is not embedded k , in text order, select all non-special format word spacing {d i,j }; According to the sequence order, select the spacing d of the unembedded displacement i,j , the watermark Δ k The displacement Δ not embedded in k,t Pre-embed and calculate the word spacing d' after embedding i,j =d i,j +Δ k,t , whether the following conditions are met: Where d0 refers to the character spacing that can be recognized by the naked eye; If the condition is met, the spacing d of the displacement is not embedded. i,j It can be used as an embedding location and marked, otherwise it cannot be used as an embedding location; Check whether the marked character spacing of the line of text can ensure the watermark Δ k Can be fully embedded in N k If it is possible, the embedding step is performed. If not, the watermark Δ k The displacement that can be embedded in the watermark is incompletely embedded, and the next line is changed to repeat the test until the watermark Δ k Can be fully embedded in N k Repeat the above steps until all watermark rows can be embedded, where N k >1 is the error tolerance factor to ensure that the complete watermark line can be extracted as much as possible when the text line is not saved completely; Select the next watermark line and the next unembedded text line, and repeat the above steps until all watermark lines can be embedded; A4, watermark embedding For the watermark Δ that can be embedded k and the corresponding word spacing {d i,j }; Position each character spacing d i,j and its corresponding watermark displacement. After each complete embedding of a line of watermark, it will be repeatedly embedded from the next group spacing until the remaining character spacing is insufficient to embed a line of watermark. Then, part or all of the watermark of the next line of watermark will be embedded into the remaining character spacing. Regardless of whether it is completely embedded or not, this embedding does not affect the subsequent embedding plan of the next line of watermark.

2. The dark watermark encryption system according to claim 1, characterized in that: The configurable editing tool controls text elements during information generation to achieve dark watermark embedding; the virtual printing tool forms a dark watermark on paper text or electronic text during the copy and fax process; The sample recognition component extracts and recognizes samples to generate recognizable text; the feature recognition component extracts the feature information carried by the recognizable text; the dark watermark extraction component obtains the recognized feature information and decodes it accordingly according to the encoding scheme of the dark watermark generation module, and uses the code to reconstruct the dark watermark.

3. The dark watermark encryption system according to claim 2, characterized in that: The element information extracted by the dark watermark extraction component includes chart element information, and the steps for extracting the chart element information include: Acquire or synthesize chart data used in the system, and perform partial cropping and rotation on chart samples to expand the diversity of sample data; A graph feature extraction model is established based on the variable convolution residual network ResNet, and feature fusion is performed using the feature pyramid network FPN. In each fusion process, the multi-level target detection model Cascade RCNN is used to check and adjust the feature extraction results, and graph samples are input into the model for graph area detection training. A multi-dimensional feature extraction model for graphs is established based on the deep residual ResNet network. The backbone network output is used as input, and the feature pyramid network FPN is used to perform convolution operations to fuse features of different dimensions. The prediction graph for optimizing the model is generated through the Head detection network, and the graph samples are input into the model for text block detection training. The model is trained using samples. Once the training results meet the requirements, the chart samples to be analyzed are input into the model for chart area detection and text block detection.

4. The dark watermark encryption system according to claim 1, characterized in that: The implementation of dark watermark embedding based on the virtual printing tool includes: using the dynamic link library to build a printer hardware port monitor, establishing a communication link between the virtual printer and the hardware port driver: using the port monitor to obtain the printing task, calling the StartDocPort function to use the Windows API to establish a data link, and realizing data exchange with the hardware through read and write functions; customizing the virtual printing handler, establishing the Botify registry key value to determine the printer type, establishing the Printers registry key value to determine the monitoring status, and establishing the Creat path definition function to determine the output path.

5. The dark watermark encryption system according to claim 1, characterized in that: Controlling text elements to achieve dark watermark embedding means using character spacing configuration, text format configuration, and text information configuration related elements to achieve dark watermark embedding; Character spacing configuration refers to controlling the character spacing and line spacing based on the parameters generated by the dark watermark generation module to achieve dark watermark encoding embedding; Text format configuration refers to implementing embedding of a specific dark watermark by adjusting and controlling one or more format configuration parameters based on parameters generated by a dark watermark generation module through a dedicated text format; Virtual printing parameters refer to virtual printing parameters that are not directly related to information content; The format configuration parameters include table row and column spacing parameters; the dark watermark generation module converts the code into row and column spacing parameter values ​​of one or more rows in the table, generates format configuration parameters and applies them to the table style in the text. The format configuration parameters refer to the parameters used to control the text style and layout information style during the text editing process.

6. The dark watermark encryption system according to claim 5, characterized in that: The virtual printing parameters include pixel values, and the specific steps of generating a dark watermark based on the pixel values ​​include: Establish a standardized text database including all kinds of texts used in the system, and determine the standard pixel value database of text blocks in various formats based on unified virtual printing parameters; According to the dark watermark code given by the dark watermark generation module and the preset pixel variable Δ, the pixel variable sequence {Δ k }, Δ k ={Δ k,1 ,Δ k,2 ...Δ k,t ...Δ k,T } represents the pixel variable sequence corresponding to the watermark code value of the kth row; Get the text line sequence {R i } and the standard pixel value of each line of text; Select the text line R that is not embedded with pixel variables in the order of sequence. i , R i Represents the i-th text line in the text sequence; The watermark Δ k The corresponding pixel variable is embedded in the text block in the text line, and the pixel variable Δ is embedded in the text block. k,t The pixel value of the text block can be expressed as q j represents the pixel value of the jth text block in the text sequence, and n>1 indicates the number of text blocks that are limited to single-pixel variable embedding; Check whether the line of text can guarantee the watermark Δ k Can be fully embedded in N k If it is possible, the embedding step is performed. If not, the next line is changed and the test is repeated until the watermark Δ k Can be fully embedded in N k Repeat the above steps until all watermark rows can be embedded, where N k >1 is the error tolerance factor to ensure that the complete watermark line can be extracted as much as possible when the text line is not saved completely; after the end, select the next line of watermark and the next line of text that is not embedded, and repeat the above steps until all watermark lines can be embedded.

7. The traceability method based on the dark watermark encryption system of claim 1 is characterized in that: The steps include: Step 1: Establish a dark watermark traceability network, including establishing the software and hardware platform and transmission network of the dark watermark generation module, encoding module, and printing and decoding module, and publishing the corresponding configurable editing tools and virtual printing tools in the embedded modules at the information generation and transmission nodes in the system; Step 2: Create dark watermarks for the corresponding nodes according to the rules of the dark watermark generation component, use the encoding component to encode the dark watermarks and store them in the encoding module; Step 3: For editable information, during the information generation and transmission process, the embedding module is used to request the dark watermark code of the corresponding node from the encoding module to the encoding component, and the dark watermark is generated in the information file through the configurable editing tool and the virtual printing tool; For non-editable information, the information data is obtained by decoding the graphic recognition component and the element recognition component in the system, and the dark watermark code of the corresponding node is requested from the encoding module to the encoding component by using the embedding module, and the dark watermark is generated in the information file by the virtual printing tool; Step 4: Use the decoding module to decode the message to be analyzed, obtain the dark watermark content, and extract the dark watermark according to the dark watermark content to determine the transmission node.

8. The dark watermark encryption system storage medium according to claim 1, characterized in that: include: Dark watermark module storage medium, configurable editing tool storage medium, virtual printing tool storage medium, dark watermark encoding storage medium; The dark watermark module storage medium refers to a storage medium storing programs or driver data for creating or updating various functional modules of the dark watermark encryption system; Configurable editing tool storage medium refers to a storage medium storing the body and plug-in data used to create or update the configurable editing tool software or driver; Virtual printing tool storage medium refers to a storage medium storing a body, plug-in or driver data used to create or update virtual printing tool software or driver; The dark watermark encoding storage medium refers to the storage medium storing the dark watermark data of each node in the dark watermark encryption system.

9. The dark watermark encryption system storage medium according to claim 8, characterized in that: Storage media include PC, hard disk or mobile device.

Images