A time authentication-based video stream block time evidence watermarking method

Abstract

The application relates to the technical field of digital video processing, and discloses a video stream block time evidence watermarking method based on time authentication, which comprises the following steps: receiving a continuous video code stream and dividing the video code stream into multiple continuous video blocks, extracting video coding layer data of the first video block, performing a hash operation to generate a reference hash value; sending the reference hash value to a time watermarking device, performing a digital signature operation on the reference hash value in combination with an absolute timestamp and a signature key to generate a digital signature; intercepting a bit segment from the digital signature as a characteristic sequence to form cascade digest data; and modulating preset macro block syntax elements of the first frame of the next video block by using the cascade digest data, and outputting reconstructed video coding layer data. The method effectively reduces processor occupation and memory consumption of the system, enables video content to meet real-time transmission requirements in a monitoring scene while guaranteeing time sequence and data anti-fake capability.

Description

Technical Field

[0001] This invention relates to the field of digital video processing technology, specifically to a time-based authentication method for video stream block time-based watermarking. Background Technology

[0002] Video stream digital watermarking technology is an important application of digital watermarking in scenarios involving continuous, high-volume video data. In business scenarios such as judicial evidence preservation and security monitoring, the system must achieve real-time embedding and detection of watermark features without interrupting the video stream or significantly increasing processing latency, and the watermark must be robust against conventional video processing operations such as video compression and format transcoding.

[0003] To address the characteristics of video streams, existing technologies typically employ robust watermarking methods based on the spatiotemporal domain or deep learning. Spatiotemporal watermarking usually operates in the decoded pixel domain, such as superimposing periodic signals on the luminance or chrominance components within a specific time period, or selecting textured regions in the video frame for feature embedding. Deep learning methods utilize neural network models to learn the most ideal embedding position and intensity in the video. However, these methods usually require full pixel-level decoding and re-encoding of the original video stream when performing feature embedding. This approach not only incurs significant computational overhead but also causes severe transmission latency, making it difficult to meet the evidence storage requirements of high-concurrency, real-time video streams.

[0004] To address the computational latency issue, video watermarking techniques based on the compression domain have also been developed in this field. This technique directly embeds fine-tuning of encoding parameters during the video encoding process (such as H.264 / AVC and H.265 / HEVC standards). This method has low computational overhead and is easily integrated into existing encoders.

[0005] However, in practical scenarios involving the preservation of time-based evidence in continuous video streams, the aforementioned existing technologies still have significant shortcomings. On the one hand, traditional compressed field watermarks are typically only used to embed static copyright identifiers, lacking a cryptographic mechanism to bind physical time to the continuous content of the video, and thus failing to provide a robust, tamper-proof time proof. On the other hand, some systems attempt to directly attach timestamps and hash signatures to the application layer of the video stream for evidence preservation, but this purely application-layer external data is easily stripped away by attackers using conventional video format conversion, transcoding, or filtering tools. Once the application-layer attached data is maliciously discarded, the evidence chain for the video will be completely broken, preventing the system from issuing a legally valid authentication conclusion for the examined video. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a time-based authentication method for video stream block time-based evidence watermarking. This method solves the problem that existing video stream evidence watermarking technologies struggle to construct evidence chains that achieve synchronous and rigorous cryptographic time authentication effectiveness and resistance to malicious full data stripping while ensuring low computational latency.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a video stream block time-based authentication watermarking method, comprising the following steps:

[0008] Receive a continuous video stream and divide it into multiple consecutive video blocks. Extract the video coding layer data of the first video block and perform a hash operation to generate a base hash value.

[0009] The baseline hash value is sent to a time watermarking device, and a digital signature operation is performed on the baseline hash value in combination with the absolute timestamp and the signature key to generate a digital signature.

[0010] Bit segments are extracted from the digital signature as feature sequences to form concatenated digest data;

[0011] The concatenated digest data is used to modulate the preset macroblock syntax elements of the first frame of the next video block, and the reconstructed video coding layer data is output.

[0012] The baseline hash value, the absolute timestamp, and the digital signature are encapsulated into a supplemental enhancement information data packet and inserted before the reconstructed video coding layer data;

[0013] An incremental hash operation is performed on the current video block containing the supplementary enhancement information data packet and the reconstructed video coding layer data to generate an incremental hash value. A new digital signature is requested, and the modulation and insertion operations are performed cyclically to construct a continuous temporal hash chain covering the entire video stream.

[0014] Preferably, before receiving the continuous video stream and dividing it into multiple consecutive video blocks, the method further includes:

[0015] Obtain a time synchronization reference through a network time protocol or a precision time protocol, and adjust the local clock of the hardware timekeeping module.

[0016] Read the basic master key in the hardware cryptographic card and execute the key derivation algorithm based on hash message authentication code to generate the signature key for the current video stream session;

[0017] A video decoding buffer pool with a circular queue structure is established in physical memory, and a context-based adaptive binary arithmetic code parsing library is loaded to obtain control permissions for the underlying elements of the video compression domain.

[0018] Preferably, the step of receiving a continuous video stream and dividing it into multiple consecutive video blocks, extracting the video coding layer data of the first video block, and performing a hash operation to generate a base hash value includes:

[0019] The number of video frames is determined by multiplying the inherent frame rate of the video bitstream by a preset duration, and the number of video frames is used as the physical boundary condition for dividing the video blocks.

[0020] The image order count is extracted by parsing the title information of each network abstraction layer unit in the first video block, and all video frames in the first video block are rearranged according to the decoding order based on the value of the image order count.

[0021] Excluding video parameter sets, sequence parameter sets, and image parameter sets, pure video coding layer data is extracted. The video coding layer data and the feature metadata of the first video block are then concatenated in binary and input into a cryptographically secure hash function to output the baseline hash value.

[0022] Preferably, the step of performing a digital signature operation on the base hash value by combining the absolute timestamp and the signature key to generate a digital signature, and extracting bit segments from the digital signature as feature sequences to construct concatenated digest data, includes:

[0023] The absolute timestamp, the base hash value, and the unique physical identifier of the time watermarking device are concatenated into a binary sequence.

[0024] The digital signature is generated by executing an elliptic curve cryptography-based digital signature algorithm on the concatenated binary sequence using the signature key.

[0025] A hash operation is performed on the digital signature structure, and the last valid bits of the output hash value binary sequence are fixedly extracted as the feature sequence.

[0026] Preferably, the step of modulating the preset macroblock syntax elements of the first frame of the next video block using the concatenated digest data to output the reconstructed video coding layer data includes:

[0027] Perform local entropy decoding on the data payload of the image macroblock at a preset position in the first frame of the next video block to restore the preset macroblock syntax elements containing quantization parameters or motion vectors;

[0028] The least significant bits of consecutive preset macroblock syntax elements are replaced sequentially with the binary bits of the concatenated digest data.

[0029] A local entropy re-encoding operation is performed on the preset macroblock syntax element after replacing the least significant bit to generate the reconstructed video coding layer data.

[0030] Preferably, the step of encapsulating the base hash value, the absolute timestamp, and the digital signature into a supplementary enhanced information data packet includes:

[0031] The baseline hash value, the absolute timestamp, and the digital signature are encapsulated into a network abstraction layer unit of unregistered user data type, and the network abstraction layer unit type value is set to a set identifier to construct the supplementary enhanced information data packet.

[0032] Preferably, the step of performing an incremental hash operation on the current video block containing the supplementary enhancement information data packet and the reconstructed video coding layer data to generate an incremental hash value includes:

[0033] A cryptographic hash context state machine that maintains its active state in memory;

[0034] The hash update function is called to sequentially input the supplementary enhancement information data packet and the reconstructed video coding layer data into the cryptographic hash context state machine to trigger iterative update of the internal parameter array;

[0035] After the data block for the current video block has been input, the hash termination function is called to output the incremental hash value.

[0036] Preferably, during the cyclic execution of the modulation and insertion operations, the method further includes:

[0037] When the end signal of the video stream is captured, the current video block is defined as the final block;

[0038] Extract the display timestamp of the last frame of the final block, and reconstruct the feature metadata of the final block based on the actual number of valid video frames captured in the final block;

[0039] The reconstructed feature metadata is used to participate in the incremental hash value calculation of the final block, and the end status flag and the total playback duration parameter of the current video stream are written into the corresponding generated supplementary enhancement information data packet.

[0040] Preferably, after constructing a continuous temporal hash chain covering the entire video stream, the method further includes a step of tracing and verifying the continuous temporal hash chain:

[0041] Extract application layer supplementary enhancement information data packets from each video block of the video stream to be verified, retrieve the public key to perform asymmetric cryptographic verification on the encapsulated digital signature, and recalculate the actual hash value and compare it bit by bit with the encapsulated historical hash record;

[0042] When the application layer supplementary enhancement information data packet is detected to be continuously missing, the macroblock syntax elements of the underlying video coding layer of the video stream to be verified are extracted, and the least significant bit is read and concatenated to restore the feature sequence to be detected.

[0043] Calculate the Shannon information entropy of the feature sequence to be detected. When the Shannon information entropy is greater than a preset judgment threshold, it is determined that the application layer supplementary enhancement information data packet of the video stream to be verified has been maliciously stripped.

[0044] A video stream block time-based authentication watermarking system, and a video stream block time-based authentication watermarking method according to any one of claims 1-9, characterized in that it includes:

[0045] The video acquisition device module is used to acquire raw video data and perform encoding, outputting a continuous video bitstream containing network abstraction layer units;

[0046] The video processing engine module is used to receive the continuous video bitstream and divide it into video blocks; extract the video coding layer data of the first video block and perform hash operation to generate a base hash value; use concatenated digest data to modulate the preset macroblock syntax elements of the first frame of the next video block to output the reconstructed video coding layer data; encapsulate the base hash value, absolute timestamp, and digital signature into a supplementary enhancement information data packet and insert it before the reconstructed video coding layer data; and perform incremental hash operation on the current video block to construct a continuous temporal hash chain.

[0047] The time source system module is used to provide a time synchronization reference via network time protocol or precision time protocol;

[0048] The time watermarking device module includes a cryptographic card and a timekeeping module, which is used to adjust the local clock according to the time synchronization benchmark, receive the benchmark hash value or incremental hash value, and perform digital signature calculations to generate the digital signature by combining the absolute timestamp and the signature key.

[0049] The traceability verification system module is used to extract application layer supplementary enhancement information data packets of each video block of the video stream to be verified and perform explicit verification. When the application layer supplementary enhancement information data packets are continuously missing, the module extracts the macroblock syntax elements of the underlying video coding layer of the video stream to be verified, splices them to restore the feature sequence to be detected, calculates the Shannon information entropy, and performs implicit anti-stripping judgment.

[0050] This invention provides a time-based authentication method for video stream block time-based watermarking. It has the following beneficial effects:

[0051] 1. This invention reduces the processing latency of video data notarization by constructing a continuous-time hash chain at the application layer. By dividing the continuous video stream into discrete blocks according to a preset duration, only the pure video coding layer data after excluding parameter sets is extracted for the baseline hash calculation. A digital signature generated by combining a high-precision absolute timestamp is encapsulated as a supplementary enhanced information data packet and inserted before the video coding layer data of the next block. Since the incremental hash calculation only relies on the newly input data block, it eliminates the need for pixel-level full decoding and recoding of the complete historical video stream, effectively reducing the system's processor usage and memory overhead. This allows the video content to meet the real-time transmission requirements of monitoring scenarios while maintaining chronological order and data anti-counterfeiting capabilities.

[0052] 2. To address the destructive behavior of malicious attackers who discard supplementary enhancement information data packets entirely through format conversion or filtering tools, this invention extracts a pseudo-random bitstream as a feature sequence from the application-layer digital signature. It then calls a context-based adaptive binary arithmetic coding parsing library to replace this feature sequence bit by bit with the least significant bit of the quantization parameters or motion vector in the first frame of the next block. This operation directly writes the feature sequence into the macroblock syntax elements of the underlying video coding structure. Even when the supplementary enhancement information data packet is physically broken and lost, the system can still extract the feature sequence from the macroblock syntax elements for anti-counterfeiting determination, thus overcoming the technical deficiency of traditional external watermarks in terms of poor anti-dropout capability.

[0053] 3. In the post-event evidence collection stage, the verification system of this invention can not only verify the public key signature by extracting supplementary enhancement information and comparing historical hash records bit by bit to achieve basic verification of video file integrity; but also, when supplementary enhancement information is continuously missing, it can restore the least significant bit of the underlying macroblock syntax element through local entropy decoding and calculate its Shannon information entropy. When the calculated information entropy is greater than the judgment threshold, the system can directly determine that the supplementary enhancement information data packet of the video stream has been maliciously stripped, providing a technical basis at the underlying parameter level for video anti-tampering analysis. Attached Figure Description

[0054] Figure 1 This is a schematic diagram of the system module connections of the present invention;

[0055] Figure 2 This is a schematic diagram of the method steps of the present invention. Detailed Implementation

[0056] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0057] See attached document Figure 2 , Figure 2 This is a flowchart of a video stream block-based watermarking method based on time authentication and encoding parameter concatenation according to an embodiment of the present invention. The present invention provides a video stream block-based watermarking method based on time authentication and encoding parameter concatenation, comprising the following steps:

[0058] S10, System initialization and time base synchronization. The video processing engine and time watermarking device start up and establish a communication connection. The time watermarking device obtains the time synchronization base from the time source system and generates a signature key using its internal cryptographic card. The video processing engine establishes a video decoding buffer pool and loads a context-based adaptive binary arithmetic encoding parsing library;

[0059] Specifically, S10 includes the following sub-steps:

[0060] S11, the video processing engine and the time watermarking device start up and establish a communication connection. The video processing engine initiates a Transmission Control Protocol (TCP) connection request to the time watermarking device through the system's network interface layer. After receiving the request, the time watermarking device and the two parties establish a Transport Layer Security (TLS) encrypted channel. The specific feature of the TLS is TLS 1.3. During the channel establishment phase, the time watermarking device sends registration information containing the device's hardware fingerprint and public key certificate to the video processing engine. The video processing engine verifies the validity of the obtained public key certificate to confirm the identity of the time watermarking device.

[0061] For the underlying logic of the transport layer security protocol handshake mechanism and certificate validity verification, those skilled in the art can use the standard asymmetric encryption verification process for configuration. The protocol interaction process is a well-known technology in this field and will not be described in detail here.

[0062] S12, the time watermarking device obtains a time synchronization reference from the time source system and generates a signature key using an internal cryptographic card. The time watermarking device then sends a time synchronization message to the external time source system. This synchronization operation uses either a network time protocol or a precise time protocol.

[0063] The system has strict threshold limits for the accuracy of time synchronization:

[0064] In Network Time Protocol (NTP) mode, the system requires a synchronization time error of less than 10 milliseconds;

[0065] In the precise time protocol mode, the system requires a synchronization time error of less than 1 microsecond.

[0066] The time watermarking device receives the response message from the time source system, calculates the transmission delay of the current network link, and adjusts the local clock of its internal hardware timekeeping module accordingly. The hardware timekeeping module uses a temperature-controlled crystal oscillator or a rubidium atomic clock to maintain high-precision timekeeping even when the external network connection is lost. The time watermarking device locks the calibrated local clock state as the basis for subsequently generating absolute timestamps.

[0067] The time-watermarking device wakes up the internal hardware cryptographic card and reads the basic master key from the secure storage area of ​​the hardware cryptographic card.

[0068] The logic unit inside the hardware cryptographic card executes a key derivation algorithm based on hash message authentication codes to generate a dedicated signature key for signing the current video stream session. The principle behind this key derivation algorithm is that by introducing trigger data that dynamically changes with each communication session, the fixed base master key is converted into a temporary key valid for only one session.

[0069] This mechanism limits the lifespan of a single key, thereby reducing the risk of brute-force attacks. The key derivation process is as follows:

[0070]

[0071] In the formula, This represents a dedicated signature key derived for the current video stream session, with a fixed length of 256 bits; This represents a cryptographic operation function based on the hash message authentication code, which calls SHA-256 as the secure hash algorithm at the underlying level. This refers to the basic master key pre-written within the hardware cryptographic card; This represents the trigger seed data for the current communication session. The trigger seed data is composed of the current system status parameter information and the unique physical identifier of the time watermarking device.

[0072] S13, the video processing engine establishes a video decoding buffer pool and loads a context-based adaptive binary arithmetic encoding parsing library.

[0073] The video processing engine allocates a separate, contiguous storage area in the host's physical memory. It then uses this contiguous storage area to construct a video decoding buffer pool with a circular queue structure. The specific size of the video decoding buffer pool is determined by multiplying the video stream's frame rate by a preset buffer duration.

[0074] For example, when the system receives a video stream with a frame rate of 25 frames per second and the preset buffer duration is configured to 3 seconds, the video processing engine allocates memory space to the video decoding buffer pool that can hold at least 75 frames of complete video stream data, thereby avoiding overflow or packet loss of the video stream during subsequent decoding and reassembly.

[0075] The video processing engine starts the underlying video encoding and decoding framework, mounting dynamic link library files supporting the H.265 standard to the main process memory. The video processing engine initializes the context-based adaptive binary arithmetic coding parsing library. Context-based adaptive binary arithmetic coding is a lossless entropy coding technique in the H.265 video standard used for compressing macroblock data.

[0076] By loading this parsing library, which provides an application programming interface for reading and overwriting the underlying compression domain of video, the video processing engine has the control to locally extract and reconstruct the syntax elements at the video macroblock level.

[0077] S20, Video Stream Reception and Baseline Block Hash Calculation. The video processing engine receives the continuous video stream output by the video acquisition device. The video processing engine divides the video stream into continuous video blocks according to a preset duration. The preset duration is set to a range of 1 to 5 seconds, and the corresponding number of video frames is used as the block division boundary.

[0078] Because bidirectional prediction frames exist in video encoding, the display order of video frames is inconsistent with the decoding order. For the first video block, the video processing engine parses the header information of the network abstraction layer unit and reassembles the video frames according to the decoding order.

[0079] The video processing engine extracts the video coding layer data of the first video block, excludes the video parameter set, sequence parameter set and image parameter set, and performs a secure hash operation to generate a base hash value. The secure hash operation uses the SHA-256 algorithm or the SM3 algorithm.

[0080] Specifically, step S20 includes the following sub-steps:

[0081] S21, the video processing engine receives the continuous video stream output by the video capture device. The video processing engine divides the continuous video stream into multiple consecutive video blocks according to a preset duration. The preset duration is set to a range of 1 second to 5 seconds.

[0082] The range of this preset duration is determined based on the balance between system computing power overhead and anti-tampering granularity:

[0083] If the preset duration is less than 1 second, the video processing engine will need to frequently perform hash calculations and signature operations, which will lead to overload of system computing power.

[0084] If the preset duration is greater than 5 seconds, severe packet loss during network transmission can cause the hash chain of a large segment of video to break, reducing the availability of evidence.

[0085] The video processing engine calculates a specific number of video frames based on the product of the inherent frame rate of the video bitstream and a preset duration, and uses this number of video frames as the physical boundary condition for dividing the video blocks. In this step, the lower-level feature of the block division rule is implemented by treating a predetermined number of consecutive video frames as an independent data processing unit.

[0086] In step S22, for the first video block after partitioning, the video processing engine performs a decoding order reassembly operation. Due to the bidirectional prediction frames in the H.265 video coding standard, the display timestamps and decoding timestamps of video frames are inconsistent, resulting in a difference between the display order and decoding order. To ensure that the calculated hash value remains strictly consistent with the physical order of the bitstream during network transmission and storage, and to avoid the significant computational cost of re-performing complete pixel-level decoding in subsequent verification stages, the system performs a hash solidification operation based on the physical receiving order of the bitstream.

[0087] The video processing engine parses the title sequence information of each network abstraction layer unit within the first video block. The engine then extracts the image order count from the title sequence information.

[0088] The video processing engine rearranges all video frames within the first video block strictly according to the decoding order based on the numerical value of the image order count. For the structural parsing of the network abstraction layer unit header information and the extraction of image order counts, those skilled in the art can refer to existing video compression coding standards for configuration; the parsing process is well-known in the field and will not be elaborated upon here.

[0089] S23, the video processing engine extracts video coding layer data and generates a baseline hash value for the first rearranged video block. During the classification process of network abstraction layer units, the video processing engine excludes video parameter sets, sequence parameter sets, image parameter sets, and any pre-existing supplementary enhancement information.

[0090] The video processing engine retains only the video coding layer data that carries the actual pixel features of the image to ensure that the extracted data is not interfered with by external additional information. The video processing engine then performs binary concatenation and splicing of the extracted video coding layer data with the metadata of the first video block.

[0091] To ensure consistency between the data concatenation at both ends during signature verification, the feature metadata must be converted to a fixed-width binary structure before concatenation. For example, the actual number of frames contained must be converted to a 32-bit unsigned integer format. The video processing engine then calls a cryptographic hash function to perform a secure hash operation on the concatenated data stream to generate a base hash value.

[0092] The specific lower-level feature implementation of this secure hash operation uses the SHA-256 algorithm or the SM3 algorithm. The calculation process of the base hash value is as follows:

[0093]

[0094] In the formula, This represents the base hash value calculated from the first video block, and its output data length is fixed at 256 bits. This indicates the specified cryptographically secure hash operation function; This represents the clean video coding layer data stream extracted after filtering within the first video block; Operators for concatenating binary data; This represents the feature metadata of the first video block. Specifically, this feature metadata includes the actual number of frames contained in the current video block, the video resolution, and the data stream encoding format identifier. All metadata features are uniformly converted into binary values ​​with a preset bit width before data splicing operations are performed.

[0095] S30, Timestamp Signature Generation and Cascaded Digest Extraction. The video processing engine sends the base hash value to the time watermarking device. The time watermarking device combines the absolute timestamp provided by its internal timekeeping module with the signature key to calculate the base hash value and generate a digital signature.

[0096] The video processing engine receives digital signatures and extracts a fixed-length feature sequence L from the signatures as concatenated digest data. The feature sequence length L ranges from 8 bits to 32 bits, ensuring sufficient feature information while maintaining the stability of the video coding structure.

[0097] Specifically, step S30 includes the following sub-steps:

[0098] S31, the video processing engine sends the calculated base hash value to the time watermarking device via a transport layer security protocol encrypted channel. The time watermarking device receives the base hash value and obtains the current absolute timestamp through its internal hardware timekeeping module.

[0099] An absolute timestamp is used to accurately record the physical time when the base hash value was generated. The time-watermarking device concatenates the absolute timestamp, the base hash value, and a unique physical identifier embedded in the time-watermarking device itself into a binary sequence. The unique physical identifier is used to provide proof of the hardware device's identity in subsequent traceability stages.

[0100] S32, the time-watermarking device calls its internal hardware cryptographic card and uses the dedicated signature key derived during system initialization to perform a digital signature operation on the concatenated binary sequence, generating a cryptographic digital signature. The specific implementation of this digital signature operation employs either an elliptic curve cryptography-based digital signature algorithm or the SM2 national cryptographic algorithm. The principle behind this digital signature is to use the private key in an asymmetric cryptographic system to perform mathematical transformations on the business data, enabling the generated data verification identifier to be bound to the original data.

[0101] This binding relationship ensures that the evidence data containing hash values ​​and timestamps cannot be tampered with or forged by third parties, and supports any party that obtains the corresponding public key to verify the source and integrity of the data.

[0102] The principle behind using digital signature algorithms is to leverage the private key characteristics of asymmetric cryptography to ensure that the evidence data containing hash values ​​and timestamps cannot be forged, while allowing third parties to use publicly available public key certificates for legitimacy verification.

[0103] For asymmetric encryption / decryption and digital signature operations based on elliptic curve cryptography, those skilled in the art can configure and implement them according to existing cryptographic standards and specifications. The underlying mathematical operations, such as elliptic curve dot multiplication, are well-known technologies in the field and will not be elaborated upon here. The digital signature calculation process is as follows:

[0104]

[0105] In the formula, This indicates the digital signature generated by the time-watermarking device for the first video block; This indicates the specified digital signature generation function based on the elliptic curve algorithm. This represents the base hash value of the first video block submitted by the video processing engine; Operators for concatenating binary data; This represents the absolute timestamp currently output by the hardware timekeeping module; A unique physical identifier code burned into the time watermarking device; This indicates that the time-watermarking device generates a dedicated signature key for the current communication session.

[0106] S33, the time-watermarking device encapsulates the generated digital signature into a response data packet and sends it to the video processing engine. The video processing engine receives the digital signature and performs a bit truncation operation to obtain concatenated digest data.

[0107] Considering that digital signatures have structured encapsulated data headers during transmission, the video processing engine needs to eliminate interference from non-random data within the encapsulation structure. The video processing engine, according to preset truncation rules, extracts the numerical portion of the digital signature that exhibits pseudo-random characteristics and extracts a fixed-length bit segment of length L as a feature sequence.

[0108] The video processing engine extracts a fixed-length bit segment of length L from the binary sequence of the digital signature as a feature sequence according to preset truncation rules.

[0109] The feature sequences constitute the concatenated summary data used to correlate the underlying grammatical parameters of the video. The fixed length L is set to range from 8 bits to 32 bits. This length range is determined based on the robustness of the video compression coding structure: if the feature sequence length is less than 8 bits, the extracted data lacks sufficient pseudo-random statistical properties, making it difficult to form a convincing basis for anti-stripping judgment during the verification stage;

[0110] If the feature sequence length is greater than 32 bits, forcibly writing too many bits of data into the underlying macroblock of the video will destroy the entropy coding efficiency of context-based adaptive binary arithmetic coding, causing decoder errors or severe mosaic distortion in the video.

[0111] The calculations for the extraction process are as follows:

[0112]

[0113] In the formula, This represents the concatenated digest data extracted from the digital signature. This indicates the set bit truncation function. The specific implementation of this function is to perform a hash operation on the input digital signature structure and then truncate the last L valid bits of the output hash value binary sequence to eliminate the interference of fixed identifier bits caused by the signature structure encapsulation. This indicates the preset extraction length, and the variable is limited to an integer value between 8 and 32.

[0114] S40, Modulation of coding parameters and insertion of supplementary enhancement information: In order to establish the association between the underlying video data and the application layer data signature, for the next adjacent video block, the video processing engine uses the concatenated digest data to modulate the preset macroblock syntax elements of the first frame of the next video block, and completes the reconstruction of the underlying video coding layer data.

[0115] Preset macroblock syntax elements include quantization parameters or motion vectors. The video processing engine performs modulation by sequentially replacing the least significant bits of the preset macroblock syntax elements with the binary bits of the concatenated summary data.

[0116] Simultaneously, the video processing engine encapsulates the baseline hash value, absolute timestamp, and digital signature into a supplementary enhancement information data packet. The video processing engine inserts this supplementary enhancement information data packet before the reconstructed video coding layer data.

[0117] Specifically, step S40 includes the following sub-steps:

[0118] S41, the video processing engine locates an image macroblock at a preset position for the first frame of the next adjacent video block. The specific positioning rule for this preset position is to select L consecutive macroblocks after the starting offset of the first frame's image patch data region, ensuring accurate location and extraction of hidden features in subsequent verification stages. The specific lower-level features of this first frame are implemented as real-time decoding refresh frames or intra-frame prediction frames in the video sequence.

[0119] Since the H.265 video stream undergoes lossless arithmetic compression, directly modifying the binary bitstream would cause the subsequent decoding process to crash. The video processing engine calls a context-based adaptive binary arithmetic coding parsing library to perform local entropy decoding on the data payload of the located image macroblock.

[0120] The video processing engine uses local entropy decoding to restore the compressed binary bitstream into underlying data that can be independently parsed and modified. The restored data is the preset macroblock syntax element, which specifically includes quantization parameters or motion vectors. The preset macroblock syntax elements, quantization parameters, and motion vectors described in this specification constitute the specific technical terminology mapping of the underlying coding parameters in the claims.

[0121] For the underlying decoding operation logic and probability model state machine update principle of context-based adaptive binary arithmetic coding, those skilled in the art can refer to existing high-efficiency video coding standards and specifications for configuration and implementation. Its arithmetic decoding process is a well-known technology in this field and will not be described in detail here.

[0122] In step S42, the video processing engine obtains the extracted preset macroblock syntax elements and performs bit-level modulation operations on them using concatenated digest data. The principle of this modulation operation is to covertly embed externally obtained signature feature parameters into the underlying coding structure of the video frame.

[0123] Since the quantization parameters determine the compression step size of the image, and the motion vector determines the positional offset of the inter-frame prediction, when these low-level values ​​undergo a single-bit flip in the least significant bit, it only causes a tiny physical perturbation in the pixel values ​​within the corresponding macroblock. This tiny perturbation is far below the visual perception threshold of the human eye, thus enabling the writing of tamper-proof features without destroying the main content of the video.

[0124] In practice, the video processing engine extracts the binary form of macroblock syntax elements and replaces their least significant bit sequence with the feature sequence of the concatenated digest data in sequence. To ensure precise matching of the written data capacity, the total number of extracted and modified macroblock syntax elements is consistent with the bit length L of the concatenated digest data, that is, each macroblock syntax element carries 1 bit of information from the concatenated digest data.

[0125] The video processing engine overwrites the last bits of the concatenated digest data bit by bit onto the last bits of the syntax elements of consecutive macroblocks. After modulation, the video processing engine performs local entropy re-encoding on the modified macroblock syntax elements using context-based adaptive binary arithmetic coding, restoring the syntax elements to a compressed bitstream structure conforming to the H.265 standard, thus outputting the reconstructed video coding layer data. The calculation process for the underlying parameter modulation is as follows:

[0126]

[0127] In the formula, This represents the sequence of preset macroblock syntax elements modulated by the substitution operation; This indicates the least significant bit substitution function. This represents the original preset macroblock syntax element sequence reconstructed from the first frame of the next video block using local entropy decoding, which contains L syntax elements; This indicates that in step S30, the generated concatenated digest data of length L bits is extracted.

[0128] The calculation process for reconstructing the video coding layer data is as follows:

[0129]

[0130] In the formula, This represents the reconstructed video coding layer data output. Represents a context-based adaptive binary arithmetic encoding reprogramming function; This represents the modulated sequence of preset macroblock syntax elements.

[0131] S43, after the video processing engine completes the reconstruction of the underlying video encoding layer data, it synchronously performs the encapsulation of evidence storage data at the application layer.

[0132] The video processing engine concatenates the base hash value generated in step S20, the absolute timestamp obtained in step S30, and the digital signature according to the set data format.

[0133] The video processing engine encapsulates the concatenated data structure into a supplementary enhancement information data packet according to video compression standards. The specific implementation format of this supplementary enhancement information data packet is a Network Abstraction Layer (NET) unit of unregistered user data type, corresponding to NET type value 39 in the H.265 standard.

[0134] The video processing engine locates the start position of the first frame of the next video block to be processed in the video stream. The engine maintains the original display order timestamp, reference frame list structure, and time stamp parameters of the current frame, and directly inserts the constructed supplementary enhancement information data packet into the video stream, physically positioned before the reconstructed video coding layer data.

[0135] This operation allows the historical evidence records of the previous block to be physically attached to the beginning of the current block's bitstream, providing a physical data carrier for the system to build a continuous and tamper-proof time hash chain.

[0136] S50, video block hash chain iterative construction: The video processing engine performs incremental hash operation on the complete video block containing supplementary enhancement information data packets and reconstructed video coding layer data to generate the incremental hash value of the current video block.

[0137] The video processing engine sends the incremental hash value to the time watermarking device to request a new digital signature, and repeatedly executes the steps of encoding parameter modulation and supplementary enhancement information insertion to build a continuous time hash chain covering the entire video stream;

[0138] Specifically, step S50 includes the following sub-steps:

[0139] S51, the video processing engine performs a hash operation on the complete video block containing supplementary enhancement information data packets and reconstructed video coding layer data to generate an incremental hash value for the current video block. This complete video block is the bitstream format after application layer data insertion and low-level parameter modulation.

[0140] To control system computing power overhead and meet the computing requirements of real-time video stream processing, the video processing engine uses incremental hashing to calculate hash values.

[0141] The specific principle of incremental hashing lies in the fact that the video processing engine maintains a continuously active cryptographic hash context state machine in memory. In the actual encoding implementation, the video processing engine does not recalculate the complete hash of the previous block, but instead calls the hash update function (such as hash_ctx.update()) to sequentially input the newly added supplementary enhancement information data packet block and the reconstructed video coding layer data block into this state machine.

[0142] Input operations trigger iterative updates to the state machine's internal parameter array. After the current block of data has been input, a hash termination function (e.g., hash_ctx.digest()) is called to output the current incremental hash result. This data update mechanism significantly reduces processor usage while ensuring the rigor of the hash calculation result. The calculation process for the incremental hash value is as follows:

[0143]

[0144] In the formula, Indicates the current number The incremental hash value calculated for each video block, where It is a sequence of consecutive positive integers greater than 1, used to define the physical temporal position of the currently processed block in the entire hash chain; This indicates the cryptographically secure hash function used; This indicates that the element generated and inserted in the previous step is the current element. The supplemental enhancement information data packet in the video block, which carries the first... -1 video block's certificate signature; Indicates the current number The reconstructed video coding layer data output after local entropy recoding of each video block; Indicates the current number The feature metadata of each video block has the same data field structure as the feature metadata of the first video block, and is used to dynamically record the actual number of video frames contained in the current block. The concatenation operator for binary data.

[0145] In step S52, the video processing engine sends the generated incremental hash value to the time watermarking device to request a new digital signature. The time watermarking device receives the incremental hash value, obtains the absolute timestamp of the current physical time, and calls the hardware cryptographic card to perform asymmetric encryption operations using the signature key to generate a digital signature for the current video block.

[0146] The time watermarking device returns the digital signature as response data to the video processing engine. After receiving the digital signature, the video processing engine extracts a bit feature sequence as new concatenated digest data according to the preset extraction rules, which will be used when the next adjacent video block performs the underlying coding parameter modulation.

[0147] This information flow process establishes the recursive logic within the system. Since the incremental hash value of the current block is calculated based on the supplementary and enhanced information data packet containing the signature of the previous block, and the newly generated digital signature of the current block will be encapsulated and passed to the next block, this makes all discrete video blocks form a cryptographically connected dependency relationship.

[0148] The dependency relationship described in this specification, which involves historical signature data participating in the current hash operation and propagating it backwards, constitutes the specific technical terminology mapping of the continuous temporal hash chain covering the entire video stream in the claims.

[0149] S53, based on the above recursive relationship, the video processing engine cyclically executes the encoding parameter modulation and supplementary enhancement information insertion steps during video stream reception until the end-of-video stream signal is captured. The specific lower-level characteristics of the end-of-video stream signal are implemented as a network connection disconnection event at the transmission control protocol layer or a file end identifier defined in the video encapsulation protocol. After capturing the end-of-video stream signal, the video processing engine determines that the currently processed video block is the final block of that video stream segment.

[0150] In actual video stream acquisition, the number of video frames contained in the final block often fails to meet the physical boundary conditions of the preset duration.

[0151] The system classifies this block as a malformed tail block. For this malformed tail block, the video processing engine does not perform automatic empty frame padding. Instead, it extracts the display timestamp of the last frame in the block and reconstructs the block's feature metadata based on the actual number of captured valid video frames. The video processing engine uses this reconstructed feature metadata in the final block's hash value and signature calculation, i.e., by updating the formula... The variable value is matched to the actual physical length of the tail block.

[0152] For the final block, the video processing engine writes an end-state flag in specific reserved bytes of the generated final supplementary enhancement information data packet, and appends a parameter recording the total playback duration of the current video stream from the first block to the final block. The technical purpose of writing the end-state flag is to clearly define the physical termination boundary of this continuous temporal hash chain for the detection system during the subsequent verification phase.

[0153] If an attacker attempts to cover up the occurrence of certain scenes by truncating the end of a video, the verification system can quickly determine that the video file has been maliciously tampered with in terms of its temporal length by detecting the absence of the end status flag.

[0154] After the processing is completely finished, the video processing engine releases the physical memory occupied by the video decoding buffer pool, destroys the temporarily derived special signature key, and closes the encrypted communication channel with the time watermarking device.

[0155] For the underlying system instruction call logic of memory space release, security key destruction and network communication port closure, those skilled in the art can refer to the memory management specifications and network programming interfaces of standard operating systems for implementation. The resource reclamation operation is a well-known technology in this field and will not be described in detail here.

[0156] S60, the source tracing and verification system acquires the video stream to be verified, extracts supplementary enhancement information data packets for each video block, and performs hash value comparison and digital signature verification. When the supplementary enhancement information data packets are missing, the source tracing and verification system extracts the macroblock syntax elements of the underlying video coding layer of the video stream to be verified, and performs video content anti-tampering and supplementary enhancement information anti-stripping determination by restoring the feature sequence.

[0157] Specifically, step S60 includes the following sub-steps:

[0158] S61, the source tracing and verification system acquires the video stream to be verified, extracts the application layer supplementary enhancement information data packets of each video block, and performs explicit verification. The source tracing and verification system reads the network abstraction layer units in the video stream to be verified one by one according to the time sequence.

[0159] The traceability verification system identifies data packets with a network abstraction layer unit type value of 39 and extracts the encapsulated physical timestamp, historical hash record, device physical identifier, and digital signature. The system then retrieves the publicly disclosed time-watermarked device public key from the system initialization phase and performs asymmetric cryptographic verification on the extracted digital signature.

[0160] Meanwhile, the source tracing and verification system recalculates the actual hash value of the corresponding historical video block based on the same hash algorithm as in the aforementioned steps, and compares it bit by bit with the historical hash record parsed from the supplementary and enhanced information data packet.

[0161] In the actual verification logic, the system only considers the block untampered if both signature verification and hash comparison pass. If the digital signature verification passes and the hash comparison is completely consistent, the traceability verification system determines that the video block has not been tampered with after the corresponding physical time point, and the temporal continuity and data integrity of the video stream are established. The calculation logic of explicit verification is as follows:

[0162]

[0163]

[0164] In the formula, This represents a boolean result of digital signature verification, with a value of either true or false. This represents a signature verification function based on a public key cryptography system. This represents the public key of the time-based watermarking device; Indicates from the first The digital signature of the previous block is extracted from the supplementary enhancement information of each video block; This represents the extracted historical hash record of the previous block; This represents the extracted physical timestamp; This indicates the extracted device physical identifier code; The source tracing and verification system indicates that it verifies the source based on the first... The actual hash value obtained by recalculating the data of each video block; This indicates the cryptographically secure hash function used; , and These represent the first and second parts of the video stream to be verified. Supplemental enhancement information data packets, video coding layer data, and feature metadata for each block.

[0165] The conditions for the traceability verification system to determine that the data has not been tampered with are: The result is true, and binary sequence and The binary sequences are strictly equal. For the verification process of digital signatures based on public-key cryptography, those skilled in the art can call existing cryptographic security library functions for processing. The underlying verification logic is well-known in the field and will not be elaborated here.

[0166] S62, the source tracing and verification system monitors the presence of supplementary enhancement information data packets while traversing the video stream to be verified. If an attacker uses filters or video processing tools to perform a full stripping attack on the video file to cover up tampering, all supplementary enhancement information data packets in the video stream will be filtered out and discarded, indicating a physical break in the hash chain.

[0167] The source tracing verification system detects continuous missing supplementary enhancement information data packets, immediately terminates the explicit verification process, and triggers the anti-stripping judgment logic of the underlying video coding layer. This state switching logic, triggered by application layer data loss and induced by low-level parameter parsing, as described in this specification, constitutes the specific technical terminology mapping for performing video content anti-tampering and anti-stripping judgments when supplementary enhancement information data packets are missing, as stated in the claims.

[0168] S63, the source tracing verification system extracts macroblock syntax elements from the underlying video coding layer of the video stream to be verified, and performs implicit anti-stripping judgment by restoring the feature sequence. The source tracing verification system locates the macroblock at a preset position in the first frame of the video block to be verified, calls the context-based adaptive binary arithmetic code parsing library to perform local entropy decoding, and extracts macroblock syntax elements containing quantization parameters or motion vectors.

[0169] The source tracing and verification system reads the least significant bits of the binary sequence of the macroblock's syntax elements, concatenates and combines them sequentially, and reconstructs the feature sequence to be detected. The calculation process for feature sequence reconstruction is as follows:

[0170]

[0171] In the formula, This represents the feature sequence reconstructed from the underlying layer of the video block to be verified, with a data length of [missing information]. Bit; This represents the least significant bit extraction and sequence recombination function; This represents a sequence of macroblock syntax elements partially decoded from the video stream to be verified.

[0172] After obtaining the restored feature sequence, the source tracing and verification system performs statistical characteristic analysis on it. The general principle of feature sequence detection is that: in normal, unmodulated video images, the quantization parameters or motion vectors of adjacent macroblocks have high spatial correlation, and the distribution of 0s and 1s in the least significant bit sequence is not uniform, resulting in low information entropy; while the feature sequence modulated by the system originates from truncated data of cryptographic signatures, and the 0s and 1s in its binary sequence exhibit a uniform pseudo-random distribution, with the information entropy approaching the theoretical maximum value of the binary sequence.

[0173] The source tracing and verification system calculates the Shannon information entropy of the feature sequence and compares it with a preset judgment threshold. The judgment threshold is set to range from 0.9 to 0.95 bits per symbol. If the calculated information entropy is higher than the judgment threshold, it indicates that the parameters of the underlying video were written into a cryptographic digest with pseudo-random characteristics.

[0174] Based on this, the source tracing verification system outputs the final anti-stripping judgment result, confirming that the inspected video source file once carried a legitimate time-authentication watermark, but was subsequently maliciously stripped and destroyed. This provides a basis for determining the video's anti-tampering and downgrade capabilities for judicial evidence collection. The mathematical calculation process of Shannon information entropy can be implemented using conventional information theory statistical algorithms by those skilled in the art; the calculation method is a well-known technique in this field and will not be elaborated upon here.

[0175] See attached document Figure 1 This invention provides a video stream block evidence watermarking system based on time authentication and encoding parameter concatenation, including: a video acquisition device module, a video processing engine module, a time watermarking device module, a time source system module, and a traceability verification system module.

[0176] The video acquisition device module is used to acquire raw video data and perform H.265 standard encoding, outputting a video stream containing network abstraction layer units.

[0177] The video processing engine receives the video stream output by the video acquisition device and performs video stream parsing, network abstraction layer unit classification, decoding order reassembly, video block hash calculation, and supplementary enhancement information insertion operations.

[0178] The video processing engine also performs context-based adaptive binary arithmetic coding local parsing and low-level coding parameter modulation.

[0179] The time watermarking device module includes a cryptographic card and a timekeeping module. The time watermarking device is used to generate absolute timestamps, calculate digital signatures, and construct supplementary and enhanced information payloads. The time watermarking device and the video processing engine interact with each other through an encrypted channel established by a transport layer security protocol.

[0180] The time source system module establishes a connection with the time watermarking device through a network time protocol or a precise time protocol, providing a standard time synchronization reference for the time watermarking device.

[0181] The traceability verification system module communicates with the time watermarking device module to obtain the status report logs of the time watermarking device. During the video forensics phase, the traceability verification system extracts video file data and performs cross-certification operations for video time tracing and integrity.

[0182] Specifically, in terms of the physical and logical connections of the system, the data output end of the video acquisition device module is connected to the data input end of the video processing engine module; the business interaction port of the video processing engine module establishes a bidirectional communication connection with the data interface of the time watermarking device module; the time synchronization output end of the time source system module is connected to the network receiving port of the time watermarking device module; and the management port of the traceability verification system module is connected to the status monitoring port of the time watermarking device module.

[0183] The video acquisition device module is used to acquire raw video data and perform H.265 standard encoding, and outputs a continuous video stream containing network abstraction layer units to the video processing engine module.

[0184] The video processing engine module receives the video stream transmitted from the video acquisition device module. It performs video stream parsing, network abstraction layer unit classification, decoding order reassembly, video block hash calculation, and supplementary enhancement information insertion. The video processing engine module also performs context-based adaptive binary arithmetic coding local parsing and low-level coding parameter modulation.

[0185] The time watermarking device module includes a cryptographic card and a timekeeping module. It receives hash data submitted by the video processing engine module, generates an absolute timestamp, calculates a digital signature, and constructs supplementary and enhanced information payloads. The time watermarking device module and the video processing engine module exchange the aforementioned business data bidirectionally via a transport layer security protocol encrypted channel.

[0186] The time source system module establishes a connection with the time watermarking device module through a network time protocol or a precise time protocol, providing a standard time synchronization reference for the time watermarking device module.

[0187] The traceability verification system module communicates with the time watermarking device module to obtain the status report logs periodically sent by the time watermarking device module. During the video forensics stage, the traceability verification system module extracts the video file data output by the video processing engine module and performs cross-certification operations for video time traceability and integrity.

[0188] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for time-based authentication of video stream block time-based watermarking, characterized in that, Includes the following steps: Receive a continuous video stream and divide it into multiple consecutive video blocks. Extract the video coding layer data of the first video block and perform a hash operation to generate a base hash value. The baseline hash value is sent to a time watermarking device, and a digital signature operation is performed on the baseline hash value in combination with the absolute timestamp and the signature key to generate a digital signature. Bit segments are extracted from the digital signature as feature sequences to form concatenated digest data; The concatenated digest data is used to modulate the preset macroblock syntax elements of the first frame of the next video block, and the reconstructed video coding layer data is output. The baseline hash value, the absolute timestamp, and the digital signature are encapsulated into a supplemental enhancement information data packet and inserted before the reconstructed video coding layer data; An incremental hash operation is performed on the current video block containing the supplementary enhancement information data packet and the reconstructed video coding layer data to generate an incremental hash value. A new digital signature is requested, and the modulation and insertion operations are performed cyclically to construct a continuous temporal hash chain covering the entire video stream.

2. The video stream block time-based authentication watermarking method according to claim 1, characterized in that: Before receiving the continuous video stream and dividing it into multiple consecutive video blocks, the method further includes: Obtain a time synchronization reference through a network time protocol or a precision time protocol, and adjust the local clock of the hardware timekeeping module. Read the basic master key in the hardware cryptographic card and execute the key derivation algorithm based on hash message authentication code to generate the signature key for the current video stream session; A video decoding buffer pool with a circular queue structure is established in physical memory, and a context-based adaptive binary arithmetic code parsing library is loaded to obtain control permissions for the underlying elements of the video compression domain.

3. The video stream block time-based authentication watermarking method according to claim 1, characterized in that: The process of receiving a continuous video stream and dividing it into multiple consecutive video blocks, extracting the video coding layer data of the first video block, and performing a hash operation to generate a base hash value includes: The number of video frames is determined by multiplying the inherent frame rate of the video bitstream by a preset duration, and the number of video frames is used as the physical boundary condition for dividing the video blocks. The image order count is extracted by parsing the title information of each network abstraction layer unit in the first video block, and all video frames in the first video block are rearranged according to the decoding order based on the value of the image order count. Excluding video parameter sets, sequence parameter sets, and image parameter sets, pure video coding layer data is extracted. The video coding layer data and the feature metadata of the first video block are then concatenated in binary and input into a cryptographically secure hash function to output the baseline hash value.

4. The video stream block time-based authentication watermarking method according to claim 1, characterized in that: The process of performing a digital signature operation on the base hash value by combining the absolute timestamp and the signature key to generate a digital signature, and extracting bit segments from the digital signature as feature sequences to construct concatenated digest data, includes: The absolute timestamp, the base hash value, and the unique physical identifier of the time watermarking device are concatenated into a binary sequence. The digital signature is generated by executing an elliptic curve cryptography-based digital signature algorithm on the concatenated binary sequence using the signature key. A hash operation is performed on the digital signature structure, and the last valid bits of the output hash value binary sequence are fixedly extracted as the feature sequence.

5. The video stream block time-based authentication watermarking method according to claim 1, characterized in that: The process of modulating the preset macroblock syntax elements of the first frame of the next video block using the concatenated digest data to output reconstructed video coding layer data includes: Perform local entropy decoding on the data payload of the image macroblock at a preset position in the first frame of the next video block to restore the preset macroblock syntax elements containing quantization parameters or motion vectors; The least significant bits of consecutive preset macroblock syntax elements are replaced sequentially with the binary bits of the concatenated digest data. A local entropy re-encoding operation is performed on the preset macroblock syntax element after replacing the least significant bit to generate the reconstructed video coding layer data.

6. The video stream block time-based authentication watermarking method according to claim 1, characterized in that: The step of encapsulating the base hash value, the absolute timestamp, and the digital signature into a supplementary enhanced information data packet includes: The baseline hash value, the absolute timestamp, and the digital signature are encapsulated into a network abstraction layer unit of unregistered user data type, and the network abstraction layer unit type value is set to a set identifier to construct the supplementary enhanced information data packet.

7. The video stream block time-based authentication watermarking method according to claim 1, characterized in that: The step of performing incremental hash operation on the current video block containing the supplementary enhancement information data packet and the reconstructed video coding layer data to generate an incremental hash value includes: A cryptographic hash context state machine that maintains its active state in memory; The hash update function is called to sequentially input the supplementary enhancement information data packet and the reconstructed video coding layer data into the cryptographic hash context state machine to trigger iterative update of the internal parameter array; After the data block for the current video block has been input, the hash termination function is called to output the incremental hash value.

8. The video stream block time-based authentication watermarking method according to claim 1, characterized in that: The process of repeatedly performing the modulation and insertion operations also includes: When the end signal of the video stream is captured, the current video block is defined as the final block; Extract the display timestamp of the last frame of the final block, and reconstruct the feature metadata of the final block based on the actual number of valid video frames captured in the final block; The reconstructed feature metadata is used to participate in the incremental hash value calculation of the final block, and the end status flag and the total playback duration parameter of the current video stream are written into the corresponding generated supplementary enhancement information data packet.

9. A video stream block time-based authentication watermarking method according to claim 1, characterized in that: After constructing a continuous-time hash chain covering the entire video stream, the process also includes a step of tracing and verifying the continuous-time hash chain: Extract application layer supplementary enhancement information data packets from each video block of the video stream to be verified, retrieve the public key to perform asymmetric cryptographic verification on the encapsulated digital signature, and recalculate the actual hash value and compare it bit by bit with the encapsulated historical hash record; When the application layer supplementary enhancement information data packet is detected to be continuously missing, the macroblock syntax elements of the underlying video coding layer of the video stream to be verified are extracted, and the least significant bit is read and concatenated to restore the feature sequence to be detected. Calculate the Shannon information entropy of the feature sequence to be detected. When the Shannon information entropy is greater than a preset judgment threshold, it is determined that the application layer supplementary enhancement information data packet of the video stream to be verified has been maliciously stripped.

10. A video stream block time-based authentication watermarking system, comprising a video stream block time-based authentication watermarking method according to any one of claims 1-9, characterized in that, include: The video acquisition device module is used to acquire raw video data and perform encoding, outputting a continuous video bitstream containing network abstraction layer units; The video processing engine module is used to receive the continuous video bitstream and divide it into video blocks, extract the video coding layer data of the first video block and perform hash operation to generate a base hash value; The preset macroblock syntax elements of the first frame of the next video block are modulated using concatenated digest data to output reconstructed video coding layer data. The base hash value, absolute timestamp, and digital signature are encapsulated as supplementary enhancement information data packets and inserted before the reconstructed video coding layer data. Incremental hashing is performed on the current video block to construct a continuous temporal hash chain. The time source system module is used to provide a time synchronization reference via network time protocol or precision time protocol; The time watermarking device module includes a cryptographic card and a timekeeping module, which is used to adjust the local clock according to the time synchronization benchmark, receive the benchmark hash value or incremental hash value, and perform digital signature calculations to generate the digital signature by combining the absolute timestamp and the signature key. The traceability verification system module is used to extract application layer supplementary enhancement information data packets of each video block of the video stream to be verified and perform explicit verification. When the application layer supplementary enhancement information data packets are continuously missing, the module extracts the macroblock syntax elements of the underlying video coding layer of the video stream to be verified, splices them to restore the feature sequence to be detected, calculates the Shannon information entropy, and performs implicit anti-stripping judgment.

Images