A robust watermarking method for authenticating 3D printed models

Abstract

The present invention provides a robust watermarking method for authenticating the authenticity of a 3D printed model, comprising a registration phase and a verification phase. The registration phase completes the robust watermark design and embedding of the 3D printed model, and the verification phase completes the authentication of the 3D printed model. The present invention converts authentication information into a binary bit string, supplemented by a check data string and a positioning flag to form embedded binary bit watermark information. An area is selected in the model for embedding the watermark. The seam structure during the printing process is used as a carrier for the watermark, and the movement direction of the printer nozzle in the embedding area is modified according to the watermark information to achieve the embedding of the watermark information. The watermark position in the model entity is located through a target detection network, and the watermark information is extracted to determine the authenticity of the model. The present invention can ensure the integrity of the model entity during the process of authenticating the authenticity of the 3D printed model, and the watermark embedding and extraction are simple, the detection results are excellent, the anti-attack capability is strong, and the robustness is high. The authentication cost is low and the practicality is strong.

Description

Technical Field

[0001] The present invention relates to the field of 3D printing technology, and in particular to a robust watermark method for authenticating the authenticity of a 3D printed model. Background Art

[0002] As a rapid prototyping technology, 3D printing offers unparalleled advantages over traditional manufacturing in terms of object design, manufacturing speed, precision, and material selection. 3D printing has been widely used in healthcare, automotive manufacturing, aerospace, and other fields. With the advancement of 3D printing technology and the reduction in costs, product imitation has become possible. Therefore, intellectual property protection and anti-counterfeiting authentication for 3D printed models are crucial.

[0003] Several methods have been proposed for authenticating 3D printed models, most of which rely on embedded watermarks. Depending on the source of the watermark, these methods can be categorized as exogenous watermarks or those that utilize traces from the printing process. While these methods possess some authentication capabilities, they suffer from limitations in terms of watermark security, complexity, and cost. These factors make it difficult to address these challenges and hinder their widespread adoption. Summary of the Invention

[0004] In response to the shortcomings of the existing technology, the present invention provides a robust watermark method for authenticating the authenticity of 3D printed models. The present invention has the ability to efficiently authenticate the authenticity of the model, and can embed the watermark into the model and extract the watermark from the model in a simple manner. In addition, the robustness of the watermark can be improved.

[0005] The technical solution of the present invention is: a robust watermark method for authenticating the authenticity of a 3D printed model, comprising a registration phase and a verification phase. The registration phase completes the robust watermark design and embedding of the 3D printed model, and the verification phase completes the authentication of the 3D printed model.

[0006] The registration phase includes the following steps:

[0007] S1), construct watermark information;

[0008] S2), performing slicing operation on the digital model to be printed;

[0009] S3), embedding watermark information into the digital model to be printed after the slicing operation;

[0010] S4), printing the digital model embedded with the watermark information to obtain a model entity containing the authentication watermark information;

[0011] S5) Using the target detection network to detect the model entity and determine the authenticity of the entity model;

[0012] The verification phase includes the following steps:

[0013] A1) After obtaining the model entity to be inspected, find out whether there is a joint structure at the corresponding position of the model entity according to the joint structure image retained in the dedicated model authentication database;

[0014] If it exists, proceed to step A2), otherwise, the model entity to be verified is a forged model;

[0015] A2) photographing the seam structure of the model entity to be inspected using the same method and equipment as in step S6) of the registration phase, and then inputting the image into the trained object detection network for watermark recognition and extraction;

[0016] A3) Check the extracted watermark information and divide the complete watermark information according to the first and last positioning flags and the empty mark;

[0017] A4) The binary bit string containing the authentication information is reversely converted into authentication information by the method of step S1) in the registration phase, and matched with the authentication information pre-stored in the dedicated model authentication database. If the match is successful, the model entity can be identified as an authentic model, otherwise it can be identified as a counterfeit model.

[0018] Preferably, in step S1), the authentication information is converted into a binary bit string, and the embedded binary bit watermark information is formed by using the check data string and the positioning flag.

[0019] Preferably, in step S1), constructing watermark information specifically includes the following steps:

[0020] S11), converting the authentication information into a binary bit string, and obtaining a cyclic redundancy check code by performing a cyclic redundancy check code calculation on the binary bit string;

[0021] S12), performing a parity check on the cyclic redundancy check code to obtain a parity check bit; and sequentially splicing the binary bit string, the cyclic redundancy check code, and the parity check bit;

[0022] S13) adding a positioning flag bit '1' and a null flag bit to the beginning of the binary bit string;

[0023] S14) Add a null mark and a positioning mark '0' at the end of the parity bit to form a complete embedded watermark information, and save the embedded watermark information in the model authentication database.

[0024] Preferably, in step S12), the splicing is to append a cyclic redundancy check code to the end of the bit string; and to append a parity check result of the cyclic redundancy check code to the end of the CRC-5 check code.

[0025] Preferably, in step S2), the slicing operation of the digital model to be printed specifically includes the following steps:

[0026] S21), importing the digital model to be printed into the slicing software, adjusting the placement of the digital model, and selecting the location of the joint structure on the inner side of the digital model and making a mark;

[0027] S22) Slicing the digital model after setting is completed is performed by slicing software to generate a G-code file.

[0028] Preferably, in step S21), the number of layers of the selected seam structure is greater than the number of embedded watermark information.

[0029] Preferably, in step S21), during the embedding process, the head end of the watermark information corresponds to the low-order printing layer of the digital model, and the tail end of the watermark information corresponds to the high-order printing layer of the digital model.

[0030] Preferably, in step S21), the G-code segment of the outermost circle of the printing layer corresponding to the seam structure area marked in step S21) is adjusted, specifically:

[0031] When the embedded watermark information bit is '1', the G-code segment is sorted in reverse order, so that the printer nozzle moves in the opposite direction to the original setting direction when printing the outermost circle;

[0032] When the embedded watermark information bit is '0', the execution order of the G-code segment is maintained, and the movement direction of the printer nozzle is kept in the outermost circle;

[0033] When the embedded watermark information bit is empty, the original set moving direction is maintained, the starting and ending coordinates of the G-code segment are set at the edge of the model, and the coordinates of each step of the code are adjusted according to the model contour.

[0034] Preferably, in step S3), the G-code code segment at the seam structure is adjusted according to the embedded watermark information to complete the embedding of the watermark information and save it as a G-code code file to be printed.

[0035] Preferably, in step S4), the G-code file to be printed is input into a 3D printer for printing to obtain a model entity containing authentication watermark information.

[0036] Preferably, in step S5), an image of the internal joint structure of the model entity is obtained, and detection is performed through a target detection network. The detection result is analyzed and compared with the original authentication information to determine the authenticity of the model entity.

[0037] Preferably, in step S6), the target detection network adopts the YOLOv10 target detection network.

[0038] Preferably, in step A3), the verification of the watermark information specifically includes the following steps:

[0039] A31), dividing the authentication information into a watermark sequence segment, a cyclic redundancy check code segment, and a parity check bit;

[0040] A32), using the parity check bit to perform parity check on the cyclic redundancy check code;

[0041] A33) Use the verified cyclic redundancy check code to verify the binary bit string containing the authentication information. Only after both verifications are passed can we proceed to step A4), otherwise the model entity is judged to be a forged model.

[0042] The beneficial effects of the present invention are:

[0043] 1. The present invention is divided into two parts: the registration phase and the verification phase. The registration phase completes the robust watermark design and embedding of the 3D printing model, and the verification phase completes the authentication of the 3D printing model.

[0044] 2. The present invention can ensure the integrity of the model entity during the process of authenticating the authenticity of 3D printed models. The watermark embedding and extraction methods are simple, the detection results are excellent, the anti-attack capability is strong, and the robustness is high. The authentication method is low-cost and highly practical.

[0045] 3. The present invention utilizes the change in the movement direction of the printer nozzle to embed watermark information, and the anti-attack performance of the watermark carrier is enhanced;

[0046] 4. The present invention uses a trained target detection model to locate and extract watermarks from images containing seam structures, thereby ensuring authentication effectiveness and reducing authentication complexity and costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 Schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0048] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings:

[0049] like Figure 1As shown, this embodiment provides a robust watermark method for authenticating the authenticity of a 3D printed model, including a registration phase and a verification phase, wherein the registration phase includes the following steps:

[0050] S1) Construct watermark information

[0051] This embodiment converts the authentication information into a binary bit string, and uses the verification data string and the positioning flag bit to form the embedded binary bit watermark information. The specific steps include the following:

[0052] S11), obtaining a cyclic redundancy check code by performing cyclic redundancy check code calculation on the binary bit string;

[0053] S12), performing a parity check on the cyclic redundancy check code to obtain a parity check bit; and sequentially splicing the binary bit string, the cyclic redundancy check code, and the parity check bit; wherein the splicing is to append the cyclic redundancy check code to the end of the bit string; and calculating the parity check result of the cyclic redundancy check code and appending it to the end of the CRC-5 check code;

[0054] S13) adding a positioning flag bit '1' and a null flag bit to the beginning of the binary bit string;

[0055] S14) Add a null mark and a positioning mark '0' at the end of the parity bit to form a complete embedded watermark information, and save the embedded watermark information in the model authentication database.

[0056] S2) Slicing the digital model to be printed

[0057] S21) Importing the digital model to be printed into the slicing software, adjusting the placement of the digital model, and setting appropriate printing parameters, then selecting the location of the seam structure on the inner side of the digital model and marking it. In this embodiment, the number of layers marked on the seam structure should be greater than the length of the complete watermark sequence to be embedded;

[0058] S22) Slicing the digital model after setting is completed is performed by slicing software to generate a G-code file.

[0059] S3) adjusting the G-code segment at the seam structure according to the embedded watermark information to complete the embedding of the watermark information and saving it as a G-code file to be printed;

[0060] In this embodiment, the head of the watermark information corresponds to the lower-order printed layers of the digital model, and the tail of the watermark information corresponds to the higher-order printed layers of the digital model. Furthermore, the G-code segment corresponding to the outermost circle of the printed layer in the seam structure area marked in step S21) is adjusted. When the embedded watermark information bit is '1', the G-code segment is reversed, causing the printer nozzle to move in the opposite direction to the originally set direction when printing the outermost circle.

[0061] When the embedded watermark information bit is '0', the execution order of the G-code segment is maintained, and the movement direction of the printer nozzle is kept in the outermost circle;

[0062] When the embedded watermark information bit is empty, the original set moving direction is maintained, the starting and ending coordinates of the G-code segment are set at the edge of the model, and the coordinates of each step of the code are adjusted according to the model contour.

[0063] S4), inputting the G-code file to be printed into the 3D printer for printing to obtain a model entity containing authentication watermark information;

[0064] S5) Obtain an image of the internal seam structure of the model entity, and detect it through a target detection network. The detection result is parsed and compared with the original authentication information to determine the authenticity of the model entity. This embodiment uses the YOLOv10 target detection network. The YOLOv10 target detection network is trained on a large number of images of the seam structure of real printed models. It can accurately identify the seam structure of the real model and extract the watermark information embedded therein.

[0065] In this embodiment, the verification phase includes the following steps:

[0066] A1) After obtaining the model entity to be inspected, find out whether there is a joint structure at the corresponding position of the model entity according to the joint structure image retained in the dedicated model authentication database;

[0067] If it exists, proceed to step A2), otherwise, the model entity to be verified is a forged model;

[0068] A2) photographing the seam structure of the model entity to be inspected using the same method and equipment as in step S6) of the registration phase, and then inputting the image into the trained object detection network for watermark recognition and extraction;

[0069] A3) Verify the extracted watermark information and divide the complete watermark information according to the first and last positioning flags and the empty mark; specifically, the following steps are included:

[0070] A31), dividing the authentication information into a watermark sequence segment, a cyclic redundancy check code segment, and a parity check bit;

[0071] A32), using the parity check bit to perform parity check on the cyclic redundancy check code;

[0072] A33) Use the verified cyclic redundancy check code to verify the binary bit string containing the authentication information. Only after both verifications are passed can we proceed to step A4), otherwise the model entity is judged to be a forged model.

[0073] A4) The binary bit string containing the authentication information is reversely converted into authentication information by the method of step S1) in the registration phase, and matched with the authentication information pre-stored in the dedicated model authentication database. If the match is successful, the model entity can be identified as an authentic model, otherwise it can be identified as a counterfeit model.

[0074] As can be seen from Tables 1 and 2, the watermark designed by the method of this embodiment has high robustness and the authentication success rate of the detection method is high, among which:

[0075] Table 1 shows the authentication success rate of the detection method under different parameter settings for the digital model

[0076]

[0077]

[0078] Table 2 Success rate of authentication after attacking and destroying the watermark on the inner side of the digital model

[0079]

[0080] The above embodiments and descriptions are only for explaining the principles and best embodiments of the present invention. Without departing from the spirit and scope of the present invention, the present invention may be subject to various changes and improvements, which shall fall within the scope of the invention to be protected.

Claims

1. A robust watermarking method for authenticating the authenticity of 3D printed models, characterized in that: It includes the registration phase and the verification phase. The registration phase completes the robust watermark design and embedding of the 3D printing model, and the verification phase completes the authentication of the 3D printing model. The registration phase includes the following steps: S1), construct watermark information; The authentication information is converted into a binary bit string, and the check data string and the positioning flag are used to form the embedded binary bit watermark information; specifically, the following steps are included: S11), obtaining a cyclic redundancy check code by performing cyclic redundancy check code calculation on the binary bit string; S12), performing a parity check on the cyclic redundancy check code to obtain a parity check bit; and sequentially splicing the binary bit string, the cyclic redundancy check code, and the parity check bit; The splicing is to append the cyclic redundancy check code to the end of the bit string; calculate the parity check result of the cyclic redundancy check code and append it to the end of the CRC-5 check code; S13) adding a positioning flag bit '1' and a null flag bit to the beginning of the binary bit string; S14), adding a null mark and a positioning mark bit '0' at the end of the parity bit to form a complete embedded watermark information, and saving the embedded watermark information in the model authentication database; S2), performing slicing operation on the digital model to be printed; S3), embedding watermark information into the digital model to be printed after the slicing operation; S4), printing the digital model embedded with the watermark information to obtain a model entity containing the authentication watermark information; S5) Using the target detection network to detect the model entity and determine the authenticity of the entity model; The verification phase includes the following steps: A1) obtaining a model entity to be inspected and searching for a joint structure at a corresponding position of the model entity; If it exists, proceed to step A2), otherwise, the model entity to be verified is a forged model; A2) photographing the seam structure of the model entity to be inspected, and then inputting the image into the trained object detection network for watermark recognition and extraction; A3) Check the extracted watermark information and divide the complete watermark information according to the first and last positioning flags and the empty mark; The verification of watermark information specifically includes the following steps: A31), dividing the authentication information into a watermark sequence segment, a cyclic redundancy check code segment, and a parity check bit; A32), using the parity check bit to perform parity check on the cyclic redundancy check code; A33) Verify the binary bit string containing the authentication information using the verified cyclic redundancy check code. Only after both verifications pass can the process proceed to step A4). Otherwise, the model entity is judged to be a forged model. A4) Reversely convert the binary bit string containing the authentication information into authentication information, and match it with the authentication information pre-stored in the dedicated model authentication database. If the match is successful, the model entity can be identified as an authentic model, otherwise it can be identified as a counterfeit model.

2. A robust watermarking method for authenticating the authenticity of a 3D printed model according to claim 1, characterized in that: In step S2), the digital model to be printed is sliced; specifically, the following steps are included: S21), importing the digital model to be printed into the slicing software, adjusting the placement of the digital model, and selecting the location of the joint structure on the inner side of the digital model and making a mark; S22) Slicing the digital model after setting is completed is performed by slicing software to generate a G-code file.

3. A robust watermarking method for authenticating the authenticity of a 3D printed model according to claim 2, characterized in that: In step S3), the watermark information is embedded into the digital model to be printed after the slicing operation, specifically: S31) adjusting the G-code segment at the seam structure according to the embedded watermark information to complete the embedding of the watermark information and saving it as a G-code file to be printed.

4. A robust watermarking method for authenticating the authenticity of a 3D printed model according to claim 3, characterized in that: In step S31), during the embedding process, the head end of the watermark information corresponds to the low-order printing layer of the digital model, and the tail end of the watermark information corresponds to the high-order printing layer of the digital model.

5. A robust watermarking method for authenticating the authenticity of a 3D printed model according to claim 4, characterized in that: In step S31), the G-code segment of the outermost circle of the printing layer corresponding to the seam structure area marked in step S21) is adjusted.

6. A robust watermarking method for authenticating the authenticity of a 3D printed model according to claim 5, characterized in that: In step S31), when the embedded watermark information bit is '1', the G-code segment is sorted in reverse order so that the printer nozzle moves in the opposite direction to the original set direction when printing the outermost circle; When the embedded watermark information bit is '0', the execution order of the G-code segment is maintained, and the movement direction of the printer nozzle is kept in the outermost circle; When the embedded watermark information bit is empty, the original set moving direction is maintained, the starting and ending coordinates of the G-code segment are set at the edge of the model, and the coordinates of each step of the code are adjusted according to the model contour.

Images