Artwork digital copyright registration method based on block chain

Abstract

The invention discloses an artwork digital copyright registration method based on a block chain, and particularly relates to the technical field of digital copyright protection, and the method comprises the following steps: firstly, calculating a blind fingerprint for an artwork, generating a placeholder hash, writing the placeholder hash into the block chain, then opening a revelation window after the delay controlled by a verifiable random function, and submitting the fingerprint and metadata; when the similarity of the revelation works is too high, a congestion signal is generated, dynamic threshold adjustment is triggered, and deep fine similarity is obtained; fingerprint clusters are merged based on comparison results, original artworks and derivative artworks are confirmed according to detail scores after revelation is finished, and a credible copyright relationship is established; according to the method, tampering-free right confirmation of the artwork copyright information is realized, and the registration sequence is ensured to be verifiable; the similarity recognition accuracy is improved through dynamic threshold adjustment and under-chain fine comparison; and a clustering and detail scoring mechanism is used for automatically determining an original and derivative relationship, so that clear and traceable copyright affiliation is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of digital copyright protection, more particularly, the present application relates to an art digital copyright registration method based on a blockchain. BACKGROUND

[0002] With the widespread deployment of image generation artificial intelligence models, especially the trend of decentralized AI painting platforms based on diffusion models and generative adversarial networks, a large number of image content can be generated on a large scale within seconds and automatically published through an on-chain mechanism. Such platforms usually use a real-time on-chain mechanism, that is, each generated image is immediately written into the chain after generation to achieve "first-occupying" registration protection of the ownership of the work. However, in a public chain structure, this mechanism has caused serious copyright confusion problems.

[0003] For example, under a specific social hot topic, an AI drawing generator generates thousands of highly similar images in a very short time and uploads and writes them into the blockchain through a smart contract. At that time, some human artists were still in the process of creating or had not completed the scanning process, causing the high-resolution image on-chain time to be "AI-generated image" to be preempted, forming a "reversal of timestamp" factual misplacement.

[0004] Because most existing smart contracts use the logic of "first come, first registered, first ownership", the system automatically identifies AI-generated images as original registered works, and the works of human artists are considered as repeated registration or even infringement. This process is often automatically executed by an on-chain crawler without human review, resulting in the fingerprint of a legitimate original work being permanently covered or marked as non-original, making it impossible to establish rights on the chain. More seriously, once the AI-generated image spreads virally on social media, even if the artist can prove that he owns the original work, he still needs to go through a long and costly off-chain judicial or arbitration process to claim it. However, the on-chain status of copyright ownership is irreversible, and the real author almost loses the possibility of timely rights protection under the unalterable timestamp recording mechanism.

[0005] Such an attack is called "AI generator crazy writing copyright insertion attack", which is essentially a copyright pre-emption behavior created by taking advantage of the upload delay and the blind area of smart contract rules. Existing blockchain notarization solutions lack the ability to judge the similarity of image content, the complexity of visual details, and originality, and can only use time as the criterion for ownership determination, making it difficult to establish an original derivative structure relationship between multiple subjects simultaneously registered and highly similar images, and even more difficult to cope with the systematic congestion caused by the "high-density aggregation" of image fingerprints in a short time. Therefore, the present application proposes an art digital copyright registration method based on a blockchain to solve the above problems. SUMMARY

[0006] To achieve the above object, the present application provides the following technical solutions: The blockchain-based art digital copyright registration method comprises the following steps: The registration node calculates a blind fingerprint for the art to be registered, combines the blind fingerprint with a random salt value to generate a placeholder hash, and writes the placeholder hash together with a trusted timestamp into a blockchain to complete placeholder registration; The smart contract determines a delay time according to a verifiable random function, opens a reveal window when the delay time expires, and the registration node submits a complete fingerprint corresponding to the placeholder hash, a random salt value and art metadata within the reveal window; The smart contract real-time calculates the similarity distribution between the revealed fingerprints during the reveal window, and generates a congestion signal when the number of similar fingerprints exceeds a preset threshold; After the congestion signal is generated, the smart contract immediately reduces the dynamic similarity threshold and calls an off-chain multi-scale deep feature comparison algorithm to obtain a fine similarity evaluation result between the fingerprints; The smart contract merges the fingerprints with a similarity not lower than the dynamic similarity threshold into the same fingerprint cluster according to the fine similarity evaluation result, and generates a unique cluster identifier for each fingerprint cluster; At the end of the reveal window, the smart contract sorts the art in each fingerprint cluster according to the multi-scale detail score, registers the art with the highest detail score as the original art, and registers the other art in the same cluster as derivative art and establishes a correlation relationship.

[0007] In a preferred embodiment, the calculation of the blind fingerprint comprises the following steps: The registration node pre-processes the digital image of the art to be registered according to a fixed resolution and a uniform color space, and performs high-frequency noise suppression on the pixel matrix to exclude the interference of scanning device noise on the stability of the fingerprint; The registration node constructs a multi-scale discrete cosine transform feature vector based on the pre-processed pixel matrix, and inputs the feature vector into a one-way key hash function based on the art private key to generate a privacy protection feature string; The registration node applies a reversible quantization perturbation algorithm to the privacy protection feature string to reduce the probability of single-point pixel restoration while maintaining global perception consistency, obtaining a blind fingerprint that can be publicly verified and is difficult to reverse; The blind fingerprint obtained through the above three steps has uniqueness and anti-collision property, providing a secure fingerprint basis for subsequent placeholder operations.

[0008] In a preferred embodiment, the generation and writing of the placeholder hash comprises the following steps: The registration node randomly generates an encrypted random salt value matching the length of the blind fingerprint, and concatenates the blind fingerprint and the random salt value in a preset order to form a raw string to be hashed; The registration node calculates a fixed-length hash of the original string using a secure hash algorithm version 3, obtains a placeholder hash, and synchronously requests a trusted timestamp signature from a distributed timestamp authority; The registration node writes the placeholder hash, a random salt value, and its trusted timestamp as an indivisible data packet into a blockchain transaction payload and broadcasts it to the network. The placeholder is tamper-proof from the time the first confirmation block is generated. The placeholder operation achieved through the above three steps ensures that any subsequent registration request must reference the same combination of blinded fingerprints and random salt values, avoiding disputes over originality caused by the generator's massive insertion.

[0009] In a preferred embodiment, the process of determining the delay time based on a verifiable random function includes the following steps: The smart contract, after receiving the placeholder hash, calls the block header random number, node key commitment value, and current network difficulty to form an input seed, providing multiple source entropy for the verifiable random function; The smart contract places the input seed into a verifiable random function based on the elliptic curve digital signature algorithm, outputs a pseudo-random number that is unpredictable for any single node and publicly verifiable, and simultaneously generates a zero-knowledge proof; The smart contract linearly maps the delay time within the preset minimum and maximum delay intervals based on the pseudo-random number, and writes the delay time and zero-knowledge proof into the blockchain for external node verification. After the delay expires, the reveal window is automatically opened.

[0010] In a preferred embodiment, the input seed generation process includes the following steps: The smart contract reads the block header random number field of the most recent completed consensus immediately after the placeholder hash is written, and broadcasts this random number through event logs, allowing any node to verify retrospectively; The block validator calculates the node key commitment value using a one-time public key within a verification period before entering the next round of block generation, and submits it along with its digital signature to the on-chain key commitment registration contract. The smart contract verifies and immediately publishes the commitment value after verification; The smart contract continuously retrieves the difficulty fields of several latest blocks within a preset window, calculates the sliding average to form the current network difficulty, and synchronously writes it along with the timestamp into the state storage.

[0011] In a preferred embodiment, the process of generating a congestion signal includes the following steps: The smart contract performs discrete cosine transform on each pair of fingerprints within the same block after four levels of scaling, extracts the bit string and calculates the Hamming distance, and generates a multi-scale perception distance list; Taking the last n reveal records as a window, mark the fingerprints with a distance below the current similarity threshold as similar, and count the number of similar fingerprints within the window; Calculate the average of the first m window similarity counts, multiply the average by the inverse of the product of the preset safety factor as the new threshold after dynamic tightening; Trigger the congestion signal when the current window similarity count exceeds the new threshold.

[0012] In a preferred embodiment, the process of lowering the dynamic similarity threshold immediately after generating the congestion signal and obtaining the fine similarity evaluation results between fingerprints includes the following steps: The smart contract multiplies the preset similarity threshold by the preset shrinkage factor to generate a shrinkage threshold after the congestion signal is triggered, and immediately updates the global state; The smart contract synchronously broadcasts the list of fingerprints to be evaluated and the shrinkage threshold to the off-chain feature comparison service, and attaches a random challenge value to ensure that the returned result is verifiable; The off-chain comparison service extracts deep features according to the multi-scale convolutional network after receiving the instruction, calculates the cosine distance between fingerprints and returns a fine similarity matrix with zero-knowledge proof; The smart contract writes the fine similarity matrix into the fingerprint clustering mapping table after verifying the zero-knowledge proof, which is used for subsequent original right confirmation calls.

[0013] In a preferred embodiment, merging fingerprints with a similarity not lower than the dynamic similarity threshold into the same fingerprint cluster means: After receiving the fine similarity matrix, the smart contract maps fingerprints with a similarity higher than the dynamic similarity threshold, i.e. the shrinkage threshold, to edges in the graph structure, and constructs a similarity graph in real time; The smart contract calls the on-chain and set algorithm on the similarity graph to merge connected nodes from bottom to top, and automatically selects the fingerprint with the highest weight as the cluster root of the connected component according to the edge weight after each merge; The smart contract calculates a secure hash by concatenating the current block number and the random entropy of each cluster root fingerprint, generates a unique cluster identifier, and stores the cluster identifier and all fingerprints in the cluster in the state table.

[0014] In a preferred embodiment, registering the artwork with the highest detail score as the original artwork means: The smart contract performs three-layer discrete wavelet decomposition on each fingerprint in each fingerprint cluster, calculates the energy of each layer of details, and aggregates them into a multi-scale detail score according to the preset weight; The smart contract sorts the cluster members according to the multi-scale detail score from high to low, and if the difference between the highest score and the second highest score is lower than the threshold, it will add a visual entropy comparison to determine the only first; The smart contract registers the artwork with the highest score as the original artwork, and writes the derivative relationship for the remaining artworks in the same cluster in the state mapping table, records the original identifier, derivative level and timestamp.

[0015] Technical effects and advantages of the present application: The present application realizes the copyright pre-emption and the non-tamperable copyright mechanism of the registration sequence by writing the placeholder hash and the trusted timestamp into the blockchain. In the registration process proposed by the present application, the registration node first calculates the blind fingerprint of the artwork to be registered, and generates a placeholder hash combined with a random salt value, and writes the trusted timestamp into the blockchain at the same time, to complete the irreversible original placeholder. Through the above operation, the artwork copyright information is certified on the chain at a specific time point, effectively preventing disputes over ownership caused by unclear copyright order or information modification. At the same time, the irreversible and tamper-proof writing behavior enables the registration process to have legal traceability and time sequence verifiability, providing a trusted basis for digital copyright protection.

[0016] The present application realizes the stability and congestion response ability of the similarity evaluation between artworks through dynamic similarity threshold adjustment and off-chain fine comparison. In the disclosure window stage, the smart contract monitors the similarity distribution between fingerprints in real time, and generates a congestion signal automatically when the similarity concentration rises to the congestion threshold. The present application reduces the dynamic similarity threshold in real time to trigger a more fine off-chain fingerprint comparison process, thereby improving the fingerprint differentiation ability while maintaining system processing efficiency. This dynamic adjustment mechanism can adapt to actual scenarios such as sudden increase in the number of disclosure nodes and dense similar artworks, ensuring that the system's identification of differences between artworks has flexibility and stability, effectively avoiding registration conflicts and copyright confusion.

[0017] The present application realizes the objective establishment of the originality of artworks and the system marking of derivative works through the fingerprint clustering and detail scoring sorting mechanism. After the disclosure window ends, the present application sorts all artworks in the same fingerprint cluster according to the preset detail scoring method, and clearly scores the highest one as the original artwork and the others as derivative artworks. This mechanism realizes the automatic copyright of the originality of artworks, avoiding disputes and uncertainties caused by relying on subjective human judgment. At the same time, the derivative relationship between the artworks in the cluster is registered and mapped by the system, improving the structured expression ability of copyright information, making the subsequent copyright tracking, derivative management and responsibility determination more clear and traceable. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to facilitate the understanding of those skilled in the art, the present application will be further described below in conjunction with the drawings; Figure 1 The principle diagram of the blockchain-based digital copyright registration method for artworks in the present application. DETAILED DESCRIPTION

[0019] With reference to the drawings of the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described, obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work belong to the scope of protection of the present application.

[0020] With reference to Figure 1 The following examples are obtained: Example 1: a blockchain-based art digital copyright registration method, comprising the following steps: The registration node calculates the blind fingerprint of the art to be registered, combines the blind fingerprint with a random salt value to generate a placeholder hash, and writes the placeholder hash together with a trusted timestamp into the blockchain, to complete the registration placeholder. The registration node performs blind fingerprint calculation to extract a compressed feature expression with uniqueness from the original art image, which still retains the overall content structure perception ability of the image under irreversible conditions, and prevents unauthorized subjects from restoring the original image. By combining the blind fingerprint with a random salt value generated by a high-strength entropy source, a unique placeholder hash is formed, which can effectively prevent primary key conflicts caused by repeated registration of the same work at different nodes. At the same time, the addition of a trusted timestamp makes the registration behavior have public credibility in time, ensuring that the "first-occupy-then-lock" ordering principle can be accepted by on-chain validators. Finally, the writing of this combined data package into the blockchain makes the registration intention unforgeable and traceable, providing an initial binding basis for subsequent disclosure and originality right establishment operations.

[0021] The smart contract determines the delay time according to the verifiable random function, opens the disclosure window when the delay expires, and the registration node submits the complete fingerprint corresponding to the placeholder hash, the random salt value and the art metadata within the disclosure window; The setting of the delay time is determined by the smart contract calling the random function with public verification capability, which can effectively prevent the registration node from predicting the specific time point and thus preempting the disclosure order, realizing the fair ordering in the sense of probability. After the delay expires, the disclosure window is automatically opened by the smart contract, so that the registration node can only submit its actual work fingerprint, salt value for restoration and verification, and supporting work metadata within this time period, avoiding the malicious ordering manipulation caused by the artificial bundling of submission and time point on the chain. The introduction of the disclosure window makes the system have the "submission first-occupy-unified disclosure" anti-insertion mechanism, effectively improving the fairness of original works and automatically generated images in the time ordering dimension.

[0022] The smart contract calculates the similarity distribution between the disclosed fingerprints in real time during the disclosure window, and generates a congestion signal when the number of similar fingerprints exceeds the preset threshold. During the disclosure window, the system automatically performs multi-scale perception fingerprint comparison on the submitted fingerprints through the smart contract to calculate the global similarity distribution. This process is real-time, ensuring that the image similarity density level in the current window is always perceived on the chain. By setting the number of similar images as a dynamic trigger point, the system can detect a large number of similar or repeated images submitted in a batch, triggering a congestion signal to determine whether there is a non-natural, batch submission behavior in that time period. The generation of the congestion signal serves as a warning mechanism, driving the system to enter a more stringent fingerprint processing path to ensure that a large number of similar images do not occupy the rights confirmation entry in a short period of time.

[0023] After generating the congestion signal, the smart contract immediately reduces the dynamic similarity threshold and calls the off-chain multi-scale deep feature comparison algorithm to obtain fine similarity evaluation results between fingerprints. When the congestion signal is activated, the system automatically tightens the perception threshold for judging image similarity and adjusts it to a lower lower limit, thereby narrowing the acceptable range of images judged as "different works". This adjustment requires the subsequent comparison algorithm to have higher detail resolution when processing image pairs. At the same time, the smart contract submits the fingerprint list to the off-chain deep comparison service and calls the multi-scale feature extraction algorithm with convolutional neural structure to perform fine-grained similarity scoring of images in multiple dimensions such as texture, boundary, and morphology. This operation forms more reliable fine evaluation results as a key technical basis for clustering and originality confirmation.

[0024] The smart contract merges fingerprints with a similarity not lower than the dynamic similarity threshold into the same fingerprint cluster based on the fine similarity evaluation results, and generates a unique cluster identifier for each fingerprint cluster. Based on the fine similarity matrix returned by the off-chain, the system constructs a similarity relationship graph between fingerprints on the chain and identifies all image pairs with a similarity exceeding the current dynamic threshold, clustering them according to the graph structure. Each cluster is considered a group of potentially related artworks, with high similarity or derivation among its members. To achieve persistent tracking of on-chain state, the smart contract automatically generates a unique cluster identifier with a block number and a random factor for each cluster, and establishes a bidirectional mapping between the identifier and the cluster members. This step not only achieves the merging and isolation of similar works, but also provides a structural basis for originality confirmation and subsequent derivative chain labeling.

[0025] At the end of the reveal window, the smart contract sorts the artworks within each fingerprint cluster based on the multi-scale detail score, registers the artwork with the highest detail score as the original artwork, and registers the other artworks within the same cluster as derivative artworks and establishes a correlation relationship. After the reveal window is closed, the smart contract traverses all generated fingerprint clusters and performs multi-scale detail score calculation on the artwork images in each cluster. The score combines image edges, textures, and local contrast, and other multi-dimensional features to comprehensively evaluate the visual complexity and structural richness of the image. The system uses the score results for sorting and confirms the highest scorer as the original artwork within the cluster, recording its original identity in the state table. At the same time, the remaining members are marked as derivative artworks, with derivative levels set according to similarity to the original image and submission time, and a traceable original-derivative correspondence relationship is established on the chain to ensure the structured transparency of the copyright chain.

[0026] The calculation of the blinded fingerprint includes the following steps: the registration node pre-processes the digital image of the artwork to be registered according to a fixed resolution and a uniform color space, and performs high-frequency noise suppression on the pixel matrix to exclude the interference of scanning device noise on the stability of the fingerprint; the digital image is an electronic file that records the artwork pattern in the form of a pixel array, and the format can be a compressed image or an original image. The registration node first adjusts the image to a uniform square resolution of 1024 pixels in width and 1024 pixels in height, and converts it to a uniform color space, where the color space adopts a structure that decouples the luminance component and the color contrast component, and the conversion result retains the luminance channel for subsequent processing. A two-dimensional pixel matrix is constructed based on the luminance channel, where each element in the matrix represents the gray value of the image position, and the gray value range is from zero to two hundred and fifty-five. The registration node performs high-frequency noise suppression based on the pixel matrix, using a three-by-three median filter operator sliding window to process the image area to remove edge burrs and random noise points introduced during scanning or photographing. The filtered image retains the original edge profile in structure, but the high-frequency artifacts are effectively suppressed, improving the stability of subsequent feature extraction.

[0027] The registration node constructs a multi-scale discrete cosine transform feature vector based on the preprocessed pixel matrix, and inputs the feature vector into a one-way key hash function based on the artwork private key to generate a privacy protection feature string; the multi-scale discrete cosine transform refers to performing frequency domain conversion on image pixel blocks under different sizes, which is used to extract the structural features and texture features of the image. The registration node divides the image into image blocks of two sizes, i.e., eight-by-eight pixel blocks and sixteen-by-sixteen pixel blocks. After performing two-dimensional discrete cosine transform on each block, the transform coefficients are extracted, and nine energy maximum coefficients are selected from the top-left low-frequency position and expanded by column to form a set of local structural feature vectors. A total of one hundred and twenty-eight groups of feature vectors are extracted from each image, each group has a dimension of nine, and finally they are spliced into an image multi-scale structural vector with a total length of one thousand one hundred and fifty-two. To prevent direct fingerprint from inferring the original image, the vector is input into a one-way key hash function. The input of the one-way key hash function is the splicing result of the image multi-scale structural vector and the artwork private key, and the output is a fixed structure digest with a length of five hundred and twelve bits. The artwork private key is generated locally by the image registrar and is not disclosed with the image body, has the functions of identity binding and privacy protection, and ensures that the digest can only be generated in a forward direction and cannot be reversed.

[0028] The registration node applies a reversible quantization disturbance algorithm to the privacy protection feature string to reduce the probability of single-point pixel restoration while maintaining global perception consistency, and obtains a blinded fingerprint that can be publicly verified and is difficult to reverse; the reversible quantization disturbance algorithm is used to add disturbance terms to the feature string while maintaining the main perception features, to prevent local pixels from being guessed. The algorithm divides the five hundred and twelve-bit feature string into sixty-four groups, each group having eight bits. The difference between the maximum and minimum jump values of each group of eight-bit data is calculated first. If the difference is greater than sixteen, no disturbance is performed, and if the difference is less than sixteen, a disturbance factor generated by a pseudo-random function is added to the original value, and the disturbance factor ranges from zero to seven. The disturbed value is operated in modulo operation within the range of eight-bit integers to ensure that it does not overflow. Finally, the sixty-four groups of disturbed values are spliced to generate a blinded fingerprint vector that cannot be recovered. The data distribution in the blinded fingerprint vector is consistent with the main structure of the original image features, but the local pixel positions do not have restoration properties, so even if they are obtained, the original image structure cannot be accurately reconstructed, ensuring the dual security of privacy and copyright.

[0029] The blind fingerprint obtained by the above three steps has uniqueness and anti-collision property, providing a secure fingerprint basis for subsequent placeholder operation; the uniqueness of the blind fingerprint is reflected in that even if two images are very similar, as long as the pixel content or structure is slightly different, the final perturbation feature will jump globally; the anti-collision property is reflected in that the probability of hash collision of the blind fingerprint of any two different art images is less than one in one billion. The blind fingerprint can be used as the identity certificate of the on-chain registration behavior of the art, and is the unique entry point for generating the placeholder hash, the right time stamp and the copyright mapping. Due to its irreversible, secure binding, cross-system unique and other characteristics, it becomes the underlying fingerprint structure supporting the trusted operation of the entire blockchain-based art digital copyright registration mechanism.

[0030] The generation and writing of the placeholder hash include the following steps: the registration node randomly generates an encrypted-level random salt value matching the length of the blind fingerprint, and concatenates the blind fingerprint and the random salt value in a predetermined order to form a raw string to be hashed; the encrypted-level random salt value refers to a piece of random data generated under the requirement of cryptographic security, which is used to prevent the same fingerprint from producing the same result in multiple on-chain operations and ensure the unpredictability of the hash output. The registration node generates a random number sequence with the same length as the blind fingerprint by calling the built-in secure entropy source of the operating system, such as an entropy pool based on environmental noise, a thermal sensitive jitter source or a high resolution timing difference disturbance mechanism. For example, if the length of the blind fingerprint is 512 bits, the registration node will generate a random salt value consisting of 64 groups of 8-bit integers, and each group has a value ranging from 0 to 255.

[0031] When concatenating the blind fingerprint and the random salt value, the order is fingerprint first and salt value second, and the concatenation method is to append each group one by one, i.e., the first group of 8-bit numbers of the blind fingerprint is placed in the first 8 bits, and the first group of 8-bit numbers of the corresponding salt value is placed immediately after it, and so on, to form a raw string to be hashed with a length of 1024 bits. This structure avoids the prefix collision problem, so that even if two images have the same fingerprint, the resulting string will be different after adding different salt values.

[0032] The registration node calculates the fixed-length hash of the raw string to be hashed using the secure hash algorithm version 3 to obtain the placeholder hash, and simultaneously requests a trusted timestamp signature from the distributed timestamp authority; the secure hash algorithm version 3 is a one-way irreversible function defined by the national cryptographic standard, which has the characteristics of compression, anti-collision and strong expansibility. In this method, the registration node calls the secure hash algorithm version 3 function with a fixed length of 512 bits to perform one-way hashing on the aforementioned 1024-bit concatenated raw string, and outputs a fixed-length placeholder hash value of 256 bits. This hash value serves as the unique identifier for subsequent on-chain registration behavior, and has tamper-proof and duplicate detection capabilities.

[0033] After the hash is completed, the registration node will take the placeholder hash as part of the original data, and synchronize the timestamp application to the distributed timestamp authority that has joined the blockchain network. The authority will add a trusted timestamp to the received hash value through a consensus signature mechanism, including the receiving time, node signature, and network chain number index fields, forming a complete and trusted timestamp signature structure. This signature structure has timeliness, consistency, and verification repeatability, ensuring that the entire registration process has legal effect and third-party independence in time.

[0034] The registration node writes the placeholder hash, random salt value, and its trusted timestamp as an indivisible data packet into the blockchain transaction payload and broadcasts it to the network. The placeholder record is tamper-proof at the time of the first confirmed block generation; after obtaining the trusted timestamp signature, the registration node forms a complete data packet with the above three types of data. This data packet structure contains three parts: 1) placeholder hash value, occupying 256 bits, used as a unique identifier for registration; 2) encrypted random salt value, occupying 512 bits, used as a verification basis in subsequent disclosure; 3) trusted timestamp signature structure, occupying a variable-length field, typically containing signature information, time field, and chain number totaling about 1000 bits. The entire data packet is encoded as the data payload part of the blockchain transaction structure and submitted to the entire network through the broadcast mechanism, entering the transaction pool of the accounting node. Miners or validators will include this transaction in the new block when packaging the new block, and record its hash digest in the block header. With the on-chain confirmation of the new block, the entire data packet content becomes an indivisible part of the block, which can be read back by any node in the future and cannot be tampered with or withdrawn.

[0035] The placeholder operation achieved through the above three steps ensures that any subsequent registration request must reference the same combination of blind fingerprint and random salt value, avoiding disputes over original rights caused by the generator's massive placeholder casting; since the generation of the placeholder hash strictly depends on the unique combination of the blind fingerprint and the random salt value, any change in the fingerprint or use of a different salt value will cause unpredictable changes in the hash value. Therefore, in the disclosure window, if the registration node cannot submit the complete fingerprint and salt value pair corresponding to the hash, it cannot complete the subsequent right confirmation, and the system will automatically determine that the registration is invalid. At the same time, since the entire registration record has a chain timestamp, any image generator who tries to mass placeholder before the real artwork is disclosed must know the artist's image details and salt value structure, which is almost impossible to achieve in practice, effectively preventing the malicious occupation of original rights by placeholder attacks, and ensuring the priority of the real author in time and structure.

[0036] The process of determining the delay time according to the verifiable random function includes the following steps: the smart contract calls the block header random number, the node key commitment value and the current network difficulty to form an input seed after receiving the placeholder hash, which provides multi-source entropy for the verifiable random function; the block header random number refers to the unpredictable random value contained in the latest block that has completed consensus on the current main chain, and the generation method usually depends on the threshold aggregate signature result submitted by the verifier under the proof of stake mechanism, or the block hash fragment submitted by the mining node under the proof of work mechanism. For example, in a proof of stake network, the system uses a hybrid entropy value calculated and aggregated by all validators, which is usually 256 bits long and can be obtained by calling the on-chain randomness interface through the smart contract.

[0037] The node key commitment value refers to a key digest published in advance by the block node in the previous consensus period for future behavior verification, which is obtained by securely hashing the one-time public key of the node and is 256 bits long, which is used to prevent subsequent forgery of input entropy. The current network difficulty refers to the block challenge strength index automatically adjusted by the main chain within a certain time period, which can be calculated based on the average interval time of the previous 100 blocks, and in the proof of stake mechanism, it can also be represented as the weighted inverse value of the average slot distance. The value is usually a positive integer, generally between hundreds of millions and several billion. The above three values are combined into an input seed with a total length of 768 bits, which is used as the input of the verifiable random function, providing sufficient and anti-predictive multi-source entropy input for subsequent random delay generation, ensuring that any single node cannot control the delay time.

[0038] The smart contract places the input seed into the verifiable random function based on the elliptic curve digital signature algorithm, outputs a pseudo-random number that is unpredictable for any single node and can be publicly verified, and synchronously generates a zero-knowledge proof; the verifiable random function is a pseudo-random number generation mechanism with anti-tampering ability, whose output is unpredictable and can be publicly proven to the external verifier. After receiving the input seed, the smart contract calls the random function component based on the elliptic curve digital signature algorithm, and the specific curve can be a parameter family in the elliptic curve hyperbolic curve group with a security level of 256 bits, and a digital signature operation is performed on the basis of the curve.

[0039] The input seed serves as the private key's computation factor, generating a 256-bit random output through elliptic curve mapping. This output value falls between zero and 2^256 minus one. To prevent nodes from forging results, the random function simultaneously generates a zero-knowledge proof. This proof employs a non-interactive proof structure, constructing a cryptographic commitment and challenge-response pair to prove that the pseudo-random number was indeed uniquely calculated from the input seed and that the calculation process was not tampered with. This pseudo-random number serves as the basis for the final delay calculation, satisfying the requirements of unpredictability, repeatability, and determinism. Furthermore, by being jointly written into the blockchain state with the zero-knowledge proof, it allows any node to verify the legitimacy of the delay source without requiring trust.

[0040] The smart contract uses a pseudo-random number to linearly map the delay time within a preset shortest and longest delay range, and writes the delay time and zero-knowledge proof into the blockchain for external node verification. The disclosure window automatically opens after the delay expires. The system sets a uniform upper and lower limit for the delay time range; for example, the shortest delay is 50 block heights, and the longest delay is 150 block heights, meaning the disclosure window will open between 50 and 150 blocks after the placeholder registration. The smart contract normalizes the pseudo-random number to a small value between zero and one using the following formula: divide the pseudo-random number by the maximum possible value (2 to the power of 256 minus 1), multiply by the interval length of 100, add the lower limit of 50, and round the result to the final integer number of delayed blocks.

[0041] For example, if the pseudo-random number, after normalization, results in 0.45, the final delay time will be 50 plus 45, or 95 blocks. With a block interval of 12 seconds set by the blockchain platform, this delay time is approximately 19 minutes. The delay time and its corresponding zero-knowledge proof are synchronously written to the transaction log and contract state table for subsequent network participants to verify its compliance. When the on-chain height reaches the placeholder transaction confirmation height plus the number of delayed blocks, the smart contract automatically opens the disclosure window for the corresponding work, allowing registration nodes to submit the corresponding complete fingerprint, salt value, and metadata, entering the next stage of the rights confirmation process. This delay control mechanism enables "first-come, first-served" registration intent, effectively suppressing attack paths where automatically generated images seize the initiative in a very short time, while ensuring both the fairness of the disclosure window and on-chain verifiability.

[0042] The generation process of the input seed includes the following steps: the smart contract reads the block header random number field of the last completed consensus immediately after the placeholder hash is written, and broadcasts the random number through the event log, so that any node can verify it; the block header random number field refers to the random entropy value field in each newly generated block on the chain, which is generated by different methods under different consensus mechanisms. In the proof of stake consensus mechanism, this field usually comes from the random number aggregation result of the last verification period, which is the mixed random value of the entire network generated by the weighted aggregation of the verification node after all active validators sign the specified seed on the chain in the order of key signature. This field can be directly called through the random value field in the block structure on the chain; in the proof of work network, this field can be selected as a fixed offset field from the work output hash result of the previous block as a substitute.

[0043] The smart contract reads the latest block header random value of the completed consensus in the same transaction processing period after the placeholder hash is successfully written, and encapsulates the value together with the identity information of the current transaction initiating node as a chain event, which is broadcast to all full nodes through the event log form, to ensure that all network participants can verify that the random value comes from the history on the chain and has not been manipulated. As part of the input seed, this field provides uncertainty entropy input under global consensus for the random function, ensuring that it is difficult to predict and cannot be played back.

[0044] The block validator calculates the node key commitment value using a one-time public key before entering the next block in a verification period, and submits it together with its digital signature to the key commitment registration contract on the chain. The smart contract verifies and immediately publishes the commitment value; the one-time public key is a temporary encryption public key generated by the block validator before each block to prevent future information from being tampered with or replayed, which is used for pairing verification in subsequent identity verification. After generating a one-time key pair, the block validator uses the secure hash algorithm in the national cryptography standard to hash the public key, and the calculation result is a node key commitment value with a length of two hundred and fifty-six bits.

[0045] To ensure the validity and correspondence of the commitment value, the verifier also needs to sign the hash result using the current identity key, and encapsulate the commitment value and the signature result together as a submission data packet, and write it to the key commitment registration contract running on the chain. After receiving the data, the contract will match the signature and identity pair, confirm that the signature is true, the structure is correct, and the submission period is legal, and immediately write the node key commitment value as a public state in the mapping table, and bind it with its block period. This key commitment value participates in seed construction when generating pseudo-random numbers, ensuring that pseudo-random numbers not only come from on-chain data, but also contain encryption commitments independently contributed by the node, increasing the unpredictability and anti-forgery ability of the random source.

[0046] The smart contract continuously retrieves the difficulty fields of several latest blocks within a preset window, calculates a sliding average to form a current network difficulty, and synchronously writes the timestamp into the state storage; the current network difficulty is a key parameter for measuring the computational complexity of block generation, which reflects the real-time pressure and network activity of the system. The smart contract sets the sliding window length to one hundred blocks, and automatically retrieves the difficulty fields in the last one hundred block headers from the current chain height after receiving each new placeholder request. Each field is extracted and combined to form a difficulty sequence with a length of one hundred. For this sequence, the contract calculates its arithmetic mean to obtain the average block generation difficulty in the current time period, which is used to eliminate the interference of single abnormal fluctuations on random generation. If the running blockchain system is a proof-of-stake network, the average slot interval or block generation interval of each block is extracted as the logically equivalent difficulty indicator, and equivalent conversion is performed. This average value is written into the smart contract state variable as the current network difficulty, and is accompanied by the current system timestamp when written, ensuring that the input seed generated subsequently is highly consistent with the actual block generation state of the network in time, and is not affected by short-term abnormalities such as isolated blocks and delayed blocks. This network difficulty participates in the construction of the pseudo-random function input seed, together with the block header random number and node key commitment value, to form a complete three-source entropy structure, providing a high-quality unpredictable input basis.

[0047] The process of generating congestion signals includes the following steps: the smart contract performs discrete cosine transform on each pair of fingerprints in the same block after four levels of scaling, extracts the bit string and calculates the Hamming distance to generate a multi-scale perception distance list; each pair of fingerprints refers to the pair of artwork fingerprints that have successfully completed the reveal and entered the similarity detection process in the current block. The smart contract performs four-level image scaling on each fingerprint image, which includes original resolution, one-half, one-fourth, and one-eighth scaling sizes. Each level of scaling uses the conservative mean downsampling method, which uses the mean value of each block of pixels as the corresponding pixel of the target image to preserve the image contour features and reduce noise.

[0048] For each level of scaled image, the smart contract performs a two-dimensional discrete cosine transform to convert the image block into a frequency coefficient matrix. In each frequency domain matrix, the smart contract selects the eight low-frequency coefficients in the upper left corner region and combines them into a bit string with a length of eight in the order from left to right and from top to bottom. Four scaling levels will generate four groups of bit strings, totaling thirty-two bits of information, which represent the structural perception features of the image at multiple scales.

[0049] The smart contract compares each pair of bit strings between all image fingerprints in the current block, calculates the Hamming distance, i.e. the total number of different corresponding bit positions. The Hamming distances of all fingerprint pairs are recorded in a multi-scale perceptual distance list, which contains a pair of fingerprint identifiers, a distance value scaled by four levels, an average distance, and an aggregated score for each entry. This data structure is used for subsequent similarity judgment.

[0050] Marking fingerprints with a distance below the current similarity threshold as similar and counting the number of similar fingerprints within a window of the last n reveal records. The smart contract maintains a sliding window to count the number of similar fingerprints generated in the last n reveal records. For example, the window size is one hundred, which means that every time a new fingerprint reveal is completed, the system will remove the oldest data and add the latest record to ensure that the data reflects the current stage of the reveal image similarity density state.

[0051] For all fingerprint pairs in the current sliding window, the smart contract determines whether the average value of the multi-scale perceptual distance and the aggregated score is below the similarity threshold currently running in the system. If the distance value is less than the threshold, the pair of fingerprints is considered similar, and the record is recorded in the similar count value. The similar count represents the aggregated performance of the similarity of the image content in the current window, and is a key signal indicator for the system to monitor whether there is a gap attack or a large number of repeatedly generated images.

[0052] Calculate the average of the similar count of the previous m windows, multiply the average by the inverse of the product of the preset safety factors as the new threshold after dynamic tightening; To determine whether the similarity count in the current window is abnormal, the smart contract needs to establish a historical baseline. The system sets the sliding window history depth to ten, i.e. the smart contract records the similar count values in the past ten windows to form a historical count sequence. Calculate the arithmetic mean of the sequence, denoted as the historical mean. In addition, the system maintains a set of safety factors related to window state, network congestion probability, block interval changes, etc., such as window congestion factor, node feedback factor, historical reveal delay factor, etc. Each safety factor is a decimal greater than zero, and the product of all factors is the safety factor product. The system uses the inverse of the safety factor product as a risk adjustment factor, and multiplies the historical mean by the inverse to obtain the new threshold after dynamic tightening. For example, if the average similar count of the previous ten windows is one hundred and fifty, the safety factors are zero point eight, zero point nine and zero point nine five, the product is zero point six eight four, the inverse is about one point four six two, and the new threshold is one hundred and fifty times one point four six two, which is about two hundred and nineteen. This threshold represents the maximum number of similar pairs that the system can tolerate in the current window. Once the actual similar count exceeds this value, it enters the congestion state. m and n are both preset positive integers.

[0053] The congestion signal is triggered when the current window similarity count exceeds the new threshold. The smart contract continuously monitors the real-time similarity count in the sliding window, and when the value first exceeds the dynamically calculated threshold described above, the system immediately triggers the congestion signal. The triggering behavior of the congestion signal is executed in the smart contract through the event broadcast mechanism, and the congestion occurrence time and corresponding block number are marked in the on-chain state table. This signal will simultaneously activate multiple on-chain and off-chain processing procedures, including similarity judgment precision tightening, calling the off-chain deep feature comparison module, adjusting the perception hash distance calculation precision, increasing the fingerprint clustering density, freezing the permissions of addresses with abnormal frequent submission, etc. By introducing the congestion signal mechanism, the system can adopt more stringent and detailed processing strategies in the high-risk interval where the image disclosure behavior set is concentrated and the content similarity increases sharply, preventing the generator from large-scale automated casting behavior from impacting the original right ranking structure.

[0054] The process of reducing the dynamic similarity threshold immediately after generating the congestion signal and obtaining the fine similarity evaluation results between fingerprints includes the following steps: the smart contract generates a contraction threshold by multiplying the preset similarity threshold by a preset contraction coefficient after the congestion signal is triggered, and immediately updates the global state; the similarity threshold is the system reference value currently used to judge whether two groups of fingerprints are similar, and in the normal state, the value is adjusted according to the historical window mean and the safety factor, and its unit is Hamming distance or distance normalized ratio. The preset contraction coefficient is a compression ratio fixedly configured in the on-chain parameter, which is used to quickly tighten the system's tolerance for similarity after the network or window enters an abnormal state. The coefficient usually takes a value range of zero point six to zero point nine, for example, the contraction coefficient is zero point seven five, which means that after contraction, only seven percent of the original judgment distance is allowed.

[0055] At the first time after the system triggers the congestion signal, the smart contract multiplies the current preset similarity threshold by the contraction coefficient to obtain the contraction threshold. For example, if the current threshold is thirty-two-bit Hamming distance and the contraction coefficient is zero point seven five, then the contraction threshold is twenty-four-bit distance. The newly generated contraction threshold is immediately written into the contract state table, replacing the dynamic threshold as the only threshold referenced by subsequent similarity judgment and comparison operations, and the adjustment operation is also written into the event log for subsequent verification and backtracking.

[0056] The smart contract synchronously broadcasts the list of fingerprints to be evaluated and the shrinkage threshold to the off-chain feature comparison service, with a random challenge value to ensure that the returned result is verifiable. The off-chain feature comparison service is an auxiliary computing system deployed in a trusted environment, capable of performing complex neural network operations and generating proof structures. After completing the shrinkage threshold update, the smart contract extracts all the artwork fingerprints in the current disclosure window that are to be involved in clustering, forming a list of fingerprints to be compared, usually thirty to five hundred image structure vectors. The contract packages the fingerprint list and the shrinkage threshold into a data packet and broadcasts it to the off-chain comparison service through an asynchronous messaging mechanism as an initialization instruction for a comparison task. At the same time, to prevent the off-chain returned data from being forged or replayed, the contract synchronously generates a set of random challenge values based on block height, node address, and task number hash calculation, which are embedded in the data packet. The off-chain service is required to return a calculation proof bound to it after the calculation is completed, ensuring the legitimacy, non-replicability, and one-time nature of the task response.

[0057] After receiving the instruction, the off-chain comparison service extracts deep features using a multi-scale convolutional network, calculates the cosine distance between the fingerprints, and returns a fine similarity matrix with zero-knowledge proof. The off-chain comparison service performs multi-scale convolutional neural network processing on each fingerprint image to extract deep structural features of the image. The network structure is usually composed of four convolutional layers, which process the edges, textures, shapes, and regional distributions of the image at different spatial frequencies. The extraction results of each convolution layer are processed by batch normalization and output through an activation function, and finally spliced into a deep feature vector with a length of 1024, which is used to describe the overall structural semantic information of the image.

[0058] For any two image deep feature vectors, the off-chain system calculates their cosine distance as the similarity score, with the formula: the dot product of the feature vectors divided by the product of their respective lengths. A cosine distance close to one indicates that the two images are highly similar, and a cosine distance close to zero indicates that the two images are quite different. The off-chain system arranges all the cosine distances between images into a symmetric matrix, denoted as the fine similarity matrix, with a dimension of n by n, where n is the number of images involved in the comparison. The comparison service constructs a zero-knowledge proof based on the challenge value attached in the initialization instruction, proving that the matrix result is calculated from the given input and that the intermediate process has not been modified. The proof uses a non-interactive zero-knowledge structure, including a commitment item, a response item, and an encryption tag, which can be quickly confirmed by the on-chain using a unified verification algorithm.

[0059] The smart contract writes the fine similarity matrix into the fingerprint clustering mapping table after verifying the zero-knowledge proof, for subsequent originality confirmation calls. After the comparison result is returned, the smart contract first verifies the integrity of the accompanying zero-knowledge proof. The verification algorithm checks the challenge value against the shrinkage threshold to confirm the consistency of the comparison task and the response data, and also verifies that the response item in the proof and the submission result satisfy the mathematical construction rules, ensuring that the calculation result is reliable and unique in origin.

[0060] After the proof passes the verification, the contract writes the fine similarity matrix into the fingerprint clustering mapping table according to the standard structure. This mapping table is a structured state object on the chain, used to record high-precision similarity scores between image fingerprints, and is the core basis for subsequent clustering division, originality selection, and derivative identity binding. Each similarity score record consists of two image identification codes, a set of floating-point score values, a comparison period stamp, and a confirmation mark. After writing, it can be read in real time by the originality confirmation process, and supports sorting and filtering operations. Through the above process, the system switches to a higher-precision off-chain comparison path in a congested state, maintaining the response capability to high-frequency attack scenarios, while also considering the verifiability and processing efficiency of the results, building a secure, stable, and reliable foundation for clustering and rights confirmation.

[0061] Merging fingerprints with a similarity not lower than the dynamic similarity threshold into the same fingerprint cluster means that the smart contract maps fingerprints with a similarity higher than the dynamic similarity threshold, i.e., the shrinkage threshold, to edges in the graph structure after receiving the fine similarity matrix, and constructs a similarity graph in real time. The fine similarity matrix is a symmetric matrix used to record the similarity scores between all disclosed artwork fingerprints. Each cell represents the cosine distance score between two images, with a range of zero to one. The larger the value, the more similar the images. After reading the matrix, the smart contract iterates through each pair of fingerprint scores and marks the fingerprint pair with a score greater than the current shrinkage threshold as an edge in the graph structure. Each edge represents two image nodes with strong similarity in the visual structure and should belong to the same visual cluster. For example, in a comparison result containing fifty artwork images, if the similarity score between the tenth and eighteenth images is 0.92, which is higher than the current shrinkage threshold of 0.85, a directed edge from number ten to number eighteen is added to the similarity graph with a score weight label. This process continuously scans the matrix and gradually generates a set of similar nodes composed of fingerprint images and their edge connections in the graph structure.

[0062] The smart contract calls the on-chain and merges the connected nodes from bottom to top on the similar graph, and automatically selects the highest weight fingerprint as the cluster root of the connected component according to the edge weight after each merge; after the construction of the similar graph is completed, the smart contract takes it as the input and calls the on-chain deployment of the union set structure to merge the nodes. The union set is a dynamic data structure for efficiently maintaining the merging of sets and the relationship of belonging, which is implemented in the form of a mapping table in the on-chain environment, with the key being the fingerprint identifier and the value being the root node identifier of the current set to which the fingerprint belongs. In the initial state, all fingerprint nodes form a cluster by themselves, and the root node is the identifier itself. The contract traverses all edges in the similar graph according to the edge table, and for each edge processed, the following steps are executed: Step 1: call the path compression lookup algorithm to get the root nodes of the two nodes at both ends of the current edge; Step 2: if the two root nodes are different, call the rank-based merging strategy to merge the set with fewer nodes into the set with more nodes; Step 3: after the merge is completed, set the root node of the new set as the cluster representative based on the maximum edge weight, i.e., the highest weight fingerprint. The edge weight is the similarity score value between the pair of fingerprints, which is directly taken from the corresponding floating-point number in the fine similarity matrix. For example, if the edge score between node five and node twenty-four is 0.998, which is the maximum score value in the current graph, and five and twenty-four are finally merged into a cluster, then the number five is set as the cluster root node. All fingerprints belonging to this cluster will be pointed to this node as the home identifier in the subsequent steps.

[0063] The smart contract calculates the secure hash of each cluster root fingerprint by concatenating the current block sequence number and the random entropy seed, generates a unique cluster identifier, and stores the cluster identifier and all fingerprints in the cluster in the state table. After completing the cluster structure merging, the smart contract traverses all the current cluster structures, and takes each cluster's root node as the core identity to build a unique identifier. The generation method is to concatenate the image fingerprint number of the root node with the current block sequence number, and add a random string of length 256 extracted from the system entropy source, then input the whole data into the national cryptographic standard secure hash algorithm, and output a hash value of length 256 as the unique cluster identifier. Taking the fingerprint with number thirty-seven as the root node, its block sequence number is one thousand and twenty, and the random entropy seed is the binary string "0010011", the concatenated whole input string is "thirty-seven plus one thousand and twenty plus zero one zero zero one one", and the cluster identifier generated after hash calculation is like "9876543". The smart contract establishes a bidirectional mapping relationship between the identifier and all member fingerprints in the cluster, and stores it in the state mapping table on the chain. The mapping table structure is a mapping group list, each group contains a cluster identifier key and a member fingerprint list value, and each member fingerprint also points to the belonging cluster identifier in its registration mapping table.

[0064] The score weight value is determined as follows: the score weight of each edge represents the structural similarity strength between image fingerprints, which is calculated by the off-chain feature comparison service when generating the fine similarity matrix and attached to each fingerprint pair. The specific calculation method is as follows: the off-chain comparison service extracts the deep structure feature vector of any two image fingerprints, which is a 1024-dimensional real number array representing the feature expression of the image in multiple dimensions such as edge direction, texture density, and spatial distribution. For any two feature vectors, the system calculates the cosine similarity, as follows: Cosine similarity = dot product of feature vector 1 and feature vector 2 divided by the modulus of feature vector 1 multiplied by the modulus of feature vector 2, dot product is the multiplication and summation of corresponding dimension elements, and modulus is the square summation of all elements and then square root. The calculation result is floating between zero and one, where one represents complete identity and zero represents complete irrelevance.

[0065] After the calculation is completed, the value is taken as the score weight of the edge and attached to the corresponding position in the fine similarity matrix. For example, if the cosine similarity between image number five and image number nine is 0.945, this value is assigned to the edge between image number five and image number nine as the weight basis for subsequent on-chain union set merging and cluster root optimization. In actual implementation, to avoid floating point error propagation, the off-chain system performs fixed-point retention processing of the last four digits of the cosine similarity value and converts it to an unsigned fixed-length integer for storage, for example, multiplying 0.945 by 10,000 and rounding down to get 9,450, as the integer score value of the edge, which is sent to the on-chain for weight comparison and priority merging judgment. The higher the score weight, the stronger the similarity between the two fingerprint images. If there are multiple mergeable paths in multiple connected components, the smart contract preferentially selects the path with the highest score weight for merging, and sets the fingerprint with the highest score value as the cluster root node, thereby forming a cluster structure with the largest central representative.

[0066] The smart contract performs three-layer discrete wavelet decomposition on each complete fingerprint in the fingerprint cluster, calculates the energy of each layer of details, and aggregates them into multi-scale detail scores according to the preset weight; wherein the complete fingerprint refers to the feature of the digital artwork submitted by the revealing link without being disturbed. The system first performs image multi-scale decomposition on the fingerprint image, which is based on the discrete wavelet decomposition model. In the first decomposition, the image is decomposed into low-frequency information and high-frequency information in three directions, corresponding to horizontal, vertical, and diagonal details in the image; the second decomposition is performed again on the low-frequency part, obtaining the second layer of detail information; the third decomposition is performed accordingly, constructing a three-layer scale detail structure.

[0067] For each layer, the system extracts the high-frequency image region in three directions, squares the sum of the brightness values of all pixels in the region, and takes the sum as the detail energy of the layer. The sum of the detail energies of the three directions is the total detail energy of the layer. The total detail energies of the three layers are multiplied by preset weight coefficients, respectively. The first layer has the largest weight, the second layer has the second largest weight, and the third layer has the smallest weight. The weighted detail energies of the three layers are added to form a multi-scale detail score of the artwork, which represents the texture complexity and local variation intensity of the artwork at different visual scales.

[0068] The smart contract sorts the cluster members according to the multi-scale detail scores from high to low. If the difference between the highest score and the second highest score is below a threshold, a visual entropy comparison is added to determine the unique first place. After sorting, the system first determines whether the difference between the artwork with the highest score and the artwork with the second highest score is significant. If the score difference is below a set threshold, i.e., the score difference is not enough to make a significant distinction, the system activates the backup comparison mechanism and performs visual entropy analysis. Visual entropy represents the information density and complexity of image content in visual space. The calculation method is to count the frequency of each gray level in the image, then calculate the weighted sum of all gray level frequencies multiplied by their logarithms, and take the negative value as the entropy value of the image. The higher the entropy value, the more complex the image content, indicating that the artwork contains more rich detail structures and randomness, and has a stronger possibility of being original.

[0069] If the visual entropy result shows that the complexity of an artwork image is significantly higher than that of other candidates, the artwork is finally selected as the original first place. This strategy can prevent the original judgment from being ambiguous due to the too close detail scores.

[0070] The smart contract registers the first-place artwork as the original artwork and writes the derivative relationship for the remaining artworks in the same cluster in the state mapping table, recording the original identification, derivative level, and timestamp. The second-place artwork is determined, and its digital identity is written into the original registration book, including the cluster number, upload node, disclosure time, image fingerprint hash value, and current block height of the artwork. The remaining artworks in the cluster that rank lower than the original artwork are written into the derivative record table, establishing a bidirectional mapping relationship between each work and the original work. Each derivative record contains the identity number of the original work, the level number of the current derivative work, the relative position of the derivative level within the cluster, and the timestamp of completing the registration. This structure ensures that each artwork can be traced back to its source and forms a verifiable chain of originality and derivation in the system. The design goal of this mechanism is to ensure that in legal scenarios such as copyright identification, originality disputes, and responsibility determination, the system can provide a clear, stable, and tamper-proof original source identification mechanism and derivative tracking channel, while having sufficient detail discrimination ability to resist complex attack behaviors such as image fine-tuning, splicing forgery, and repeated registration, ensuring a fair and just authentication basis for digital rights of artworks.

[0071] To further optimize the implementation, the following is added: To improve the robustness and comparability of visual entropy under different image resolutions, proportional cropping or corner occlusion, the system introduces a partitioning process and redundant entropy removal mechanism before performing full-image gray entropy calculation. Specifically, the system divides the digital image into a number of fixed-size grid cells, and calculates the local entropy value of each cell based on the gray histogram. To avoid the dilution effect of large areas of blank regions on the global entropy value, the system automatically removes low-variation grids during the calculation process, with the removal criterion being that the standard deviation of pixel gray scale within the cell is below a pre-set minimum variation threshold. The final image visual entropy indicator is defined as the weighted average of the local entropy values of all valid cells, with the weights determined based on the proportion of non-zero pixels in the cell. This mechanism effectively improves the stability of entropy evaluation in the presence of scan artifacts, corner cropping and compression reconstruction scenarios, ensuring consistency and anti-bias ability in the originality determination between multiple versions of artworks.

[0072] When a registered original artwork is cancelled due to voluntary withdrawal, ownership withdrawal or failed authentication, the smart contract automatically triggers the inheritance ordering process for the cluster. The system re-executes the detail energy calculation and visual entropy evaluation for all members in the cluster, and generates a new ordering list based on the original multi-scale detail scoring model. If the difference between the score of the top artwork and the second highest score is greater than the set originality authentication threshold, the system confirms it as the new original work in the cluster; the remaining members inherit the original derivative level and automatically update their derivative path pointers. If the score difference is not sufficient to make a clear originality determination, the system marks the cluster as temporarily unauthentic, and waits for the re-evaluation process triggered by the addition of new works before making a decision. This inheritance ordering mechanism ensures the legal succession of originality relationships in the long-term evolution process, effectively avoiding the ownership gap or responsibility drift caused by the invalidation of original works.

[0073] When an artwork needs to update its registration information due to restoration, fine-tuning or style evolution, the system provides a fingerprint evolution version management mechanism. The registration node re-calculates the blinded fingerprint for the revised artwork image, and performs stability judgment based on the deep feature cosine similarity between the current fingerprint and the original fingerprint. If the similarity is higher than the version tolerance threshold, the system determines that the fingerprint belongs to the continuous version of the original work. The smart contract automatically generates a unique version identifier, which is calculated based on the original fingerprint hash, current timestamp and revision number, and has one-way irreversibility and verifiability. The system builds a version evolution chain in the state table, supporting the query of the original root node, evolution path and registration history corresponding to each version, ensuring that the artwork can maintain registration continuity and ownership integrity during content adjustment, while effectively preventing repeated registration and ownership drift behavior.

[0074] The above algorithms or formulas are all de-dimensioned to calculate their numerical values, and the results of the most recent real situation are obtained by collecting a large amount of data for software simulation. The preset parameters are set by a person skilled in the art according to the actual situation.

[0075] It should be understood that the size of the sequence number of each process described above in various embodiments of the present application does not mean the order of execution, and the execution order of each process should be determined according to its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0076] Those skilled in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. A person skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0077] Those skilled in the art can clearly understand that, for the convenience and brevity of the description, the specific working processes of the above-described devices and units can refer to the corresponding processes in the foregoing method embodiments, which will not be described here.

[0078] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1.A blockchain-based method for digital copyright registration of an artwork, characterized in that, The method comprises the following steps: The registration node calculates a blind fingerprint for the artwork to be registered, combines the blind fingerprint with a random salt value to generate a placeholder hash, and writes the placeholder hash into the blockchain together with a trusted timestamp to complete the registration of the placeholder; The smart contract determines a delay time according to a verifiable random function, opens a reveal window when the delay time expires, and the registration node submits the complete fingerprint corresponding to the placeholder hash, the random salt value and the artwork metadata within the reveal window; The smart contract calculates the similarity distribution between the revealed fingerprints in real time during the reveal window, and generates a congestion signal when the number of similar fingerprints exceeds a preset threshold; After the congestion signal is generated, the smart contract immediately reduces the dynamic similarity threshold and calls the off-chain multi-scale deep feature comparison algorithm to obtain a fine similarity evaluation result between the fingerprints; The smart contract merges the fingerprints with a similarity not lower than the dynamic similarity threshold into the same fingerprint cluster according to the fine similarity evaluation result, and generates a unique cluster identifier for each fingerprint cluster; At the end of the reveal window, the smart contract sorts the artworks in each fingerprint cluster according to the multi-scale detail score, registers the artwork with the highest detail score as the original artwork, and registers the other artworks in the same cluster as derivative artworks and establishes a correlation relationship between them. 2.The blockchain-based art digital copyright registration method of claim 1, wherein, The calculation of the blind fingerprint comprises the following steps: The registration node pre-processes the digital image of the artwork to be registered according to a fixed resolution and a uniform color space, and performs high-frequency noise suppression on the pixel matrix to eliminate the interference of scanning device noise on the stability of the fingerprint; The registration node constructs a multi-scale discrete cosine transform feature vector based on the pre-processed pixel matrix, and inputs the feature vector into a one-way key hash function based on the artwork private key to generate a privacy-protected feature string; The registration node applies an invertible quantization perturbation algorithm to the privacy-protected feature string to reduce the probability of single-pixel restoration while maintaining global perceptual consistency, thereby obtaining a blind fingerprint that can be publicly verified and is difficult to reverse. 3.The blockchain-based art digital copyright registration method of claim 1, wherein, The generation and writing of the placeholder hash comprise the following steps: The registration node randomly generates an encrypted random salt value matching the length of the blind fingerprint, and concatenates the blind fingerprint and the random salt value in a predetermined order to form a raw string to be hashed; The registration node calculates a fixed-length hash of the raw string to be hashed using the Secure Hash Algorithm version 3 to obtain the placeholder hash, and simultaneously requests a trusted timestamp signature from a distributed timestamp authority; The registration node writes the placeholder hash, the random salt value and the trusted timestamp as an indivisible data packet into the blockchain transaction payload and broadcasts it to the network, and the placeholder record is tamper-proofed as soon as the first confirmation block is generated. 4.The blockchain-based art digital copyright registration method of claim 1, wherein, The process of determining the delay time according to the verifiable random function comprises the following steps: The smart contract calls the block header random number, the node key commitment value and the current network difficulty to form an input seed after receiving the placeholder hash, which provides multiple source entropies for the verifiable random function; The smart contract inputs the input seed into the verifiable random function based on the Elliptic Curve Digital Signature Algorithm, outputs a pseudo-random number that is unpredictable for any single node and can be publicly verified, and simultaneously generates a zero-knowledge proof; The smart contract linearly maps the preset minimum delay and maximum delay interval according to the pseudo-random number to obtain a delay time, writes the delay time and zero-knowledge proof into the blockchain for external node verification, and automatically opens the reveal window after the delay expires. 5.The blockchain-based art digital copyright registration method of claim 4, wherein, The generation process of the input seed includes the following steps: The smart contract reads the random number field of the latest block header that has completed consensus immediately after writing the placeholder hash, and broadcasts the random number through the event log; The block verifier calculates the node key commitment value through the one-time public key before entering the next round of block generation, and submits it to the on-chain key commitment registration contract together with its digital signature. The smart contract verifies and immediately publishes the commitment value after verification; The smart contract continuously retrieves the difficulty fields of several latest blocks within a preset window, calculates the sliding average to form the current network difficulty, and synchronously writes it into the state storage together with the timestamp. 6.The blockchain-based art digital copyright registration method of claim 5, wherein, The process of generating a congestion signal includes the following steps: The smart contract performs discrete cosine transform on each pair of fingerprints in the same block after four levels of scaling, extracts the bit string and calculates the Hamming distance to generate a multi-scale perception distance list; Take the last n reveal records as a window, mark the fingerprints with a distance lower than the current similarity threshold as similar, and count the similar count in the window; Calculate the average of the similar count of the first m windows, multiply the average by the inverse of the product of the preset safety factor to get the new threshold after dynamic tightening; Trigger the congestion signal when the current window similar count exceeds the new threshold. 7.The blockchain-based art digital copyright registration method of claim 6, wherein, The process of reducing the dynamic similarity threshold and obtaining the fine similarity evaluation result between fingerprints after generating the congestion signal includes the following steps: The smart contract multiplies the preset similarity threshold by a preset contraction coefficient to generate a contraction threshold after the congestion signal is triggered, and immediately updates the global state; The smart contract synchronously broadcasts the list of fingerprints to be evaluated and the contraction threshold to the off-chain feature comparison service, along with a random challenge value; The off-chain comparison service extracts deep features through a multi-scale convolutional network after receiving the instruction, calculates the cosine distance between fingerprints, and returns a fine similarity matrix with zero-knowledge proof; The smart contract verifies the zero-knowledge proof and writes the fine similarity matrix into the fingerprint cluster mapping table for subsequent original right confirmation calls. 8.The blockchain-based art digital copyright registration method of claim 7, wherein, Merging fingerprints with a similarity not lower than the dynamic similarity threshold into the same fingerprint cluster means: After receiving the fine similarity matrix, the smart contract maps the fingerprints with a similarity higher than the dynamic similarity threshold, i.e. the contraction threshold, to edges in the graph structure, and constructs a similarity graph in real time; The smart contract calls the on-chain and search set algorithm on the similarity graph to merge connected nodes from bottom to top. After each merge, it automatically selects the fingerprint with the highest weight as the cluster root of the connected component according to the edge weight; The smart contract calculates the secure hash of the current block number and random entropy seed for each cluster root fingerprint, generates a unique cluster identifier, and stores the cluster identifier and all fingerprints in the cluster in the state table in a bidirectional mapping. 9.The blockchain-based art digital copyright registration method of claim 8, wherein, Registering the artwork with the highest detail score as the original artwork means: The smart contract performs three-layer discrete wavelet decomposition on each complete fingerprint in each fingerprint cluster, calculates the energy of each layer of details, and aggregates them into a multi-scale detail score according to the preset weight; The smart contract will cluster members according to the multi-scale detail score from high to low, and if the difference between the highest score and the second highest score is lower than the threshold, the visual entropy comparison will be added to determine the unique first place; The smart contract will register the first place artwork as an original artwork, and write the derivative relationship for the remaining artworks in the same cluster in the state mapping table, recording the original identification, derivative level and timestamp.

Images