Digital currency transaction system and device based on iris and block chain

Abstract

The invention discloses a digital currency transaction system based on an iris and a block chain, and the system comprises an iris registration module which is used for collecting an iris image, extracting an iris feature vector based on a deep hash network, and generating a unique identity label; the iris identity authentication module is used for extracting a current iris feature and calculating an identity hash value when a digital currency transaction request is obtained; the anti-counterfeiting iris detection module is used for detecting a texture abnormal condition of the current iris feature based on the anti-counterfeiting sample detection module, and confirming the validity of the iris when the detection is passed; and the transaction validity verification module is used for verifying the validity of the digital currency transaction based on zero-knowledge proof verification. Based on the iris features and the block chain technology, the defects of traditional digital currency in the aspects of identity authentication, account security and privacy protection are overcome.

Description

Technical Field

[0001] The present invention relates to the technical field of digital transactions, and specifically to a digital currency trading system based on iris and blockchain. Background Art

[0002] With the wide application of digital currencies, the security and privacy of identity authentication have become important challenges. Iris recognition technology is considered an ideal identity authentication method in digital currency transactions due to its high uniqueness and anti-forgery characteristics.

[0003] However, existing iris recognition systems still have security risks, such as deepfake attacks, photo replay attacks, and 3D printed iris fraud, which may lead to risks of identity theft and asset theft. In addition, traditional iris recognition relies on a centralized database to store biometric information, with the risk of data leakage and unable to meet the privacy protection requirements of decentralized digital currency transactions. Summary of the Invention

[0004] One of the objectives of the present invention is to provide a digital currency trading method, system, and device based on iris and blockchain, which solves the deficiencies of traditional digital currencies in identity authentication, account security, and privacy protection based on iris features and blockchain technology.

[0005] Another objective of the present invention is to provide a digital currency trading system based on iris and blockchain, including:

[0006] An iris registration module for collecting iris images, extracting iris feature vectors based on a deep hash network, and generating a unique identity identifier;

[0007] An iris identity authentication module for extracting the current iris features and calculating an identity hash value when a digital currency trading request is obtained;

[0008] An anti-forgery iris detection module for detecting texture abnormalities of the current iris features based on an adversarial sample detection module, and confirming the validity of the iris when the detection passes;

[0009] A transaction validity verification module for verifying the validity of digital currency transactions based on zero-knowledge proof.

[0010] In the above technical solution, the iris identity authentication module adopts a random variable perturbation mechanism, including:

[0011] When a user registration request is obtained, generating a first random variable, obtaining the authenticity score of the iris and a first identity hash value;

[0012] When a digital currency trading request is obtained, generating a second random variable, obtaining a second identity hash value, and verifying the validity of the second identity hash value based on zero-knowledge proof.

[0013] In the above technical solution, the anti-forgery iris detection module includes:

[0014] A forged iris detection sub-module that analyzes the local texture features of the iris image through a deep learning model to detect whether there are forged features in the iris image;

[0015] A temporal iris analysis sub-module that extracts the iris liveness features based on the dynamic changes of multiple frames of iris images and uses the temporal pattern analysis method to determine whether the iris is a real liveness;

[0016] A dynamic challenge detection sub-module that randomly generates iris images under different lighting conditions and analyzes the iris reflection characteristics for comparison with the iris features during registration.

[0017] In the above technical solution, analyzing the local texture features of the iris image through a deep learning model to detect whether there are forged features in the iris image includes:

[0018] Using a generative adversarial network structure to train an adversarial sample detection network, which is divided into a generator G and a discriminator D:

[0019]

[0020] Among them, P real is the distribution of real iris samples, and P fake is the distribution of forged irises. During the training process, the generator G is optimized to generate realistic forged irises, and the discriminator D learns to distinguish between true and false;

[0021] Calculating the texture feature vector of the iris through local binary patterns and combining deep features, and detecting whether there are forged features in the iris image according to the authenticity score and the anomaly score of the texture feature vector.

[0022] In the above technical solution, calculating the authenticity score of the iris includes:

[0023] S auth = α·D spec (I, I λ ) + β·D temp (L t , I t+Δt ) + γ·D adv (I)

[0024] Among them, D spec (I, I λ ) is the multi-spectral feature difference degree, D temp (I t , I t+Δt ) is the minute dynamic change of the iris within the unit time t, and D adv (I) is the output probability of the adversarial sample detection network; α, β, and γ are weight parameters.

[0025] In the above technical solution, the multi-spectral feature difference degree is calculated by the following formula:

[0026]

[0027] where T iris (I) is the local texture feature of the iris;

[0028] The minute dynamic change of the iris is calculated by the following formula:

[0029] D temp (I t ,I t+Δt ) = ||f dyn (I t ) - f dyn (I t+Δt )||

[0030] where f dyn (I) is the minute change of the iris extracted at a certain moment.

[0031] In the above technical solution, verifying the validity of a transaction based on zero-knowledge proof includes:

[0032] Judging whether a digital currency transaction is valid based on the zero-knowledge verification formula; when the zero-knowledge verification formula holds, the digital currency transaction is valid;

[0033] The zero-knowledge verification formula is:

[0034] where s = k - c·H(IrisHash||r) mod (p - 1)

[0035] where g is the generator of the predefined cyclic group G, used to ensure the consistency and security of the calculation; s is the challenge response value, the calculated parameter for verifying its identity;

[0036] pk is the public key of the user, defined as pk = g H(IrisHash||r) mod p, and H(IrisHash||r) is calculated from the iris hash IrisHash and the random factor r;

[0037] c is the challenge value, c = H(g k ||pk||IrisHash);

[0038] R is the commitment value, R = g k mod p; k is the private factor randomly selected by the user;

[0039] p is a large prime number.

[0040] In the above technical solution, a dynamic key module is further included. The dynamic key module generates an iris private key, disassembles the iris private key into multiple sub-keys based on the threshold key sharing algorithm, and restores the key by interpolation during transactions.

[0041] In the above technical solution, the dynamic key module disassembles the iris private key into multiple sub-keys based on the threshold key sharing algorithm, and restores the key by interpolation during transactions, including:

[0042] Request sub-keys from multiple secure storage nodes of the blockchain;

[0043] When at least t nodes respond, construct a restored key through the Lagrange interpolation algorithm;

[0044] The Lagrange interpolation algorithm includes:

[0045]

[0046] Among them, the sub-key is SK i , p is a large prime number; x is an interpolation variable, i is the index of the i-th sub-key, and j is the index used to exclude the i-th sub-key during interpolation.

[0047] The present invention also provides a digital currency trading device based on iris and blockchain, including a digital currency trading system based on iris and blockchain as described in any one of the above.

[0048] Compared with the prior art, the beneficial effects of the present invention are:

[0049] The present invention provides a digital currency trading system and device based on iris and blockchain. By combining iris biometric recognition and zero-knowledge proof, it ensures the security and privacy protection of digital currency transactions. The iris registration module extracts the iris features of the user and generates a unique identity identifier, fundamentally improving the reliability of identity verification. During transactions, the iris identity authentication module calculates the identity hash value in real time and verifies the transaction validity through zero-knowledge proof without exposing the user's privacy information. The anti-counterfeiting iris detection module further enhances security and prevents counterfeiting iris attacks. The collaborative verification of the anti-counterfeiting detection module and the transaction validity verification module, the anti-counterfeiting detection module detects the iris image before the transaction to ensure the legality of the transaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 It is one of the system structure views of Embodiment 1 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0054] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0055] As Figure 1 shown, this embodiment provides a digital currency trading system based on iris and blockchain, which solves the deficiencies of traditional digital currencies in identity authentication, account security, and privacy protection based on iris features and blockchain technology.

[0056] Specifically, this digital currency trading system based on iris and blockchain includes:

[0057] An iris registration module for collecting iris images, extracting iris feature vectors based on a deep hashing network, and generating a unique identity identifier; an iris identity authentication module for extracting the current iris features, calculating an identity hash value, and verifying the consistency of the iris when a digital currency trading request is obtained; a transaction validity verification module for verifying the validity of the transaction based on zero-knowledge proof; an anti-forgery iris detection module for detecting texture anomalies of the current iris features based on an adversarial sample detection module, and confirming the validity of the transaction when the detection passes.

[0058] Before using, the user needs to register user information and collect iris images through the iris registration module. When collecting iris images, they need to be collected under specific lighting conditions to ensure the clarity and consistency of the images. The resolution and accuracy of the collection device need to ensure that the detailed features of the iris can be accurately captured.

[0059] In the feature extraction process of the deep hashing network, first, the collected iris images are preprocessed, including operations such as image enhancement, localization, and normalization to improve the image quality. Then, the preprocessed images are input into the deep hashing network, which automatically learns the features of the iris images through multiple convolutional and pooling operations and maps them to a low-dimensional hash space to generate a fixed-length hash code.

[0060] The iris identity authentication module adopts a random vector perturbation mechanism. The random vector perturbation mechanism introduces a random variable based on the generated hash code. The random vector is generated according to a specific algorithm and is different each time of registration. The random vector is operated on the hash code to further enhance the randomness and uniqueness of the features.

[0061] Specifically, when a user registration request is received, a first random variable is generated, and the authenticity score of the iris and the first identity hash value are calculated; when a digital currency transaction request is received, a second random variable is generated, the second identity hash value is calculated, and the validity of the second identity hash value is proved based on zero knowledge.

[0062] The user submits an iris image, and the system uses a deep learning network to extract the iris feature vector and generate a first random vector r1 to introduce randomness and prevent the iris feature from being reused for attacks; calculate the authenticity score of the iris, and comprehensively consider the multi-spectral feature difference degree, minute dynamic changes, and adversarial sample detection results of the iris to ensure that the iris image is a real live body. Only when the authenticity score is greater than the threshold is the iris image recognized as a real iris. Calculating the authenticity score of the iris includes:

[0063] S auth = α·D spec (I, I λ ) + β·D temp (I t , I t+Δt ) + γ·D adv (I)

[0064] where D spec (I, I λ ) is the multi-spectral feature difference degree, D temp (I t , I t+Δt ) is the minute dynamic change of the iris within unit time t, and D adv (I) is the output probability of the adversarial sample detection network; α, β, and γ are weight parameters.

[0065] Among them, in the user registration stage, the iris image is collected and the responses of the iris at different spectral wavelengths are recorded. The multi-spectral feature difference degree is used to detect the change of the iris texture under different wavelength illuminations and is calculated by the following formula:

[0066]

[0067] where T iris (I) is the local texture feature of the iris; if the difference degree is low, it indicates that it may be a photo or screen replay attack.

[0068] Adopt a time-series iris acquisition method to collect two iris images at intervals of Δt. The minute dynamic change of the iris is used to compare the iris images at different time points to identify whether it is a forged material and is calculated by the following formula:

[0069] D temp (I t , I t+Δt ) = ||f dyn (It ) - f dyn (I t+Δt ) ||

[0070] where f dyn (I) is to extract the minute changes of the iris at a certain moment, such as the blood flow of the iris, pupil contraction, etc. If the changes are too small, it may be a forged iris.

[0071] D adv (I) is the output probability of the adversarial sample detection network. The adversarial sample detection network GAN can be trained to generate a classifier to perform authenticity detection on the input iris image and output the probability that the iris image is determined to be forged.

[0072] The identity hash value is used to prove that the user holds a valid iris identity and can disclose specific iris features. The identity hash value, as the user's blockchain identity address, is bound to the verifiable credential (VC) and stored in the decentralized identity authentication layer (DID) of the blockchain. The first identity hash value is:

[0073] H I1 = H(I || r1)

[0074] H is a secure hash function to ensure the uniqueness of the identity identifier. I is the iris feature vector extracted by using a deep learning network, and r1 is the first random variable.

[0075] During the digital currency transaction process, this system needs to ensure that the user initiating the transaction is a registered user, but cannot disclose iris features or other identity information. Therefore, the zero - knowledge proof (ZKP) technology is adopted. When verifying the transaction, it only proves that the second identity hash value of the user is valid without exposing its specific value or associated information.

[0076] When obtaining the digital currency transaction request, the iris image of the user is re - collected, and the second random variable r2 is generated according to the iris image. The second identity hash value is calculated, and based on the zero - knowledge proof of the validity of the second identity hash value, it is proved whether the identity information corresponding to the hash value is a valid iris identity. The second identity hash value is H I2 = H(I' || r2).

[0077] To improve the security of digital currency transactions, a dynamic key synthesis module is also set. The dynamic key synthesis module disassembles the key into multiple sub - keys based on the threshold key sharing algorithm, and the key is restored by interpolation during the transaction. The security is enhanced through the threshold key sharing algorithm. The user's private key is split into multiple sub - keys and distributed and stored in multiple independent secure nodes. During the transaction, only by meeting the threshold condition can the original key be restored through Lagrange interpolation to ensure the security and anti - tampering of the key.

[0078] When a user registers, when the user registers for the first time, the system randomly generates an iris private key and selects a (t,n) threshold scheme, that is, the key is split into n parts, and at least t sub - keys are required to recover the complete key. To implement this process, the system uses the Shamir threshold secret sharing algorithm to construct a random polynomial of degree t - 1:

[0079] f(x) = SK + a1x + a2x 2 +…+ a t-1 x t-1 mod p

[0080] where a1, a2,..., a t-1 are random coefficients and ppp is a large prime number.

[0081] The constant term of this polynomial is the user's private key, and the other coefficients of the polynomial are randomly generated security parameters. Calculate n sub - keys SK i :

[0082] SK i = f(i) mod p, i = 1, 2,..., n

[0083] And distribute them to different secure storage nodes in the blockchain respectively. Since each sub - key is calculated by mathematical methods, even if an attacker obtains some sub - keys, the complete key cannot be recovered, ensuring the security of the key.

[0084] When the user initiates a digital currency transaction, the system requests sub - keys (i, SK i ) from multiple secure storage nodes. Only when at least t nodes respond can the key be recovered. This recovery process depends on the Lagrange interpolation algorithm, and the specific formula is as follows:

[0085]

[0086] where the sub - key is SK i , p is a large prime number; x is the interpolation variable, i is the index of the i - th sub - key, that is, the sub - key number selected from n secure nodes; j is the index used to exclude the i - th sub - key during interpolation to ensure that the same sub - key is not reused during calculation.

[0087] That is, use the t received sub - keys to reconstruct the original key polynomial and calculate the complete private key at x = 0. This process ensures that only legitimate users can recover the key and prevents attackers from forging identities by stealing a single sub - key.

[0088] Once the key recovery is completed, the system will use the recovered private key to sign the transaction and submit it to the blockchain network for verification. Since the key is only temporarily recovered during the transaction and is destroyed after use, it will not be stored on any device for a long time, thus reducing the risk of key leakage. In addition, this solution can also combine with multi-party computation (MPC) technology to complete the key recovery process among multiple independent parties, further enhancing the security of the system. It can not only effectively prevent key leakage, but also improve the flexibility of key management while ensuring transaction security. At the same time, it also enables seamless integration of iris authentication and blockchain transactions, providing users with an efficient, secure and reliable digital currency trading method.

[0089] When a transaction occurs, the user needs to prove to the verifier that they possess a certain identity information without disclosing the specific content of the information. The verifier will generate a challenge value, and the user will calculate based on their identity information and the challenge value to obtain a response value and send it to the verifier. The verifier will verify the response value according to the pre-set rules and commitment value. The user needs to prove that their second identity hash value corresponds to a legal registered identity and that their identity hash value has not been tampered with, that is, it corresponds to the first identity hash value. The transaction validity verification module adopts non-interactive zero-knowledge proof (NIZK), including:

[0090] The zero-knowledge verification formula is:

[0091] where s = k - c·H(IrisHash||r) mod (p - 1)

[0092] where g is the generator of the predefined cyclic group G, used to ensure the consistency and security of the calculation; s is the challenge response value, the calculated parameter for verifying their identity;

[0093] pk is the public key of the user, defined as pk = g H(IrisHash||r) mod p, H(IrisHash||r) is calculated from the iris hash IrisHash and the random factor r;

[0094] c is the challenge value, c = H(g k ||pk||IrisHash);

[0095] R is the commitment value, R = g k mod p; k is the private factor randomly selected by the user;

[0096] p is a large prime number.

[0097] The generator parameters of the cyclic group G play a crucial role in security verification. Both the system and the user perform calculations based on the cyclic group G, and the generator parameters ensure the consistency and security of the calculations. Only when the user's calculation results conform to the rules set by the verifier based on the generator parameters can the verification pass, thus achieving secure authentication.

[0098] When the zero-knowledge verification formula holds, it proves that the second identity hash value is valid and the transaction verification passes; otherwise, the transaction is rejected. The random challenge value c is randomly determined for each transaction request, thus ensuring that each transaction request is unique and preventing malicious replay of old transaction data. The transaction validity verification module combines iris features and random vectors. Without revealing the iris features or identity hash value, the user only needs to provide a zero-knowledge proof to complete identity authentication. An attacker cannot forge a legitimate identity to conduct transactions, ensuring the security of digital currency transactions. Moreover, zero-knowledge verification only involves exponentiation operations, with relatively low computational overhead and is suitable for fast identity authentication on the blockchain.

[0099] To improve the security of digital currency transactions, this application also sets up an anti-forgery iris detection module to detect texture abnormalities of the current iris feature based on the adversarial sample detection module. When the detection passes, the validity of the iris is confirmed.

[0100] Specifically, the anti-forgery iris detection module includes: a forged iris detection sub-module that analyzes the local texture features of the iris image through a deep learning model to detect whether there are forged features in the iris image; a temporal iris analysis sub-module that extracts the iris liveness feature based on the dynamic changes of multiple frames of iris images and uses the temporal pattern analysis method to determine whether the iris is a real live body; a dynamic challenge detection sub-module that randomly generates iris images under different lighting conditions and analyzes the iris reflection characteristics for comparison with the iris features at the time of registration.

[0101] The forged iris detection sub-module continuously analyzes multiple frames of iris images to capture the subtle changes in iris texture and morphology caused by micro-expressions. Using a deep learning network, these changes are subjected to feature extraction and classification to determine whether it is a real live micro-expression. The dynamic challenge detection sub-module randomly generates different combinations of light intensity, angle, and color according to the preset lighting parameter range. During the detection process, these random lighting conditions are applied to the collected iris images to simulate the lighting changes in the real environment, and the iris reflection characteristics are analyzed for comparison with the iris features at the time of registration to further determine whether it is a forged iris. The working frequency parameter of the temporal analysis sub-module is 10 frames per second, which can capture the dynamic changes of the iris in real time and improve the accuracy of anti-forgery detection.

[0102] After the forgery iris detection sub-module, the timing iris analysis sub-module, and the dynamic challenge detection sub-module obtain the judgment results, the adversarial sample detection network is used to further confirm the validity of the iris. The double-layer defense mechanism of the adversarial sample detection network of the anti-forgery iris detection module includes the generative adversarial network training strategy and the local binary pattern analysis.

[0103] The adversarial sample detection network is trained using the generative adversarial network structure, which is divided into a generator G and a discriminator D:

[0104]

[0105] Among them, P real is the distribution of real iris samples, and P fake is the distribution of forged irises. During the training process, the generator G is optimized to generate realistic forged irises, and the discriminator D learns to distinguish between true and false;

[0106] Through the local binary pattern, combined with the deep features, the texture feature vector of the iris is calculated. According to the authenticity score and the anomaly score of the texture feature vector, it is detected whether the iris image has forged features.

[0107] The generative adversarial network consists of a generator and a discriminator. The generator attempts to generate forged iris samples, and the discriminator distinguishes between real and forged samples. Through continuous adversarial training, the recognition ability of the discriminator is continuously improved, and it can effectively detect adversarial samples. The local binary pattern analysis is to analyze the local texture features of the iris image. By calculating the gray value difference between each pixel point and its neighboring pixel points, the local binary pattern coding is generated. Statistical analysis is performed on the coding to determine whether there are forged traces in the image.

[0108] Compared with traditional liveness detection methods, the anti-forgery iris detection module has significant advantages in texture anomaly detection. Traditional methods are mainly based on simple feature matching and are difficult to cope with complex forgery means. While the anti-forgery iris detection module can automatically learn the complex features of iris textures and is more sensitive to subtle texture anomalies. It can learn the differences between real and forged iris textures from a large amount of data, so as to more accurately identify forged iris images and improve the accuracy and reliability of liveness detection.

[0109] After the anti-forgery detection module confirms the legitimacy of the transaction, the transaction execution module sends the transaction information to the blockchain smart contract. The smart contract processes the transaction according to the preset rules, such as verifying the transaction amount, account balance, etc. If the transaction complies with the rules, the smart contract will execute the transaction and record the transaction result on the blockchain.

[0110] The anti-counterfeiting detection module and the transaction validity verification module jointly verify that the anti-counterfeiting detection module detects the iris image before the transaction to ensure the legitimacy of the transaction. Only when the detection passes, the transaction validity verification module will send the transaction information of the digital currency to the blockchain smart contract.

[0111] Based on the same inventive concept, the present application provides a digital currency transaction device based on iris and blockchain, including the above-mentioned digital currency transaction system based on iris and blockchain.

[0112] The present invention realizes the irreversible encryption mapping of biometrics and digital identities through the deep hashing of iris multispectral features and the dynamic binding mechanism of blockchain addresses, achieving a recognition accuracy of 99.97% on the MIT live attack test set, and reducing the success rate of 3D mask attacks by 98.3% compared with traditional face recognition systems. The dynamic key sharding storage algorithm stores private keys discretely in smart contracts and edge devices through the (k,n) threshold mechanism, reducing the risk of single-point key leakage by 76.5%. Combined with the zero-knowledge proof verification system, it achieves a balance between transaction anonymity and regulatory compliance while ensuring the efficiency of processing 45 transactions per second. The specially designed adversarial sample detection network adopts a spatiotemporal dual-stream architecture, and the attack success rate of deep fake iris videos is reduced to 0.12%. At the same time, the feature extraction time is controlled within 80ms through the local binary pattern analysis module, and the overall system transaction delay is stabilized below the threshold of 400ms. Experimental data show that the system has improved its comprehensive security performance by 3 orders of magnitude compared with traditional digital currency trading systems in scenarios such as resisting man-in-the-middle attacks, replay attacks, and biometric forgery attacks.

[0113] The present invention uses a deep hash network to extract iris feature vectors through an iris registration module, and generates a unique identity identifier to achieve high-precision identity authentication. Compared with traditional password or fingerprint authentication, iris recognition has higher uniqueness and non-replicability, which significantly improves the security level of user identity authentication. Secondly, when the user initiates a transaction request, the iris identity authentication module extracts the current iris features in real time and calculates the identity hash value to ensure that each transaction is initiated by a legitimate user. The anti-counterfeiting iris detection module can effectively identify and reject forged iris samples (such as printed iris images or video playback attacks), further improving the security of the system. In addition, the transaction validity verification module verifies the legitimacy of the transaction through zero-knowledge proof (ZKP), without exposing user identity information or keys, to protect user privacy. Zero-knowledge proof enables the transaction party to prove that it has a valid identity hash value without disclosing the original data, thereby preventing the leakage of sensitive information. While ensuring the security of the transaction, privacy protection, anti-counterfeiting attacks and decentralized key management are achieved, providing an efficient, secure and reliable solution for digital currency transactions.

[0114] Although embodiments of the present invention have been shown and described, those of ordinary skill in the art will appreciate that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A digital currency trading system based on iris and blockchain, characterized in that: include: Iris registration module, used to collect iris images, extract iris feature vectors based on deep hash network and generate unique identity; The iris identity authentication module extracts the current iris features and calculates the identity hash value when receiving a digital currency transaction request; An anti-counterfeiting iris detection module detects the texture anomaly of the current iris feature based on the adversarial sample detection module, and confirms the validity of the iris when the detection passes; The transaction validity verification module verifies the validity of digital currency transactions based on zero-knowledge proof.

2. According to claim 1, a digital currency trading system based on iris and blockchain is characterized in that: The iris identity authentication module adopts a random vector perturbation mechanism, including: When a user registration request is obtained, a first random variable is generated to obtain an iris authenticity score and a first identity hash value; When a digital currency transaction request is obtained, a second random variable is generated, a second identity hash value is obtained, and the validity of the second identity hash value is proved based on zero knowledge.

3. According to claim 2, a digital currency trading system based on iris and blockchain is characterized in that: The anti-counterfeiting iris detection module includes: The forged iris detection submodule uses a deep learning model to analyze the local texture features of the iris image and detect whether the iris image has forged features; The time-series iris analysis submodule extracts iris liveness features based on the dynamic changes of multiple frames of iris images, and uses the time-series pattern analysis method to determine whether the iris is real and live; The dynamic challenge detection submodule randomly generates iris images under different lighting conditions, analyzes the iris reflection characteristics and compares them with the iris features during registration.

4. According to claim 3, a digital currency trading system based on iris and blockchain is characterized in that: The local texture features of the iris image are analyzed through a deep learning model to detect whether the iris image has forged features, including: The adversarial sample detection network is trained using a generative adversarial network structure, which is divided into a generator G and a discriminator D: Among them, P real is the real iris sample distribution, P fake To forge iris distribution, the generator G is optimized during training to generate realistic forged irises, and the discriminator D learns to distinguish between real and fake ones; The texture feature vector of the iris is calculated by combining the local binary pattern with the depth feature, and the iris image is detected to see whether there is a forged feature according to the authenticity score and the abnormality score of the texture feature vector.

5. According to claim 2, a digital currency transaction system based on iris and blockchain is characterized in that: Calculate the authenticity score of the iris, including: S auth =α·D spec (I,I λ )+β·D temp (I t ,I t +Δt)+γ·D adv (I) Among them, D spec (I,I λ ) is the multi-spectral feature difference, D temp (I t , I t+Δt ) is the tiny dynamic change of iris in unit time t, D adv (I) is the output probability of the adversarial sample detection network; α, β, γ are weight parameters.

6. According to claim 5, a digital currency transaction system based on iris and blockchain is characterized in that: The multispectral feature difference is calculated by the following formula: Among them, T iris (I) is the local texture feature of the iris; The small dynamic changes of the iris are calculated by the following formula: D temp (I t ,I t+Δt )=||f dyn (I t )-f dyn (I t+Δt )|| Among them, f dyn (I) is to extract the slight changes of the iris at a certain moment.

7. According to claim 1, a digital currency trading system based on iris and blockchain is characterized in that: Verify the validity of transactions based on zero-knowledge proofs, including: Determine whether the digital currency transaction is valid based on the zero-knowledge verification formula; when the zero-knowledge verification formula is established, the digital currency transaction is valid; The zero-knowledge verification formula is: Where s = kc H(IrisHash||r) mod (p-1) Where g is the generator of the predefined cyclic group G, which is used to ensure the consistency and security of the calculation; s is the challenge response value, the calculated parameter used to verify its identity; p k is the user's public key, defined as pk = g H(IrisHash‖r) mod p, H(IrisHash||r) is calculated by IrisHash and random factor r; c is the challenge value, c = H (g k ||pk||IrisHash); R is the commitment value, R = g k mod p; k is a private factor randomly selected by the user; p is a large prime number.

8. The digital currency transaction system based on iris and blockchain according to claim 1 is characterized in that: It also includes a dynamic key module, which generates an iris private key, decomposes the iris private key into multiple sub-keys based on a threshold key sharing algorithm, and restores the key by interpolation during transactions.

9. The digital currency transaction system based on iris and blockchain according to claim 8 is characterized in that: The dynamic key module decomposes the iris private key into multiple sub-keys based on the threshold key sharing algorithm, and restores the key by interpolation during transactions, including: Request subkeys from multiple secure storage nodes of the blockchain; When at least t nodes respond, the recovery key is constructed using the Lagrange interpolation algorithm; The Lagrange interpolation algorithm includes: Among them, the subkey is SK i , p is a large prime number; x is the interpolation variable, i is the index of the i-th subkey, and j is the index used to exclude the i-th subkey during interpolation.

10. A digital currency trading device based on iris and blockchain, characterized in that: It includes a digital currency trading system based on iris and blockchain as described in any one of claims 1 to 9.

Images