An electronic screen content tampering detection system and method based on dynamic watermarking

Abstract

The present invention discloses an electronic screen content tampering detection system and method based on dynamic watermarking. The system includes a cloud server, a display terminal electronic screen, and a security monitoring terminal. The method includes: S1, generating and uploading a display terminal key seed; S2, the cloud server generating a frame watermark key packet based on the key seed and the video; S3, downloading the video to be played and obtaining the corresponding frame watermark key packet through encrypted communication; S4, calculating the dynamic offset and rendering and outputting the watermarked image; S5, reconstructing the original image through inverse transformation based on the frame watermark key and calculating the actual offset through the phase correlation method; S6, determining tampering according to the residual metric and triggering a shutdown operation. The present invention constructs a more secure electronic screen content tampering detection scheme, monitors the legality of the video content of the electronic screen, automatically identifies and blocks illegal playback, and prevents the illegal spread of bad content.

Description

Technical Field

[0001] The present invention relates to the technical field of network electronic screen content security information technology, and particularly to an electronic screen content tampering detection system and method based on dynamic watermarking. Background Art

[0002] With the advancement of the construction of smart cities, electronic screens have become the core carriers for information dissemination in public places and are widely used in scenarios such as square advertising screens, transportation hub information screens, and digital signage in commercial complexes. However, public data shows that the number of current public display content tampering incidents increases annually, involving attack forms such as illegal advertisement implantation, sensitive information tampering, and malicious code injection.

[0003] In this context, real-time video tampering detection technology has become the core defense line for ensuring public information security, and many related technology patents have also emerged. For example, CN117336570A, a video anti-tampering system, method, electronic device, and medium based on digital watermarking. This solution adds a digital watermark to the original video and determines whether video tampering has occurred by detecting the digital watermark during video playback. The defect of this solution is that the watermark information in the video is transmitted together with the video, and there is a risk that the video watermark information can be cracked. There is also a replay attack in the case where the watermark information is not cracked, which may cause the video to not play properly. Summary of the Invention

[0004] In order to solve the security problems existing in the existing video tampering detection technology, the present invention proposes an electronic screen content tampering detection system and method based on dynamic watermarking, constructs a more secure electronic screen content tampering detection scheme, monitors the legality of the electronic screen video content, automatically identifies and blocks illegal playback, and prevents the illegal dissemination of bad content.

[0005] The present application discloses an electronic screen content tampering detection system based on dynamic watermarking, including a cloud server, a display terminal electronic screen, and a security monitoring terminal. A plurality of display terminal electronic screens are connected to the cloud server to obtain video content to be played. Each display terminal electronic screen is connected to a security monitoring terminal, and all security monitoring terminals are connected to the cloud server through TLS encrypted communication.

[0006] Preferably, the display terminal electronic screen is internally provided with a GPU module supporting programmable shaders, which is used for the stage of processing vertex coordinate transformation in the GPU rendering pipeline, and a dynamic offset is superimposed after the vertex coordinate transformation.

[0007] Preferably, the security monitoring terminal is internally provided with an FPGA module, which is used for capturing the output signal of the display terminal electronic screen.

[0008] The present application also discloses a method for detecting electronic screen content tampering based on dynamic watermarking, which is implemented based on the above-mentioned system for detecting electronic screen content tampering based on dynamic watermarking, and includes the following steps:

[0009] S1. Generate a display terminal key seed, bind the display terminal electronic screen to the security monitoring terminal, and upload the mapping relationship and the key seed;

[0010] S2. The cloud server generates a frame watermark key package based on the key seed and the video;

[0011] S3. The display terminal electronic screen downloads the video to be played, and requests the cloud server to obtain the frame watermark key package corresponding to the video through encrypted communication;

[0012] S4. The display terminal electronic screen calculates the dynamic offset, the vertex shader superimposes the dynamic offset, and performs image rendering to output the watermarked image;

[0013] S5. The acquisition terminal electronic screen captures the current watermarked frame image output, reconstructs the original image through inverse transformation based on the frame watermark key, and calculates the actual offset through the phase correlation method;

[0014] S6. Calculate the residual metric using the actual offset and the theoretical offset. When three consecutive frames of images satisfy that the residual is greater than the preset value, shut down the display terminal electronic screen and report to the cloud server.

[0015] Preferably, the S1 includes the following steps:

[0016] S11. The screen driver chip with a unique serial number of the display terminal electronic screen integrates a physically unclonable function to generate an anti-physical cloning key seed;

[0017] S12. Pair and bind the display terminal electronic screen to the security monitoring terminal to establish a mapping relationship;

[0018] S13. Transmit the mapping relationship, the device serial numbers of the display terminal electronic screen and the security monitoring terminal, and the key seed to the cloud server through TLS encrypted communication.

[0019] Preferably, the S2 includes the following steps:

[0020] S21. Generate a master key corresponding to the display terminal electronic screen for deriving the watermark key for each frame, which is generated by performing a hash calculation on the hardware identifier:

[0021]

[0022] Wherein, is the master key, is the encryption hash function, is the key seed;

[0023] S22. Derive and independently generate a frame key for each frame of the video through the master key and the frame number:

[0024]

[0025] Among them, is the frame key of the video frame ; is the counter mode encryption operation based on ; is the frame number.

[0026] S23. Calculate the content hash value of the video frame :

[0027]

[0028] Combine the hash value with the frame key to generate a watermark key:

[0029]

[0030] Among them, is the hash-based message authentication code algorithm;

[0031] S24. After performing frame-by-frame processing of S21 - S23 on the video, obtain the frame watermark key packet of the video, and store it in the cloud server corresponding to the video number and the serial number of the display terminal electronic screen device.

[0032] Preferably, the S4 includes the following steps:

[0033] S41. Obtain a key seed through the GPU module built in the display terminal electronic screen;

[0034] S42. Calculate the frame content hash value of the video frame to be played :

[0035]

[0036] Generate the watermark key for the frame to be played:

[0037]

[0038] Among them, is the frame number of the video frame to be played ;

[0039] S43. Calculate the dynamic offset according to the watermark key for the frame to be played, that is, the offset used for dynamic adjustment of vertex coordinates:

[0040]

[0041] Among them, is a uniform distribution mapping function, is a normalization coefficient;

[0042] At the same time, the key corresponding to the frame is obtained from the frame watermark key packet acquired from the cloud server , and the theoretical offset is calculated:

[0043]

[0044] Among them, is the offset component of the theoretical offset in the horizontal direction, is the offset component of the theoretical offset in the vertical direction;

[0045] S44. Through the stage of processing vertex coordinate transformation in the GPU module rendering pipeline, after the vertex coordinate transformation, the vertex shader superimposes the dynamic offset ;

[0046] The transformed vertex set is rasterized into screen pixels by the GPU module, and the frame image with watermark is rendered and output.

[0047] Preferably, the S5 includes the following steps:

[0048] S51. Use the FPGA module built in the security monitoring terminal to collect the current frame image with watermark output by the display terminal electronic screen, denoted as ;

[0049] S52. Utilize the theoretical offset , and reconstruct the original image through inverse transformation to obtain the theoretically watermark-free image:

[0050]

[0051] Among them, is the inverse transformation function;

[0052]

[0053] represents the image to be inversely transformed, , are the coordinates of the image pixels, is the inverse transformation offset, is the offset components in the horizontal and vertical directions, and the symbol represents the translation transformation operation in image processing, represents the inverse translation transformation of the coordinates, represents the original coordinates after applying the inverse translation operation;

[0054] S53. Calculate the actual offset through the phase correlation algorithm:

[0055]

[0056] Among them, represents the two-dimensional discrete Fourier transform, represents the two-dimensional inverse Fourier transform, is the complex conjugate operator, represents the peak of the Fourier transform, and the corresponding position is the actual offset of the two images, is the offset component of the actual offset in the horizontal direction, is the offset component of the actual offset in the vertical direction.

[0057] Preferably, the S6 includes the following steps:

[0058] Calculate the residual metric using the actual offset and the theoretical offset:

[0059]

[0060] When three consecutive frames of images satisfy where is a preset value, it is determined that the currently played content on the electronic screen has been tampered with, a message indicating that there is illegal playback on the display terminal electronic screen is sent to the cloud server, and the display terminal electronic screen is shut down.

[0061] Advantages of the present invention:

[0062] (1) The watermark information is separated from the video, and the downloaded video does not contain watermark information, eliminating the possibility of attackers cracking by analyzing the video watermark information after downloading, with higher security.

[0063] (2) The watermark embedding is transferred from the pixel domain to the geometric domain, taking advantage of the irreversible characteristics of the GPU rendering pipeline, making the watermark information unable to be extracted by conventional image processing means.

[0064] (3) In the process of generating the watermark information, the master key is bound to the key seed generated by the physical unclonable function of the electronic screen driver chip, realizing "one machine, one key" hardware-level protection, and the watermark information generation is bound to the video frame, realizing the dynamic change of the watermark, which can prevent replay attacks.

[0065] (4) The pseudo-random offset applied to the vertex coordinates is at the sub-pixel level, with almost zero loss of image quality. Description of the Drawings

[0066] Figure 1 is the overall framework diagram of the electronic screen content tampering detection system according to the embodiment of the present invention.

[0067] Figure 2 This is the flowchart of the method for detecting electronic screen content tampering based on dynamic watermark in an embodiment of the present invention. Detailed implementation manners

[0068] To make the objectives, technical solutions and advantages of the present application more clear and understandable, the following examples are given with reference to the accompanying drawings to further elaborate on the present application in detail.

[0069] An embodiment of the present application discloses a system for detecting electronic screen content tampering based on dynamic watermark, and its overall framework is as Figure 1 shown, including a cloud server, a display terminal electronic screen, and a security monitoring terminal. Multiple display terminal electronic screens are connected to the cloud server to obtain video content to be played. Each display terminal electronic screen is connected to a security monitoring terminal, and all security monitoring terminals are connected to the cloud server through TLS encrypted communication. The display terminal electronic screen is built-in with a GPU module supporting programmable shaders, which is used in the stage of the GPU rendering pipeline to process vertex coordinate transformation, and a dynamic offset is superimposed after vertex coordinate transformation. The security monitoring terminal is built-in with an FPGA module, which is used to capture the output signal of the display terminal electronic screen. In this embodiment, the FPGA model is Decklin 8K Pro G2.

[0070] Another embodiment of the present application discloses a method for detecting electronic screen content tampering based on dynamic watermark. Based on the above system for detecting electronic screen content tampering based on dynamic watermark, its implementation process is as Figure 2 shown, including the following steps:

[0071] S1. Generate a display terminal key seed, bind the display terminal electronic screen to the security monitoring terminal, and upload the mapping relationship and the key seed.

[0072] S11. The screen driver chip with a unique serial number of the display terminal electronic screen (NVIDIA Quadro driver board in this embodiment) integrates a physically unclonable function (PUF) to generate an anti-physical cloning key seed .

[0073] S12. Pair and bind the display terminal electronic screen to the security monitoring terminal to establish a mapping relationship;

[0074] S13. Transmit the mapping relationship, the device serial numbers of the display terminal electronic screen and the security monitoring terminal, and the key seed to the cloud server through TLS encrypted communication.

[0075] S2. The cloud server generates a frame watermark key packet based on the key seed and the video.

[0076] S21. Generate the master key corresponding to the display terminal's electronic screen , which is used to derive the watermark key for each frame and is generated by performing a hash calculation on the hardware identifier:

[0077]

[0078] Among them, is the master key, is the cryptographic hash function, is the key seed.

[0079] S22. Independently generate a frame key for each frame of the video , specifically derived through the master key and the frame number in the AES-CTR mode:

[0080]

[0081] Among them, is the frame key of the video frame , is the encryption operation in the counter mode based on , is the frame number.

[0082] S23. Calculate the content hash value for the video frame :

[0083]

[0084] Combine the hash value with the frame key to generate the watermark key :

[0085]

[0086] Among them, is the hashed message authentication code algorithm.

[0087] S24. After performing frame-by-frame processing of S21 - S23 on the video, obtain the frame watermark key packet of the video, and store it in the cloud server corresponding to the video number and the serial number of the display terminal's electronic screen device.

[0088] S3. The display terminal's electronic screen downloads the video to be played and requests the cloud server to obtain the frame watermark key packet corresponding to the video to be played through TLS encrypted communication.

[0089] S4. The display terminal's electronic screen calculates the dynamic offset and performs image rendering output. The display terminal's electronic screen calculates the dynamic offset, the vertex shader superimposes the dynamic offset, and performs image rendering output to obtain the image with watermark.

[0090] S41. Obtain a key seed through the GPU module built in the display terminal's electronic screen.

[0091] S42. For the video frame to be played Calculate the frame content hash value :

[0092]

[0093] Generate the watermark key for the frame to be played :

[0094]

[0095] Among them, is the frame number of the video frame to be played of.

[0096] S43. According to the watermark key for the frame to be played Calculate the dynamic offset, that is, the offset used for dynamic adjustment of vertex coordinates :

[0097]

[0098] Among them, is the uniform distribution mapping function, is the normalization coefficient, which takes the value of in this embodiment.

[0099] At the same time, obtain the key corresponding to the frame from the frame watermark key package obtained from the cloud server , and calculate the theoretical offset :

[0100]

[0101] Among them, is the horizontal offset component of the theoretical offset, is the vertical offset component of the theoretical offset;

[0102] S44. Through the stage in the GPU module rendering pipeline that processes vertex coordinate transformation, after vertex coordinate transformation, the vertex shader superimposes the dynamic offset . The transformed vertex set is rasterized into screen pixels by the GPU module, and the frame image with watermark (dynamic offset) is rendered and output.

[0103] S5. Detect the watermark frame by frame. Collect the current frame image with watermark output by the terminal electronic screen, reconstruct the original image through inverse transformation based on the frame watermark key, and calculate the actual offset by the phase correlation method.

[0104] S51. Use the FPGA module built into the security monitoring terminal to capture the DP / HDMI output signal at 120 fps, and collect the current frame image with a watermark (dynamic offset) output by the display terminal's electronic screen, denoted as .

[0105] S52. Utilize the theoretical offset to reconstruct the original image through inverse transformation and obtain the theoretical watermark-free image :

[0106]

[0107] where is the inverse transformation function;

[0108]

[0109] represents the image to be inversely transformed, , are the coordinates of the image pixels, is the inverse transformation offset, is the offset component in the horizontal and vertical directions, and the symbol represents the translation transformation operation in image processing, represents the inverse translation transformation of the coordinates, represents the original coordinates after applying the inverse translation operation.

[0110] S53. Calculate the actual offset through the phase correlation algorithm:

[0111]

[0112] where represents the two-dimensional discrete Fourier transform, represents the two-dimensional inverse Fourier transform, is the complex conjugate operator, represents the peak of the Fourier transform, and the corresponding position is the actual offset of the two images, is the offset component of the actual offset in the horizontal direction, is the offset component of the actual offset in the vertical direction.

[0113] S6. Calculate the residual metric using the actual offset and the theoretical offset. When three consecutive frame images satisfy that the residual is greater than the preset value, shut down the display terminal's electronic screen and report to the cloud server.

[0114] Calculate the residual metric using the actual offset and the theoretical offset:

[0115]

[0116] When three consecutive frames of images satisfy At this time, is a preset value, and in this embodiment, the value is , it is determined that the currently played content on the electronic screen has been tampered with and is being illegally played. A message indicating that there is illegal playback on the display terminal's electronic screen is sent to the cloud server, and the power supply of the screen is disconnected by triggering the power relay to shut down the display terminal's electronic screen.

[0117] In a specific embodiment, the effect of the solution of the present application is illustrated through a comparative experiment. The experimental environment is as follows:

[0118] Hardware: NVIDIA Quadro RTX 6000 GPU (CUDA cores, 24 GB of video memory);

[0119] Software: Unity rendering engine, CUDA 11.6, custom watermark embedding / extraction module;

[0120] Dataset: 3D model library (including 1,000 high-precision models), 4K video sequence (60fps dynamic scene).

[0121] The comparison baselines are as follows:

[0122] Traditional static watermark: Static watermark based on DCT / DWT, embedded in the frequency domain of the final rendered frame.

[0123] Traditional dynamic watermark: Pixel-domain dynamic watermark, generating invisible watermarks dynamically in the pixel domain or at certain fixed positions.

[0124] The results of the anti-attack test and quantization metrics obtained from the comparative experiment are shown in Tables 1 and 2, where Table 1 is the visual quality evaluation result and Table 2 is the watermark defense success rate result.

[0125] Table 1 Visual Quality Evaluation (PSNR / SSIM)

[0126]

[0127] As can be seen from Table 1, the solution proposed in the present application is significantly better than the traditional solutions in terms of PSNR (>39 dB) and SSIM (>0.96).

[0128] Table 2 Watermark Defense Success Rate (Success Rate %)

[0129]

[0130] As can be seen from Table 2, the traditional solution is vulnerable to analysis and cracking due to the fixed watermark content. Even for dynamic watermarks, since the watermark content is not separated from the video, there is still a risk of being cracked. In contrast, the solution proposed in this application significantly enhances security through dynamic watermark keys and pipeline-level embedding.

[0131] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and the descriptions in the specification are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and all such changes and improvements fall within the scope of the present invention as claimed. The scope of the present invention claimed is defined by the appended claims and their equivalents.

Claims

1. A method for detecting content tampering of an electronic screen based on dynamic watermarking, characterized in that, It is implemented by an electronic screen content tampering detection system based on dynamic watermarking. The detection system includes a cloud server, a display terminal electronic screen, and a security monitoring terminal. Multiple display terminal electronic screens are connected to the cloud server to obtain video content to be played. Each display terminal electronic screen is connected to a security monitoring terminal, and all security monitoring terminals are connected to the cloud server through TLS encrypted communication; It includes the following steps: S1. Generate a display terminal key seed, bind the display terminal electronic screen to the security monitoring terminal, and upload the mapping relationship and the key seed; S2. The cloud server generates a frame watermark key packet based on the key seed and the video; S3. The display terminal electronic screen downloads the video to be played and requests the cloud server to obtain the frame watermark key packet corresponding to the video through encrypted communication; S4. The display terminal electronic screen calculates the dynamic offset, and the vertex shader superimposes the dynamic offset and performs image rendering to output the watermarked image; S5. Collect the current watermarked frame image output by the terminal electronic screen, reconstruct the original image through inverse transformation based on the frame watermark key, and calculate the actual offset through the phase correlation method; S6. Calculate the residual metric using the actual offset and the theoretical offset. When three consecutive frames of images satisfy that the residual is greater than the preset value, shut down the display terminal electronic screen and report it to the cloud server.

2. The method for detecting electronic screen content tampering based on dynamic watermark according to claim 1, wherein, The S1 includes the following steps: S11. The screen driver chip with a unique serial number of the display terminal electronic screen integrates a physically unclonable function to generate an anti-physical cloning key seed; S12. Pair and bind the display terminal electronic screen to the security monitoring terminal to establish a mapping relationship; S13. Transmit the mapping relationship, the device serial numbers of the display terminal electronic screen and the security monitoring terminal, and the key seed to the cloud server through TLS encrypted communication.

3. The method for detecting the tampering of the electronic screen content based on the dynamic watermark according to claim 2, wherein, The S2 includes the following steps: S21. Generate a master key corresponding to the display terminal electronic screen for deriving the watermark key for each frame, which is generated by performing a hash calculation on the hardware identifier: Among them, is the master key, is the encryption hash function, is the key seed; S22. Derive an independent frame key for each frame of the video through the master key and the frame number: Among them, is the frame key of the video frame , is the encryption operation based on the counter mode of , is the frame number; S23. Calculate the content hash value for the video frame ​ Combine the hash value with the frame key to generate a watermark key: Among them, is a hash-based message authentication code algorithm; S24. After performing frame-by-frame processing of S21-S23 on the video, obtain the frame watermark key packet of the video, and store it in the cloud server corresponding to the video number and the device serial number of the display terminal electronic screen.

4. The method for detecting electronic screen content tampering based on dynamic watermark according to claim 3, wherein The S4 includes the following steps: S41. Obtain the key seed through the GPU module built in the display terminal electronic screen; S42. Video frame to be played Calculate the hash value of the frame content: Generate the watermark key for the frame to be played: Among them, is the video frame to be played and is the frame number. S43. Calculate a dynamic offset according to the watermark key of the frame to be played, that is, the offset for dynamically adjusting the vertex coordinates: ​ Among them, is a uniform distribution mapping function, is a normalization coefficient; Meanwhile, obtain the key corresponding to the frame from the frame watermark key packet acquired from the cloud server , and calculate the theoretical offset: Among them, is the offset component of the theoretical offset in the horizontal direction, is the offset component of the theoretical offset in the vertical direction; S44. Through the stage of processing vertex coordinate transformation in the GPU module rendering pipeline, after vertex coordinate transformation, the vertex shader superimposes a dynamic offset ; The transformed vertex set is rasterized into screen pixels by the GPU module, and the watermarked frame image is rendered and output.

5. The method for detecting electronic screen content tampering based on dynamic watermark according to claim 4, characterized in that, The S5 includes the following steps: S51. Use the FPGA module built in the safety monitoring terminal to collect the current watermarked frame image output by the display terminal's electronic screen, denoted as ; S52. Using the theoretical offset , the original image is reconstructed through inverse transformation to obtain the theoretically watermark-free image: Among them, is the inverse transformation function; Represents the image to be inverse-transformed, and are the coordinates of the image pixels, is the inverse transformation offset, is the offset components in the horizontal and vertical directions, and the sign represents the translation transformation operation in image processing, represents the inverse translation transformation of the coordinates, represents the original coordinates after applying the inverse translation operation; S53. Calculate the actual offset through the phase correlation algorithm: Among them, represents the two-dimensional discrete Fourier transform, represents the two-dimensional inverse Fourier transform, is the complex conjugate operator, represents the peak value of the Fourier transform, and the corresponding position is the actual offset between the two images, is the offset component of the actual offset in the horizontal direction, is the offset component of the actual offset in the vertical direction.

6. The method for detecting electronic screen content tampering based on dynamic watermark according to claim 5, characterized in that The S6 includes the following steps: Calculate the residual metric using the actual offset and the theoretical offset: When three consecutive frames of images satisfy then is a preset value, it is determined that the content currently played on the electronic screen has been tampered with, a message indicating that there is illegal playback on the display terminal's electronic screen is sent to the cloud server, and the display terminal's electronic screen is shut down.

Images