Color image encryption method based on finger vein secret key and off-axis digital holography

Abstract

The invention discloses a color image encryption method based on a finger vein secret key and off-axis digital holography. The color image encryption method comprises the encryption steps that firstly, red, green and blue channel data of a color plaintext image are separated; performing encryption based on DNA coding on each channel by using the finger vein data of the encrypted user; thirdly, embedding the DNA coding encryption result into the host image by using a discrete wavelet transform-based technology to generate an encrypted watermark image; and finally, performing encryption based on a Gyrator transform domain and off-axis digital holography on the encrypted watermark image by using the finger vein data and the digital key to generate a ciphertext hologram. And a decryption step: firstly, carrying out identity verification on a decryption user by using a finger vein authentication technology based on the convolutional neural network, decrypting the ciphertext hologram by using the finger vein data stored in the database after the authentication is successful, and recovering a plaintext image. The encryption method provided by the invention has the advantages of high security, finger vein key robustness, digital key sensitivity, large key space and high decryption quality.

Description

Technical Field

[0001] The present invention relates to the field of information security technology, and in particular to a color image encryption method based on finger vein key and off-axis digital holography. Background Art

[0002] In order to prevent color images from being modified and stolen during communication transmission, people pay more and more attention to the use of encryption technology to protect the information of color images. Among the many color image encryption technologies, optical color image encryption technology based on optical principles has the characteristics of fast encryption speed, high security, high parallelism, rich key types, and the ability to process large-capacity images, which has attracted widespread attention from domestic and foreign research groups. Therefore, since the image encryption method based on double random phase encoding (DRPE) proposed by Refregier and Javidi in 1995, color image encryption methods based on DRPE technology have been continuously proposed by various research groups and scholars. For example, Sui et al. proposed an optical color image encryption method based on Gyrator transform domain DRPE and Arnold transform. Yildirim et al. proposed an optical color image encryption method based on Fourier transform domain DRPE and DNA encoding. Later, related technologies such as interference, diffraction imaging, single pixel imaging, and computer holography were gradually introduced into optical color image encryption technology, and a variety of novel optical color image encryption schemes emerged. For example, Tay et al. proposed an optical color image encryption method based on interference. Chen et al. proposed an optical color image encryption method based on Fresnel domain coherent diffraction imaging. Qu et al. proposed an optical color image encryption method based on single pixel imaging and Arnold transform. Liu et al. proposed an optical color image encryption method based on computer-generated holography and chaotic function.

[0003] However, the current optical color image encryption technology often has the problem of high key complexity, especially the random mask key is usually difficult for users to remember or carry, and there is also the problem of lack of connection between the key and the legitimate user. Once the key is lost or stolen, unauthorized attackers can obtain sensitive security information in the security system in an illegal way, thus seriously threatening the system security. Biometric keys are an effective means to solve the above problems, because biometric keys are convenient, unique, and will not be lost, forgotten, or stolen. In recent years, many research groups at home and abroad have made useful research on optical color image encryption methods based on biometric keys. For example, Su et al. proposed a color image encryption method based on fingerprint key and coaxial digital holography. Wang et al. proposed a color image encryption method based on fingerprint key and diffraction imaging. Later, Zhang et al. proposed a color image encryption method based on palmprint key and Fresnel transform domain DRPE. In addition, Abuturab et al. proposed a color image encryption method based on iris key and unequal spectrum decomposition (USD).

[0004] However, the fingerprint key and palm print key used in the above encryption method are both traditional biometric keys that require contact sensors for authentication, which have the disadvantages of easy sensor wear and user cross-contamination. In addition, since the acquisition of iris information is too sensitive to ambient light, the iris information of legitimate users is difficult to be correctly acquired by the authentication system under complex lighting conditions, resulting in decryption failure. Compared with fingerprint keys, palm print keys and iris keys, finger vein keys have the natural advantages of non-contact collection, easy acquisition, and strong anti-interference ability, and are an effective way to solve the above problems. Therefore, the present invention proposes a color image encryption method that combines finger vein keys with off-axis digital holography technology. Summary of the invention

[0005] The purpose of the present invention is to provide a color image encryption method based on finger vein key and off-axis digital holography, aiming to achieve optical encryption of color images with high security, large key space, contactless authentication, good robustness of finger vein key, and easy storage and carrying of ciphertext.

[0006] To achieve the above-mentioned purpose of the invention, the technical solution provided by the present invention is as follows: a color image encryption method based on finger vein key and off-axis digital holography, comprising an encryption step and a decryption step:

[0007] The encryption comprises the following steps:

[0008] S1: Encrypted user data generation based on Swin-Transformer network and quantum walk system: First, use the infrared camera to collect the encrypted user's finger vein image Z(x, y), and then input Z(x, y) into the Swin-Transformer network to extract the global vein feature to obtain the encrypted user vein feature model Z E At the same time, the hash value SH of Z(x,y) is extracted by using the finger vein image hash value data extraction technology, and SH is combined with the encrypted user vein feature model Z E Save it to the encrypted user database, and then use the quantum initial value generation technology to generate four sets of quantum initial values ​​for SH Last pair The quantum matrix generation technology based on the quantum walk system is used to generate the encryption key in the next encryption step, namely the finger vein quantum DNA encryption mask VQEM, the finger vein quantum DNA encryption scrambled index VQEI, and the first finger vein quantum phase mask VQPM. 1 , Second finger vein quantum phase mask VQPM 2 , and save the above key to the encrypted user database.

[0009] S2: Watermark image generation based on DNA coding and discrete wavelet transform: First, the color image channel separation technology is used to perform channel separation operation on the color plaintext image P(x, y) to obtain the red channel data P of P(x, y). R (x, y), green channel component P G (x, y), blue channel component P B (x,y). Then VQEM and VQEI are used as encryption keys to encrypt P R (x,y),P G (x,y),P B (x, y) performs the image encryption operation based on DNA coding to obtain the encryption result P RR (x,y),P GR (x,y),P BR (x,y). Finally, the encrypted watermark image generation technology based on discrete wavelet transform is used to convert P RR (x,y),P GR (x,y),P BR (x, y) is embedded into the host image W(x, y) to obtain the encrypted watermark image W E (x,y).

[0010] S3: Image encryption based on Gyrator transform domain and off-axis digital holography: First, the double random phase coding encryption technology based on Gyrator transform domain and the first finger vein quantum phase mask VQPM in the encrypted user database are used.1 , Second finger vein quantum phase mask VQPM 2 And three sets of portable rotating digital keys to encrypt the watermark image W E (x,y) performs encryption operation to obtain the encrypted light wave G o (u, v), and then use a set of simple digital keys to generate a structured phase mask SPM, and then use SPM and phase modulation technology to modulate the reference light wave G r (u,v) is phase modulated to obtain the encrypted reference light wave G rE (u,v), and then record G respectively o (u,v) and G rE The light intensity of (u,v) is I o (x,y) and I r (x, y), and finally use the off-axis digital holographic technology to encrypt the light wave G o (u,v) and the encrypted reference light wave G rE (u,v) produces interference to obtain the ciphertext hologram I(x,y).

[0011] At this point, the encryption step is completed.

[0012] The decryption comprises the following steps:

[0013] J1: Finger vein authentication and decryption key acquisition based on forward reasoning calculation of convolutional neural network: First, the finger vein image of the user is decrypted and the encrypted user finger vein feature model Z in the encrypted database E As the input of the forward reasoning calculation of the convolutional neural network, the output result is then input into the Softmax activation function for calculation and encrypted user finger vein feature model Z m The similarity probability VS, if VS>δ, the system outputs the hash value SH of the encrypted user's finger vein image, and then performs quantum initial value mapping on SH to generate four sets of quantum initial values, namely Last pair The quantum matrix generation technology based on quantum walk is used to generate the decryption key required in the next decryption step, namely the finger vein quantum DNA decryption mask VQDM, the finger vein quantum DNA decryption index VQDI, and the first finger vein quantum conjugate phase mask. Second finger vein quantum conjugation phase mask Otherwise, the authentication fails and the decryption is terminated.

[0014] J2: Decrypted watermark image reconstruction based on Fresnel inverse diffraction and Gyrator transform: First, the encrypted light wave intensity I o (x,y), encrypted reference light wave intensity I r(x, y) and the DC component suppression technology based on the light intensity subtraction method suppresses the DC component of the ciphertext hologram I(x, y) to obtain the suppression result I ◇ (x,y), and the decryptor needs to input a set of correct digital keys to generate the decrypted structured phase mask Then use and phase modulation technology for the reference light wave G r (u,v) is phase modulated to obtain the decryption reference light wave G rd (u,v), then make I ◇ (x,y) and G rd After multiplying (u, v), a Fresnel inverse diffraction operation with a wavelength of λ and a diffraction distance of Z is performed on the multiplication result, and the optical information mask OIM is used to filter the inverse diffraction operation result to generate the decryption object light wave Finally, the first finger vein quantum conjugation phase mask is used Second finger vein quantum conjugation phase mask And the dual random phase decoding technology based on Gyrator transform domain Perform the decoding operation to obtain the decrypted watermark image W D (x,y).

[0015] J3: Decrypted watermark image extraction based on DNA decoding and discrete wavelet transform: First, the watermark image extraction technology based on discrete wavelet transform is used to extract the decrypted watermark image W D (x,y) performs extraction operation to obtain the extraction result Then VQDM and VQEI are used as decryption keys to Perform DNA-encoded image decryption operations to obtain decrypted results Last pair Perform image channel merging operations to obtain the final decryption result

[0016] At this point, the decryption step is completed.

[0017] The above-mentioned color image encryption method based on finger vein key and off-axis digital holography, the encryption step S1: generating an encrypted user database based on Swin-Transformer network and quantum walk, comprises the following steps:

[0018] (i) Encrypted finger vein image acquisition based on infrared camera: Use infrared camera to collect encrypted user's finger vein image Z(x,y);

[0019] (ii) Generation of encrypted user finger vein feature model based on Swin-Transformer network: Input Z(x, y) into the Swin-Transformer network to generate the encrypted user finger vein feature model Z E And save it to the encrypted user database.

[0020] (iii) Finger vein hash data extraction and quantum initial value generation: First, the hash value SH of the encrypted user's finger vein image Z(x,y) is extracted using the fingerprint vein hash data extraction technology and saved in the encrypted user database. Next, quantum initial value mapping is performed on SH to obtain four sets of quantum initial values.

[0021]

[0022] (iv) Quantum matrix generation based on quantum walk system: First, four sets of quantum initial values ​​are obtained using quantum initial value mapping. Constrain the motion trajectories of the four walkers respectively, where It is used to constrain the walking range of the walker. After being constrained, the walking range of the walker in the two-dimensional rectangular coordinate system (ξ,η) is Used to constrain the number of steps taken by the walker; Used to constrain the initial state of the coin After being constrained It can be represented by a linear combination of the state vector |κ>(κ=0,1) bundles, namely:

[0023]

[0024] Used to constrain the coin operator After being constrained for:

[0025]

[0026] Next, control the Walker within range D n (n=1,2,3,4) Walking inside steps, and the walker needs to use the Adjust to the coin state. Therefore, each step of the walker can use the unitary operator To express:

[0027]

[0028] in, represents the tensor operation, H represents the unit operator in the unitary operator, and They are respectively expressed as the displacement operators of the nth (n=1, 2, 3, 4) walker in the ξ direction and η direction of the two-dimensional rectangular coordinate system:

[0029]

[0030] Where mod is the modulus operator, |ξ,τ,0> and |ξ,τ,1> represent the state vector of the walker after walking τ steps in the positive and negative directions of ξ, |τ,η,0> and |τ,η,1> represent the state vector of the walker after walking τ steps in the positive and negative directions of η, and <ξ,η,0| and <ξ,η,1|) are the conjugates of the state vectors |ξ,η,0> and |ξ,η,1> in the positive and negative directions of ξ and η, respectively. The nth (n=1,2,3,4)th walker walks Status after step With the initial system state ψ> 0,n The evolution relationship can be expressed as:

[0031]

[0032] in Unitary operator The nth (n=1,2,3,4)th walker passes The probability matrix of arriving at the two-dimensional space (X, Y) after one step is:

[0033]

[0034] in Mobile for walkers The unitary operator after the step. Finally, the probability matrix After being arranged into a two-dimensional matrix, they are normalized to obtain the encryption key required in the next encryption step, namely, the finger vein quantum DNA mask VQEM, the finger vein quantum DNA encryption scrambled index VQEI, and the first finger vein quantum phase mask VQPM. 1 , Second finger vein quantum phase mask VQPM 2 .

[0035] In the above-mentioned color image encryption method based on finger vein key and off-axis digital holography, the encryption step S2: generating an encrypted watermark image based on DNA coding and wavelet transform, comprises the following steps:

[0036] (i) DNA-encoded color plaintext image encryption: First, the color plaintext image P(x, y) is subjected to channel separation to obtain the red channel data P R (x, y), green channel component P G (x, y), blue channel component PB (x,y). Next, we will R (x,y),P G (x,y),P B (x, y) and VQEM and VQEI in the encrypted database perform DNA encoding operations to obtain the binary encoding result P RD (α,β),P GD (α,β),P BD (α,β), VQEM D 、VQEI D , then P RD (α,β),P GD (α,β),P BD (α, β) and VQEM D Perform the XOR operation to obtain the XOR result P RK (α,β),P GK (α,β),P BK (α,β):

[0037]

[0038] Where XOR(·) represents an exclusive OR operation. Then VQEI D As the scrambling index, RK (α,β),P GK (α,β),P BK (α, β) performs the scrambling operation to obtain the scrambling result P RE (α,β),P GE (α,β),P BE (α,β):

[0039]

[0040] Among them, SO(·) represents the scrambling operation. Finally, P RE (α,β),P GE (α,β),P BE (α, β) performs DNA reverse encoding operation to obtain DNA encryption result P RR (x,y),P GR (x,y),P BR (x,y).

[0041] (ii) Encrypted watermark image generation based on wavelet transform: First, discrete wavelet transform is performed on the host image W(x,y) to decompose W(x,y) into four sub-band images: LL(x,y), LH(x,y), HL(x,y), and HH(x,y). Then, LH(x,y), HL(x,y), and HH(x,y) are replaced by P RR (x,y),PGR (x,y),P BR (x, y), and finally perform inverse discrete wavelet transform on the replacement result to obtain the encrypted watermark image W E (x,y):

[0042]

[0043] Where DWT(·) represents discrete wavelet transform, IDWT(·) represents inverse discrete wavelet transform, Indicates a replace operation.

[0044] In the above-mentioned color image encryption method based on finger vein key and off-axis digital holography, the encryption step S3: image encryption based on Gyrator transform domain and off-axis digital holography comprises the following steps:

[0045] (i) Encrypted light wave generation based on double random phase coding in Gyrator transform domain: First, the first finger vein quantum phase mask VQPM is used 1 , Second finger vein quantum phase mask VQPM 2 And three sets of easy-to-remember digital key pairs for changing angles E (x,y) is encrypted by double random phase coding based on Gyrator transformation to obtain the encrypted light wave G o (u,v):

[0046]

[0047] in, Expressed as the transformation angle μ i Gyrator transform of (i=0,1,2). Taking the complex amplitude P of a certain input t-1 (u,v) as an example, after transformation the angle is μ t The output complex amplitude P obtained after Gyrator transformation t (u′,v′) is:

[0048]

[0049] Where exp(·) is the exponential operation and j is the imaginary unit

[0050] (ii) Off-axis digital holographic interferometry under reference light modulation: First, a set of simple digital keys is used to generate the structured phase mask SPM required for encryption:

[0051]

[0052] Where λ is the wavelength of the light wave, r is the radius of the Fresnel zone plate, f is the focal length of the Fresnel zone plate, and p is the topological charge of the radial Hilbert mask. is the spatial azimuth of the radial Hilbert mask, and arg(·) is the phase operation. Then, the reference light wave G is r (u,v) is phase modulated to obtain the encrypted reference light wave G re (u,v), and then record the encrypted reference light wave G re (u,v) and encrypted light wave G o (u,v) light intensity I o (x, y) and I r (x, y) and store them in the encrypted database respectively. Finally, the off-axis digital holographic technology is used to make G o (u,v) and G re (u,v) interferes to produce the ciphertext hologram I(x,y):

[0053] I(x,y)=ODHE(FrT λ,Z (G o (u,v)),G re (u,v)

[0054] =|ZP(FrT λ,Z (G o (u,v)))+G r (u,v)×exp(jSPM) 2

[0055] Among them, ZP(·) represents the zero-filling operation, FRT λ,Z (·) represents the Fresnel diffraction operation with a wavelength of λ and a distance of Z, and ODHE(·) represents the off-axis digital holographic encoding operation.

[0056] The above-mentioned color image encryption method based on finger vein key and off-axis digital holography, the decryption step J1: finger vein authentication and decryption key acquisition based on forward reasoning calculation of convolutional neural network, includes the following steps:

[0057] (i) Decrypted finger vein image acquisition based on infrared camera: Use infrared camera to collect and decrypt the user's finger vein image

[0058] (ii) Decryption data acquisition based on finger vein authentication: First, decrypt the user's finger vein image and the encrypted user finger vein feature model Z in the encrypted database m As the input of the inference calculation based on the forward propagation principle of the convolutional neural network, the output result is then input into the Softmax activation function to calculate and encrypted user finger vein feature model Z E Similar probability VS:

[0059]

[0060] Where FP(·) represents the forward reasoning calculation process of the convolutional neural network, and Softmax(·) represents the Softmax activation function. Assume that the K-dimensional vector As the input of the Softmax activation function, the calculation result is output for:

[0061]

[0062] If VS is greater than the preset threshold δ, the system outputs the hash value SH of the encrypted user's finger vein image Z(x,y) and enters the next decryption step. Otherwise, the authentication fails and the decryption is terminated.

[0063] (iii) Decryption key generation based on quantum initial value mapping and quantum walk: First, the same as sub-step (iii) in encryption step S1, the quantum initial value mapping is performed on SH to obtain four sets of quantum initial values Then will They are used as the input of the quantum walk system respectively, and the output results are arranged into a two-dimensional matrix and normalized to obtain the finger vein quantum DNA mask VQDM, the finger vein quantum DNA scrambled index VQEI, and the first finger vein quantum phase mask VQPM. 1 , Second finger vein quantum phase mask VQPM 2 , followed by VQPM 1 、VQPM 1 Perform conjugation operations respectively to obtain the first finger vein quantum conjugation phase mask Second finger vein quantum conjugation phase mask

[0064]

[0065] Among them, conj(·) represents the conjugation operation. Finally, the finger vein quantum DNA mask VQDM, the finger vein quantum DNA scrambled index VQEI, the first finger vein quantum conjugated phase mask Second finger vein quantum conjugation phase mask As the decryption key in the next decryption step.

[0066] In the above-mentioned color image encryption method based on finger vein key and off-axis digital holography, the decryption step J2: reconstruction of the decrypted watermark image based on Fresnel diffraction and Gyrator transform comprises the following steps:

[0067] (i) DC component suppression based on light intensity subtraction: The encrypted hologram I(x, y) is compared with the encrypted object light wave intensity I in the encryption database. o(x, y), encrypted reference light wave intensity I r (x, y) can be subtracted to obtain the DC component suppression result I of I(x, y) ◇ (x, y):

[0068] I ◇ (x, y) = I (x, y) - I o (x, y)-I r (x, y)

[0069] (ii) Decryption light wave reconstruction based on Fresnel inverse diffraction: First, the decryption user needs to enter a set of correct digital keys and perform the same operation as sub-step (iii) in the encryption step S2 to obtain the decrypted structured phase mask. Then use For the reference light wave G r (u,v) is phase modulated to obtain the decryption reference light wave G rd (u,v). Next, I ◇ (x, y) and G rd (u,v) and perform a Fresnel inverse diffraction operation on the result with a wavelength of λ and a diffraction distance of Z to obtain the light wave containing the decryption object. The spectrum distribution is then filtered by constructing an optical information mask (OIM) to obtain the decrypted optical wave.

[0070]

[0071] Among them, IFrt λ，Z (·) is the Fresnel inverse diffraction with wavelength λ and diffraction distance Z,

[0072] (iii) Decrypted watermark image reconstruction based on double random phase decoding in Gyrator transform domain: The first finger vein quantum conjugate phase mask Second finger vein quantum conjugation phase mask VQPM * 2 As the decryption phase key, Gyrator transforms the angle μ i (i=0,1,2) is used as the decryption digital key to decrypt the object light wave. The decrypted watermark image W is obtained by using the double random phase decoding operation based on the Gyrator transform domain. D (x,y):

[0073]

[0074] In the above-mentioned color image encryption method based on finger vein key and off-axis digital holography, the decryption step J3: extracting the decrypted watermark image based on DNA decoding and discrete wavelet transform comprises the following steps:

[0075] (i) Decryption data extraction based on wavelet transform: Decryption watermark image W is extracted using decryption data extraction technology based on wavelet transform. D Decrypted data in (x,y)

[0076] Among them, LL D To decrypt the watermark image W D Low-frequency information in the horizontal and vertical directions in (x,y).

[0077] (ii) Decryption result acquisition based on DNA decoding: First, the decrypted data is obtained in the same manner as sub-step (i) in encryption step S2. Finger vein quantum DNA decryption mask VQDM, finger vein quantum DNA decryption index VQDI perform DNA encoding operation to obtain binary encoding result VQDI D 、VQDM D Then VQDI D As scrambling index, Perform the inverse scrambling operation to get the inverse scrambling result

[0078]

[0079] Among them, ISO(·) represents the inverse scrambling operation. VQDM D Perform an XOR operation to obtain the XOR result

[0080] Then, XOR the results separately Perform DNA reverse encoding operation to obtain the red channel decryption result Green channel decryption results Blue channel decryption results Finally, the channel merging technology is used to Perform the merge operation to obtain the final decryption result

[0081] Compared with the prior art, the present invention has the following beneficial effects:

[0082] (1) The finger vein key of the present invention is easy to carry, the digital key is easy to remember, and there is no phase mask key, which effectively reduces the complexity of the key.

[0083] (2) The present invention uses the Swin-Transformer network to extract the encrypted user vein feature model in the encryption stage. In the decryption stage, the finger vein authentication technology based on the forward reasoning calculation of the convolutional neural network is used to perform security authentication on the decrypted finger vein, thereby effectively improving the robustness and security of the finger vein key, and further improving the practicality of the finger vein key in the optical image encryption method.

[0084] (3) The present invention sets the rotation angle in the double random phase encoding process based on the Gyrator transform domain and the optical parameters of the structured light phase mask (SPM) in the off-axis digital holographic encryption process as digital keys, thereby increasing the types and quantity of digital keys, expanding the key space and thus improving the security of the encryption system.

[0085] (4) The present invention utilizes holographic ciphertext generation technology based on structured light illumination and off-axis digital holography, so that a color plaintext image is encrypted into an amplitude-type grayscale ciphertext, which is easier to store and transport than traditional complex amplitude ciphertext.

[0086] (5) Compared with the original color plaintext image, the decrypted image of the present invention has no color distortion, brightness shift, or contrast change, thereby achieving efficient and secure color image encryption, thereby ensuring the integrity of the decrypted image's color, details, and layer information. BRIEF DESCRIPTION OF THE DRAWINGS

[0087] Figure 1 This is an encryption flow chart of a color image encryption method based on finger vein key and off-axis digital holography.

[0088] Figure 2 This is a flow chart of the encryption sub-step S1 in the color image encryption method based on finger vein key and off-axis digital holography.

[0089] Figure 3 The schematic diagram of the Swin-Transformer network in the encryption step S1, where PP is the program patch partition function and LE is the linear embedding function.

[0090] Figure 4 Schematic diagram of two Swin-Transformer modules (STB) in the Swin-Transformer network. LN is the layer normalization function; W-MSA is the multi-head attention mechanism; MLP is the multi-layer perceptron.

[0091] Figure 5 This is a flow chart of the encryption sub-step S2 in the color image encryption method based on finger vein key and off-axis digital holography.

[0092] Figure 6Schematic diagram of an optical device to assist in describing the encryption steps, where P is a polarizer, OL is an objective lens, PH is a pinhole, CL is a collimating lens, BS is a beam splitter prism, M is a reflector, and L is a lens.

[0093] Figure 7 This is a flow chart of the encryption sub-step S3 in the color image encryption method based on finger vein key and off-axis digital holography.

[0094] Figure 8 The decryption flow chart of the color image encryption method based on finger vein key and off-axis digital holography.

[0095] Fig. 9 This is a flow chart of the decryption sub-step J1 in the color image encryption method based on finger vein key and off-axis digital holography.

[0096] Fig.10 This is a flow chart of the decryption sub-step J2 in the color image encryption method based on finger vein key and off-axis digital holography.

[0097] Fig.11 This is a flowchart of the decryption sub-step J3 in the color image encryption method based on finger vein key and off-axis digital holography.

[0098] Fig.12 (a) is the color plaintext image "Peppers"; (b1)-(b3) are the data of the red channel, green channel, and blue channel of the color plaintext image "Peppers" respectively; (c) is the encrypted user finger vein image Z(x, y) captured by the infrared camera; (d1)-(d4) are the vein quantum DNA encryption mask VQEM, finger vein quantum DNA encryption index VQEI, and first finger vein quantum phase mask VQPM generated according to (c) respectively. 1 , Second finger vein quantum phase mask VQPM 2 ; (e1)-(e3) are the results of (a1)-(a3) after DNA encryption; (f) and (g) are the host image W(x, y) and the encrypted watermark image W E (x, y); (h1)-(h2) are the encrypted light waves G o (u,v) is the amplitude distribution and phase distribution; (j) is the structured phase mask SPM; (k) is the encrypted hologram I(x,y).

[0099] Fig.13 (a1)-(a3) are the decryption results of the red channel, green channel, and blue channel of the color plaintext image “Peppers” when the key is completely correct, and (b) is the decryption result of the color plaintext image “Peppers” when the key is completely correct.

[0100] Fig.14 It is the decryption result when the digital key is correct and the finger vein key is wrong.

[0101] Fig.15 (a)-(c) are the robustness analysis results of finger vein keys when subjected to translation attack, rotation attack, and brightness attack, respectively.

[0102] Fig.16 (a)-(f) are the rotation keys μ respectively 0 , rotate the key μ 1 , rotate the key μ 2 , wavelength key λ, focal length key f, and topological charge number key p. DETAILED DESCRIPTION

[0103] The present invention is further described below in conjunction with the embodiments and drawings, but they are not intended to limit the present invention.

[0104] The design idea of ​​the present invention is: firstly, an infrared camera is used to obtain the encrypted user's finger vein image Z(x, y), and then Z(x, y) is input into the Swin-Transformer network to extract the global vein feature, thereby obtaining the encrypted user vein feature model Z E At the same time, the hash value SH of the encrypted user's finger vein image is extracted from Z(x, y) using the finger vein image hash value extraction technology. Next, the finger vein encryption key generation technology based on quantum initial value mapping and quantum walk is used to generate the finger vein quantum DNA encryption mask VQEM, the finger vein quantum DNA encryption index VQEI, and the first finger vein quantum phase mask VQPM. 1 , Second finger vein quantum phase mask VQPM 2 Then, the color image channel separation technology is used to separate the color plaintext image P(x, y) so that the red channel component P R (x, y), green channel component P G (x, y), blue channel component P B Then, VQDM and VQEI are used as keys to encrypt P using DNA-based image encryption technology. R (x,y),P G (x,y),P B (x, y) is encrypted to obtain the encrypted data P BE (x,y),P GE (x,y),P BE (x,y). Next, we introduce the host image W(x,y) and use the watermark image embedding technology based on discrete wavelet transform to embed P BE (x,y),P GE (x,y),PBE (x,y) is embedded into W(x,y) to obtain the watermarked encrypted image W E (x,y). Then VQPM 1 、VQPM 2 As the phase key pair W E (x,y) is encrypted by double random phase coding based on Gyrator transformation to obtain the encrypted light wave G E (u,v), and then introduce the reference light G r (u,v) and use G E (u,v) and the holographic ciphertext generation technology based on structured light illumination and off-axis digital holography generate the final ciphertext I(x,y).

[0105] Embodiment: The present invention provides a color image encryption method based on finger vein key and off-axis digital holography, including: Figure 1 The encryption steps shown and Figure 8 Decryption steps shown:

[0106] The encryption steps are as follows Figure 1 As shown, the specific steps include:

[0107] S1: Encrypted user data generation based on Swin-Transformer network and quantum walk system: First, use the infrared camera to collect the encrypted user's finger vein image Z(x, y), and then input Z(x, y) into the Swin-Transformer network to extract the global vein feature to obtain the encrypted user vein feature model Z E At the same time, the hash value SH of Z(x,y) is extracted by using the finger vein image hash value data extraction technology, and SH is combined with the encrypted user vein feature model Z E Save it to the encrypted user database, and then use the quantum initial value generation technology to generate four sets of quantum initial values ​​for SH Last pair The quantum matrix generation technology based on the quantum walk system is used to generate the encryption key in the next encryption step, namely the finger vein quantum DNA encryption mask VQEM, the finger vein quantum DNA encryption scrambled index VQEI, and the first finger vein quantum phase mask VQPM. 1 , Second finger vein quantum phase mask VQPM 2 , and save the above key to the encrypted user database. The process is as follows Figure 2 As shown, the specific steps include:

[0108] (i) Encrypted finger vein image acquisition based on infrared camera: Use infrared camera to collect encrypted user's finger vein image Z(x,y);

[0109] (ii) Generation of encrypted user finger vein feature model based on Swin-Transformer network: Input Z(x,y) into Figure 3 The encrypted user finger vein feature model Z is generated in the Swin-Transformer network composed of a program patch partition function (PP), a linear embedding function (LE) and multiple consecutive Swin-Transformer Blocks (STBs) E And save it to the encrypted user database. Figure 4 As shown, it includes two head attention mechanisms (W-MSA), four layer normalization functions (LN), and two multi-layer perceptrons (MLP).

[0110] (iii) Finger vein hash data extraction and quantum initial value generation: First, the hash value SH of the encrypted user's finger vein image Z(x,y) is extracted using fingerprint vein hash data extraction technology and saved in the encrypted user database. Next, quantum initial value mapping is used to obtain four sets of quantum initial values ​​for SH.

[0111]

[0112] (iv) Quantum matrix generation based on quantum walk system: First, four sets of quantum initial values ​​are obtained using quantum initial value mapping. Constrain the motion trajectories of the four walkers respectively, where It is used to constrain the walking range of the walker. After being constrained, the walking range of the walker in the two-dimensional rectangular coordinate system (ξ,η) is Used to constrain the number of steps taken by the walker; Used to constrain the initial state of the coin After being constrained It can be represented by a linear combination of the state vector |κ>(κ=0,1) bundles, namely:

[0113]

[0114] Used to constrain the coin operator After being constrained for:

[0115]

[0116] Next, control the Walker within range D n (n=1,2,3,4) Internal walking SH 2 n steps, and the walker needs to use the Adjust to the coin state. Therefore, each step of the walker can use the unitary operator To express:

[0117]

[0118] in, represents the tensor operation, H represents the unit operator in the unitary operator, and They are respectively expressed as the displacement operators of the nth (n=1, 2, 3, 4) walker in the ξ direction and η direction of the two-dimensional rectangular coordinate system:

[0119]

[0120] Where mod is the modulus operator, |ξ,τ,0> and |ξ,τ,1> represent the state vector of the walker after walking τ steps in the positive and negative directions of ξ, |τ,η,0> and |τ,η,1> represent the state vector of the walker after walking τ steps in the positive and negative directions of η, and <ξ,η,0| and <ξ,η,1|) are the conjugates of the state vectors |ξ,η,0> and |ξ,η,1> in the positive and negative directions of ξ and η, respectively. The nth (n=1,2,3,4)th walker walks Status after step With the initial system state |ψ> 0,n The evolution relationship can be expressed as:

[0121]

[0122] in Unitary operator The nth (n=1,2,3,4)th walker passes The probability matrix of arriving at the two-dimensional space (X, Y) after one step is:

[0123]

[0124] in Mobile for walkers The unitary operator after the step. Finally, the probability matrix After being arranged into a two-dimensional matrix, they are normalized to obtain the encryption key required in the next encryption step, namely, the finger vein quantum DNA mask VQEM, the finger vein quantum DNA encryption scrambled index VQEI, and the first finger vein quantum phase mask VQPM. 1 , Second finger vein quantum phase mask VQPM 2 .

[0125] S2: Watermark image generation based on DNA coding and discrete wavelet transform: First, the color image channel separation technology is used to perform channel separation operation on the color plaintext image P(x, y) to obtain the red channel data P of P(x, y). R (x, y), green channel component P G (x, y), blue channel component P B (x,y). Then VQEM and VQEI are used as encryption keys to encrypt P R (x,y),P G (x,y),P B (x, y) performs the image encryption operation based on DNA coding to obtain the encryption result P RR (x,y),P GR (x,y),P BR (x,y). Finally, the encrypted watermark image generation technology based on discrete wavelet transform is used to convert P RR (x,y),P GR (x,y),P BR (x, y) is embedded into the host image W(x, y) to obtain the encrypted watermark image W E (x,y). Its flow chart is as follows Figure 5 As shown, the specific steps include:

[0126] (i) DNA-encoded color plaintext image encryption: First, the color plaintext image P(x, y) is subjected to channel separation to obtain the red channel data P R (x, y), green channel component P G (x, y), blue channel component P B (x,y). Next, we will R (x,y),P G (x,y),P B (x, y) and VQEM and VQEI in the encrypted database perform DNA encoding operations to obtain the binary encoding result P RD (α,β),P GD (α,β),P BD (α,β), VQEM D 、VQEI D , then P RD (α,β),P GD (α,β),P BD (α, β) and VQEM D Perform the XOR operation to obtain the XOR result P RK (α,β),P GK (α,β),P BK (α,β):

[0127]

[0128] Where XOR(·) represents an exclusive OR operation. Then VQEI D As the scrambling index, RK (α,β),P GK (α,β),P BK (α, β) performs the scrambling operation to obtain the scrambling result P RE (α,β),P GE (α,β),P BE (α,β):

[0129]

[0130] Among them, SO(·) represents the scrambling operation. Finally, P RE (α,β),P GE (α,β),P BE (α, β) performs DNA reverse encoding operation to obtain DNA encryption result P RR (x,y),P GR (x,y),P BR (x,y).

[0131] (ii) Encrypted watermark image generation based on wavelet transform: First, discrete wavelet transform is performed on the host image W(x,y) to decompose W(x,y) into four sub-band images: LL(x,y), LH(x,y), HL(x,y), and HH(x,y). Then, LH(x,y), HL(x,y), and HH(x,y) are replaced by P RR (x,y),P GR (x,y),P BR (x, y), and finally perform inverse discrete wavelet transform on the replacement result to obtain the encrypted watermark image W E (x,y):

[0132] Where DWT(·) represents discrete wavelet transform, IDWT(·) represents inverse discrete wavelet transform, Indicates a replace operation.

[0133] S3: Image encryption based on Gyrator transform domain and off-axis digital holography: First, the double random phase coding encryption technology based on Gyrator transform domain and the first finger vein quantum phase mask VQPM in the encrypted user database are used. 1 , Second finger vein quantum phase mask VQPM 2 And three sets of portable rotating digital keys to encrypt the watermark image W E (x,y) performs encryption operation to obtain the encrypted light wave Go (u, v), and then use a set of simple digital keys to generate a structured phase mask SPM, and then use SPM and phase modulation technology to modulate the reference light wave G r (u,v) is phase modulated to obtain the encrypted reference light wave G rE (u,v), and then record G respectively o (u,v) and G rE The light intensity of (u,v) is I o (x,y) and I r (x, y), and finally use the off-axis digital holographic technology to encrypt the light wave G o (u,v) and the encrypted reference light wave G rE (u,v) produces interference to obtain the encrypted hologram I(x,y). The optical path diagram is as follows Figure 6 As shown, the flowchart of this step is as follows Figure 7 As shown, the specific steps include:

[0134] (i) Encrypted light wave generation based on double random phase coding in Gyrator transform domain: First, the first finger vein quantum phase mask VQPM is used 1 , Second finger vein quantum phase mask VQPM 2 And three sets of easy-to-remember digital key pairs for changing angles E (x,y) is encrypted by double random phase coding based on Gyrator transformation to obtain the encrypted light wave G o (u,v):

[0135]

[0136] in, Expressed as the transformation angle μ i Gyrator transform of (i=0,1,2). Taking the complex amplitude P of a certain input t-1 (u,v) as an example, after transformation the angle is μ t The output complex amplitude P obtained after Gyrator transformation t (u′,v′) is:

[0137]

[0138] Where exp(·) is the exponential operation and j is the imaginary unit

[0139] (ii) Off-axis digital holographic interferometry under reference light modulation: First, a set of simple digital keys is used to generate the structured phase mask SPM required for encryption:

[0140]

[0141] Where λ is the wavelength of the light wave, r is the radius of the Fresnel zone plate, f is the focal length of the Fresnel zone plate, and p is the topological charge of the radial Hilbert mask. is the spatial azimuth of the radial Hilbert mask, and arg(·) is the phase operation. Then, the reference light wave G is r (u,v) is phase modulated to obtain the encrypted reference light wave G re (u,v), and then record the encrypted reference light wave G re (u,v) and encrypted light wave G o (u,v) light intensity I o (x, y) and I r (x, y) and store them in the encrypted database respectively. Finally, the off-axis digital holographic technology is used to make G o (u,v) and G re (u,v) interferes to produce the ciphertext hologram I(x,y):

[0142] I(x,y)=ODHE(FrT λ,Z (G o (u,v)),G re (u,v)

[0143] =|ZP(FrT λ,Z (G o (u,v)))+G r (u,v)×exp(jSPM)| 2

[0144] Among them, ZP(·) represents the zero-filling operation, FRT λ,Z (·) represents the Fresnel diffraction operation with a wavelength of λ and a distance of Z, and ODHE(·) represents the off-axis digital holographic encoding operation.

[0145] The decryption steps are as follows: Figure 8 As shown, the specific steps include:

[0146] J1: Finger vein authentication and decryption key acquisition based on Transformer network: First, decrypt the user's finger vein image and the encrypted user finger vein feature model Z in the encrypted database E As the input of the forward reasoning calculation of the convolutional neural network, the output result is then input into the Softmax activation function for calculation and encrypted user finger vein feature model Z m The similarity probability VS, if VS>δ, the system outputs the hash value SH of the encrypted user's finger vein image, and then performs the quantum initial value mapping operation on SH to generate four sets of quantum initial values, namely Last pair The quantum matrix generation technology based on quantum walk is used to generate the decryption key required in the next decryption step, namely the finger vein quantum DNA decryption mask VQDM, the finger vein quantum DNA decryption index VQDI, and the first finger vein quantum conjugate phase mask. Second finger vein quantum conjugation phase mask Otherwise, the authentication fails and the decryption is terminated. The flow chart is as follows Fig. 9 As shown, the specific steps include:

[0147] (i) Decrypted finger vein image acquisition based on infrared camera: Use infrared camera to collect and decrypt the user's finger vein image

[0148] (ii) Decryption data acquisition based on finger vein authentication: First, decrypt the user's finger vein image and the encrypted user finger vein feature model Z in the encrypted database m As the input of the inference calculation based on the forward propagation principle of the convolutional neural network, the output result is then input into the Softmax activation function to calculate and encrypted user finger vein feature model Z E Similar probability VS:

[0149]

[0150] Where FP(·) represents the forward reasoning calculation process of the convolutional neural network, and Softmax(·) represents the Softmax activation function. Assume that the K-dimensional vector As the input of the Softmax activation function, the calculation result is output for:

[0151]

[0152] If VS is greater than the preset threshold δ, the system outputs the hash value SH of the encrypted user's finger vein image Z(x,y) and enters the next decryption step. Otherwise, the authentication fails and the decryption is terminated.

[0153] (iii) Decryption key generation based on quantum initial value mapping and quantum walk: First, the same as sub-step (iii) in encryption step S1, the quantum initial value mapping is performed on SH to obtain four sets of quantum initial values Then will They are used as the input of the quantum walk system respectively, and the output results are arranged into a two-dimensional matrix and normalized to obtain the finger vein quantum DNA mask VQDM, the finger vein quantum DNA scrambled index VQEI, and the first finger vein quantum phase mask VQPM. 1 , Second finger vein quantum phase mask VQPM 2 , followed by VQPM1 、VQPM 1 Perform conjugation operations respectively to obtain the first finger vein quantum conjugation phase mask Second finger vein quantum conjugation phase mask

[0154]

[0155] Among them, conj(·) represents the conjugation operation. Finally, the finger vein quantum DNA mask VQDM, the finger vein quantum DNA scrambled index VQEI, the first finger vein quantum conjugated phase mask Second finger vein quantum conjugation phase mask As the decryption key in the next decryption step.

[0156] J2: Decrypted watermark image reconstruction based on Fresnel inverse diffraction and Gyrator transform: First, the encrypted light wave intensity I o (x,y), encrypted reference light wave intensity I r (x, y) and the DC component suppression technology based on the light intensity subtraction method suppresses the DC component of the ciphertext hologram I(x, y) to obtain the suppression result I ◇ (x,y), and the decryptor needs to input a set of correct digital keys to generate the decrypted structured phase mask Then use and phase modulation technology for the reference light wave G r (u,v) is phase modulated to obtain the decryption reference light wave G rd (u,v), then make I ◇ (x,y) and G rd After multiplying (u, v), a Fresnel inverse diffraction operation with a wavelength of λ and a diffraction distance of Z is performed on the multiplication result, and the optical information mask OIM is used to filter the inverse diffraction operation result to generate the decryption object light wave Finally, the first finger vein quantum conjugation phase mask is used Second finger vein quantum conjugation phase mask And the dual random phase decoding technology based on Gyrator transform domain Perform the decoding operation to obtain the decrypted watermark image W D (x,y). Its flow chart is as follows Fig.10 As shown, the specific steps include:

[0157] (i) DC component suppression based on light intensity subtraction: The encrypted hologram I(x, y) is compared with the encrypted object light wave intensity I in the encryption database. o (x, y), encrypted reference light wave intensity I r(x, y) can be subtracted to obtain the DC component suppression result I of I(x, y) ◇ (x, y):

[0158] I ◇ (x, y) = I (x, y) - I o (x, y)-I r (x, y)

[0159] (ii) Decryption light wave reconstruction based on Fresnel inverse diffraction: First, the decryption user needs to enter a set of correct digital keys and perform the same operation as sub-step (iii) in the encryption step S2 to obtain the decrypted structured phase mask. Then use For the reference light wave G r (u,v) is phase modulated to obtain the decryption reference light wave G rd (u,v). Next, I ◇ (x, y) and G rd (u,v) and perform a Fresnel inverse diffraction operation on the result with a wavelength of λ and a diffraction distance of Z to obtain the light wave G containing the decryption object. rd The spectrum distribution of (u,v) is then filtered by constructing an optical information mask OIM to obtain the decrypted optical wave

[0160]

[0161] Among them, IFrt λ，Z (·) is the Fresnel inverse diffraction with wavelength λ and diffraction distance Z,

[0162] (iii) Decrypted watermark image reconstruction based on double random phase decoding in Gyrator transform domain: The first finger vein quantum conjugate phase mask Second finger vein quantum conjugation phase mask As the decryption phase key, Gyrator transforms the angle μ i (i=0,1,2) is used as the decryption digital key to decrypt the object light wave. The decrypted watermark image W is obtained by using the double random phase decoding operation based on the Gyrator transform domain. D (x,y):

[0163]

[0164] J3: Decrypted watermark image extraction based on DNA decoding and discrete wavelet transform: First, the watermark image extraction technology based on discrete wavelet transform is used to extract the decrypted watermark image W D (x,y) performs extraction operation to obtain the extraction result Then VQDM and VQEI are used as decryption keys to Perform DNA-encoded image decryption operations to obtain decrypted results Last pair Perform image channel merging operations to obtain the final decryption result Its flow chart is as follows Fig.11 As shown, the specific steps include:

[0165] (i) Decryption data extraction based on wavelet transform: Decryption watermark image W is extracted using decryption data extraction technology based on wavelet transform. D Decrypted data in (x,y)

[0166] Among them, LL D To decrypt the watermark image W D Low-frequency information in the horizontal and vertical directions in (x,y).

[0167] (ii) Decryption result acquisition based on DNA decoding: First, the decrypted data is obtained in the same manner as sub-step (i) in encryption step S2. Finger vein quantum DNA decryption mask VQDM, finger vein quantum DNA decryption index VQDI perform DNA encoding operation to obtain binary encoding result VQDI D 、VQDM D Then VQDI D As scrambling index, Perform the inverse scrambling operation to get the inverse scrambling result

[0168]

[0169] Among them, ISO(·) represents the inverse scrambling operation. VQDM D Perform an XOR operation to obtain the XOR result

[0170] Then, XOR the results separately Perform DNA reverse encoding operation to obtain the red channel decryption result Green channel decryption results Blue channel decryption results Finally, the channel merging technology is used to Perform the merge operation to obtain the final decryption result

[0171] The content of the present invention is further explained below in conjunction with the accompanying drawings:

[0172] See also Figure 1 , to encrypt the color image. First, select Fig.12 The color image (“Peppers”) with a pixel count of 256×256 shown in (a) is used as the original plaintext image P(x,y) to be encrypted. The color image channel separation technology is used to separate the color plaintext image P(x,y) to obtain the red channel component P of P(x,y). R (x, y), green channel component P G (x, y), blue channel component P B (x,y)(respectively Fig.12 (b1)-(b3)). Then, an infrared camera is used to obtain Fig.12 (c) shows the encrypted user finger vein image Z(x,y). Next, the quantum matrix generation technology based on quantum walk is used to generate the following Fig.12 (d1)-(d4) shows the vein quantum DNA mask VQEM, the finger vein quantum DNA scramble index VQEI, and the first finger vein quantum phase mask VQPM 1 , Second finger vein quantum phase mask VQPM 2 Then use VQEM and VQEI as keys to identify P R (x,y),P G (x,y),P B (x,y) performs DNA encryption operation to obtain Fig.12 The encryption result P of (e1)-(e3) RR (x,y),P GR (x,y),P BR (x,y), then introduce Fig.12 (f) shows the host image W(x,y), and uses the watermark image embedding technology based on discrete wavelet transform to embed P RR (x,y),P GR (x,y),P BR (x,y) is embedded into W(x,y) to obtain Fig.12 (g) The encrypted watermark image W E (x,y), and then VQPM 1 、VQPM 2 As the phase key and set three sets of transformation angle keys (μ 1 =π / 6, μ 2 =π / 3, μ 3 =π / 2) for W E (x,y) performs double random phase coding encryption based on the Gyrator transform domain to obtain the encrypted light wave G o (u,v) whose amplitude and phase are respectively Fig.12 (h1)-(h2) As shown, the reference light wave used in the off-axis digital holographic encoding is set to be a uniform plane light wave G with a wavelength λ of 632.8 nm and not coaxial with the object light wave. r (u,v), and set Fig.12 The structured light phase mask SPM shown in (i) is composed of a Fresnel zone plate with a focal length of f = 30 mm and a radial Hilbert mask with a topological charge of p = 6. r (u, v) is phase modulated, and the Fresnel diffraction distance d is set to 1500 mm. The ciphertext hologram generated by the off-axis digital holographic interference under the reference light modulation described in sub-step (ii) of the encryption step S3 is as follows: Fig.12 (j) as shown.

[0173] See also Figure 8 The decryption steps shown in the figure are used to decrypt the ciphertext hologram I(x, y). The required decryption key includes the decrypted user's finger vein image, a set of digital keys consisting of three rotation angles (μ 1 , μ 2 , μ 3 ) and another set of digital keys consisting of focal length f, wavelength λ, and topological charge number p. When all the keys are correct, the decryption results of the red channel, green channel, and blue channel of the color image ("Peppers") are as follows Fig.13 (a1)-(a3) show that the final decryption result is as follows Fig.13 (b). Fig.13 From the decryption results in , we can see that when all the keys are correct, the original plaintext image can be reconstructed with high quality.

[0174] The security of the finger vein key in the present invention is examined below: The applicant randomly collected more than 500 finger vein images from different attackers and used them as false decryption fingerprint keys. Fig.14 The decryption result when the decryption user's finger vein key is wrong is given. Fig.14 It can be seen that when an error occurs in decrypting the user's finger vein key, the vein similarity VS between the wrong finger vein key and the correct finger vein key is less than the preset threshold δ (0.75), resulting in the decryption being stopped. It can be seen that the finger vein key in the present invention has strong security.

[0175] Next, we examine the robustness of the finger vein key in the present invention, including translation robustness, rotation robustness, and illumination robustness. First, we test the ability of the finger vein key to resist translation attacks by translating the number of pixels of the correct finger vein image. The test results are shown in Figure 2. Fig.15As shown in (a). From the authentication result in 15(a), it can be seen that although the similarity between the decrypted finger vein key and the encrypted finger vein key decreases with the increase of the number of translation pixels, the vein similarity corresponding to the finger vein key with a translation pixel number of 60Pix is ​​still greater than the threshold (0.75), so that effective authentication can still be achieved, and then high-quality decryption can be completed. Subsequently, the finger vein key is rotated to test the ability of the finger vein key to resist rotation attacks. The test results are shown in Fig.15 (b). Fig.15 The authentication result in (b) shows that even if the finger vein key is rotated 90°, the vein similarity corresponding to it is still greater than the threshold (0.75), so effective authentication can still be achieved, and high-quality decryption can be completed. Secondly, the ability of the finger vein key to resist illumination attacks is tested by changing the illumination offset value. The test results are shown in Fig.15 (c) shows the finger vein key The result of the illumination attack after the illumination offset value is ρ can be expressed as:

[0176]

[0177] Where E is a matrix of all 1s. Fig.15 The authentication result in (c) shows that even if the finger vein key is subjected to an illumination attack with an illumination offset of ρ = ±40, the corresponding vein similarity is still greater than the threshold (0.75), so effective authentication can still be achieved, and high-quality decryption can be completed. Fig.15 It can be proved that the finger vein key in the present invention has good robustness against translation attacks and rotation attacks.

[0178] Finally, the sensitivity of the digital key in the present invention is investigated, wherein the correlation coefficient (CC) is used as an evaluation function to quantitatively evaluate the correlation between the decryption result and the original plaintext image, namely:

[0179]

[0180] In the formula, f M (x,y) and Respectively represent the grayscale values ​​of the original plaintext image and the decrypted result at (x, y) in channel M (M = R, G, B), and Respectively represent the average pixel grayscale values ​​of the original plaintext image and the decrypted result in channel M (M = R, G, B). Fig.16 (a)-(f) show the rotation key μ 0 , rotate the key μ 1 , rotate the key μ 2 , wavelength key λ, focal length key f, topological charge number key p. In addition, when the rotation key μ0 Deviation Δμ 0 = 1%, when the rotation key μ 1 Deviation Δμ 1 = 1%, when the rotation key μ 2 Deviation Δμ 2 =1%, when the wavelength key λ deviates Δλ=1nm, when the focal length key f deviates Δf=1mm, when the topological charge key p deviates Δp=1, the decryption results of the red, green and blue channels are marked on Fig.16 (a)-(f). Fig.16 It can be seen from the sensitivity curve and the erroneous decryption results that when any digital key has a slight deviation, the CC between the decryption results of each channel and the plaintext image drops sharply to close to 0. At this time, the decryption results all show obvious noise distribution, resulting in the original plaintext image cannot be correctly restored, thereby proving the security of the digital key in the present invention and indicating that the digital key in the present invention has high sensitivity.

[0181] The above is an explanation of the specific implementation of the present invention, rather than a limitation of the present invention. Those skilled in the relevant technical field can also make various equivalent technical solutions without departing from the scope of the present invention, so all equivalent technical solutions should be included in the patent protection scope of the present invention.

Claims

1. A color image encryption method based on finger vein key and off-axis digital holography, characterized in that: The encryption steps include: S1: Encrypted user data generation based on Swin-Transformer network and quantum walk system: First, use the infrared camera to collect the encrypted user's finger vein image Z(x, y), and then input Z(x, y) into the Swin-Transformer network to extract the global vein feature to obtain the encrypted user vein feature model Z E ; At the same time, the hash value SH of Z(x,y) is extracted by using the finger vein image hash value data extraction technology, and SH is combined with the encrypted user vein feature model Z E Save it to the encrypted user database, and then use the quantum initial value generation technology to generate four sets of quantum initial values ​​for SH Last pair The encryption key in the next encryption step is generated by using the quantum matrix generation technology based on the quantum walk system, namely the finger vein quantum DNA encryption mask VQEM, the finger vein quantum DNA encryption scrambled index VQEI, the first finger vein quantum phase mask VQPM1, and the second finger vein quantum phase mask VQPM2, and the above keys are saved in the encryption user database; S2: Watermark image generation based on DNA coding and discrete wavelet transform: First, the color image channel separation technology is used to perform channel separation operation on the color plaintext image P(x, y) to obtain the red channel data P of P(x, y). R (x, y), green channel component P G (x, y), blue channel component P B (x, y); then VQEM and VQEI are used as encryption keys to encrypt P R (x,y),P G (x,y),P B (x, y) performs the image encryption operation based on DNA coding to obtain the encryption result P RR (x,y),P GR (x,y),P BR (x, y); Finally, the encrypted watermark image generation technology based on discrete wavelet transform is used to convert P RR (x,y),P GR (x,y),P BR (x, y) is embedded into the host image W(x, y) to obtain the encrypted watermark image W E (x,y); S3: Image encryption based on Gyrator transform domain and off-axis digital holography: First, the watermark image W is encrypted using the dual random phase coding encryption technology based on the Gyrator transform domain and the first finger vein quantum phase mask VQPM1, the second finger vein quantum phase mask VQPM2 in the encryption user database and three sets of portable rotating digital key pairs. E (x,y) performs encryption operation to obtain the encrypted light wave G o (u, v), and then use a simple set of digital keys to generate a structured phase mask SPM, and then use SPM and phase modulation technology to modulate the reference light wave G r (u,v) is phase modulated to obtain the encrypted reference light wave G rE (u,v), and then record G respectively o (u,v) and G rE The light intensity of (u,v) is I o (x,y) and I r (x, y), and finally use the off-axis digital holographic technology to encrypt the light wave G o (u,v) and the encrypted reference light wave G rE (u,v) produces interference to obtain the ciphertext hologram I(x,y); At this point, the encryption step is completed; Decryption steps: J1: Finger vein authentication and decryption key acquisition based on forward reasoning calculation of convolutional neural network: First, the finger vein image of the user is decrypted and the encrypted user finger vein feature model Z in the encrypted database E As the input of the forward reasoning calculation of the convolutional neural network, the output result is then input into the Softmax activation function for calculation and encrypted user finger vein feature model Z m The similarity probability VS, if VS>δ, the system outputs the hash value SH of the encrypted user's finger vein image, and then performs quantum initial value mapping on SH to generate four sets of quantum initial values, namely Last pair The quantum matrix generation technology based on quantum walk is used to generate the decryption key required in the next decryption step, namely the finger vein quantum DNA decryption mask VQDM, the finger vein quantum DNA decryption index VQDI, and the first finger vein quantum conjugate phase mask. Second finger vein quantum conjugation phase mask Otherwise, if the authentication fails, the decryption is terminated; J2: Decrypted watermark image reconstruction based on Fresnel inverse diffraction and Gyrator transform: First, the encrypted light wave intensity I o (x,y), encrypted reference light wave intensity I r (x, y) and the DC component suppression technology based on the light intensity subtraction method suppresses the DC component of the ciphertext hologram I(x, y) to obtain the suppression result I ◇ (x,y), and the decryptor needs to input a set of correct digital keys to generate the decrypted structured phase mask Then use and phase modulation technology for the reference light wave G r (u,v) is phase modulated to obtain the decryption reference light wave G rd (u,v), then make I ◇ (x,y) and G rd After multiplying (u, v), a Fresnel inverse diffraction operation with a wavelength of λ and a diffraction distance of Z is performed on the multiplication result, and the optical information mask OIM is used to filter the inverse diffraction operation result to generate the decryption object light wave Finally, the first finger vein quantum conjugation phase mask is used Second finger vein quantum conjugation phase mask And the dual random phase decoding technology based on Gyrator transform domain Perform the decoding operation to obtain the decrypted watermark image W D (x,y); J3: Decrypted watermark image extraction based on DNA decoding and discrete wavelet transform: First, the watermark image extraction technology based on discrete wavelet transform is used to extract the decrypted watermark image W D (x,y) performs extraction operation to obtain the extraction result Then VQDM and VQEI are used as decryption keys to Perform DNA-encoded image decryption operations to obtain decrypted results Last pair Perform image channel merging operations to obtain the final decryption result 2. The color image encryption method based on finger vein key and off-axis digital holography according to claim 1, characterized in that: The specific process of the encryption step S1 is as follows: (i) Encrypted finger vein image acquisition based on infrared camera: Use infrared camera to collect encrypted user's finger vein image Z(x,y); (ii) Generation of encrypted user finger vein feature model based on Swin-Transformer network: Input Z(x, y) into the Swin-Transformer network to generate the encrypted user finger vein feature model Z E And save it to the encrypted user database; (iii) Finger vein hash data extraction and quantum initial value generation: First, the hash value SH of the encrypted user's finger vein image Z(x,y) is extracted using the fingerprint vein hash data extraction technology and saved in the encrypted user database. Next, quantum initial value mapping is performed on SH to obtain four sets of quantum initial values. (iv) Quantum matrix generation based on quantum walk system: First, four sets of quantum initial values ​​are obtained using quantum initial value mapping. Constrain the motion trajectories of the four walkers respectively, where It is used to constrain the walking range of the walker. After being constrained, the walking range of the walker in the two-dimensional rectangular coordinate system (ξ,η) is Used to constrain the number of steps taken by the walker; Used to constrain the initial state of the coin After being constrained It is represented by a linear combination of the state vector |κ>(κ=0,1) bundles: Used to constrain the coin operator After being constrained for: Next, control the Walker within range D n (n=1,2,3,4) Walking inside steps, and the walker needs to use the Adjust to the coin state, so each step of the walker uses the unitary operator To express: in, represents the tensor operation, H represents the unit operator in the unitary operator, and They are respectively expressed as the displacement operators of the nth (n=1, 2, 3, 4) walker in the ξ direction and η direction of the two-dimensional rectangular coordinate system: Where mod is the modulus operator, |ξ,τ,0> and |ξ,τ,1> represent the state vector of the walker after walking τ steps in the positive and negative directions of ξ, |τ,η,0> and |τ,η,1> represent the state vector of the walker after walking τ steps in the positive and negative directions of η, and <ξ,η,0| and <ξ,η,1|) are the conjugates of the state vectors |ξ,η,0> and |ξ,η,1> in the positive and negative directions of ξ and η, respectively. The nth (n=1,2,3,4)th walker walks Status after step With the initial system state ψ> 0,n The evolution relationship can be expressed as: in Unitary operator The nth (n=1,2,3,4)th walker passes The probability matrix of arriving at the two-dimensional space (X, Y) after one step is: in Mobile for walkers The unitary operator after the step; finally, the probability matrix They are arranged into two-dimensional matrices and then normalized to obtain the encryption keys required in the next encryption step, namely the finger vein quantum DNA mask VQEM, the finger vein quantum DNA encryption scrambled index VQEI, the first finger vein quantum phase mask VQPM1, and the second finger vein quantum phase mask VQPM2.

3. The color image encryption method based on finger vein key and off-axis digital holography according to claim 1, characterized in that: The specific process of the encryption step S2 is as follows: (i) DNA-encoded color plaintext image encryption: First, the color plaintext image P(x, y) is subjected to channel separation to obtain the red channel data P R (x, y), green channel component P G (x, y), blue channel component P B (x,y), then P R (x,y),P G (x,y),P B (x, y) and VQEM and VQEI in the encrypted database perform DNA encoding operations to obtain the binary encoding result P RD (α,β),P GD (α,β),P BD (α,β), VQEM D 、VQEI D , then P RD (α,β),P GD (α,β),P BD (α, β) and VQEM D Perform the XOR operation to obtain the XOR result P RK (α,β),P GK (α,β),P BK (α,β): Where XOR(·) represents an exclusive OR operation. Then VQEI D As the scrambling index, RK (α,β),P GK (α,β),P BK (α, β) performs the scrambling operation to obtain the scrambling result P RE (α,β),P GE (α,β),P BE (α,β): Among them, SO(·) represents the scrambling operation. Finally, P RE (α,β),P GE (α,β),P BE (α, β) performs DNA reverse encoding operation to obtain DNA encryption result P RR (x,y),P GR (x,y),P BR (x,y); (ii) Encrypted watermark image generation based on wavelet transform: First, discrete wavelet transform is performed on the host image W(x,y) to decompose W(x,y) into four sub-band images: LL(x,y), LH(x,y), HL(x,y), and HH(x,y). Then, LH(x,y), HL(x,y), and HH(x,y) are replaced by P RR (x,y),P GR (x,y),P BR (x, y), and finally perform inverse discrete wavelet transform on the replacement result to obtain the encrypted watermark image W E (x,y): Where DWT(·) represents discrete wavelet transform, IDWT(·) represents inverse discrete wavelet transform, Indicates a replace operation.

4. The color image encryption method based on finger vein key and off-axis digital holography according to claim 1, characterized in that: The specific process of the encryption step S3 is as follows: (i) Encryption light wave generation based on double random phase coding in Gyrator transform domain: First, the first finger vein quantum phase mask VQPM1, the second finger vein quantum phase mask VQPM2 and three sets of easy-to-remember transformation angle digital key pairs W are used. E (x,y) is encrypted by double random phase coding based on Gyrator transformation to obtain the encrypted light wave G o (u,v): in, Expressed as the transformation angle μ i Gyrator transform of (i=0,1,2). Taking the complex amplitude P of a certain input t-1 (u,v) as an example, after transformation the angle is μ t The output complex amplitude P obtained after Gyrator transformation t (u′,v′) is: Where exp(·) is the exponential operation and j is the imaginary unit; (ii) Off-axis digital holographic interferometry under reference light modulation: First, a set of simple digital keys is used to generate the structured phase mask SPM required for encryption: Where λ is the wavelength of the light wave, r is the radius of the Fresnel zone plate, f is the focal length of the Fresnel zone plate, and p is the topological charge of the radial Hilbert mask. is the spatial azimuth of the radial Hilbert mask, and arg(·) is the phase operation. Then, the reference light wave G is r (u,v) is phase modulated to obtain the encrypted reference light wave G re (u,v), and then record the encrypted reference light wave G re (u,v) and encrypted light wave G o (u,v) light intensity I o (x, y) and I r (x, y) and store them in the encrypted database respectively; finally, use the off-axis digital holographic technology to make G o (u,v) and G re (u,v) interferes to produce the ciphertext hologram I(x,y): I(x,y)=ODHE(FrT λ,Z (G o (u,v)),G re (u,v)) =|ZP(FrT λ,Z (G o (u,v)))+G r (u,v)×exp(jSPM)| 2 Among them, ZP(·) represents the zero-filling operation, FRT λ,Z (·) represents the Fresnel diffraction operation with a wavelength of λ and a distance of Z, and ODHE(·) represents the off-axis digital holographic encoding operation.

5. The color image encryption method based on finger vein key and off-axis digital holography according to claim 1, characterized in that: The specific process of the decryption step J1 is as follows: (i) Decrypted finger vein image acquisition based on infrared camera: Use infrared camera to collect and decrypt the user's finger vein image (ii) Decryption data acquisition based on finger vein authentication: First, decrypt the user's finger vein image and the encrypted user finger vein feature model Z in the encrypted database m As the input of the inference calculation based on the forward propagation principle of the convolutional neural network, the output result is then input into the Softmax activation function to calculate and encrypted user finger vein feature model Z E Similar probability VS: Where FP(·) represents the forward reasoning calculation process of the convolutional neural network, and Softmax(·) represents the Softmax activation function. Assume that the K-dimensional vector As the input of the Softmax activation function, the calculation result is output for: If VS is greater than the preset threshold δ, the system outputs the hash value SH of the encrypted user's finger vein image Z(x,y) and enters the next decryption step. Otherwise, the authentication fails and the decryption is terminated. (iii) Decryption key generation based on quantum initial value mapping and quantum walk: First, the same as sub-step (iii) in encryption step S1, the quantum initial value mapping is performed on SH to obtain four sets of quantum initial values Then will They are respectively used as the input of the quantum walk system, and the output results are arranged into a two-dimensional matrix and normalized to obtain the finger vein quantum DNA mask VQDM, the finger vein quantum DNA scrambled index VQEI, the first finger vein quantum phase mask VQPM1, and the second finger vein quantum phase mask VQPM2. Then, conjugate operations are performed on VQPM1 and VQPM2 to obtain the first finger vein quantum conjugate phase mask Second finger vein quantum conjugation phase mask Among them, conj(·) represents the conjugation operation. Finally, the finger vein quantum DNA mask VQDM, the finger vein quantum DNA scrambled index VQEI, the first finger vein quantum conjugated phase mask Second finger vein quantum conjugation phase mask As the decryption key in the next decryption step.

6. The color image encryption method based on finger vein key and off-axis digital holography according to claim 1, characterized in that: The specific process of the decryption step J2 is as follows: (i) DC component suppression based on light intensity subtraction: The encrypted hologram I(x, y) is compared with the encrypted object light wave intensity I in the encryption database. o (x, y), encrypted reference light wave intensity I r (x, y) can be subtracted to obtain the DC component suppression result I of I(x, y) ◇ (x, y): I ◇ (x,y)=I(x,y)-I o (x,y)-I r (x,y) (ii) Decryption light wave reconstruction based on Fresnel inverse diffraction: First, the decryption user needs to enter a set of correct digital keys and perform the same operation as sub-step (iii) in the encryption step S2 to obtain the decrypted structured phase mask. Then use For the reference light wave G r (u,v) is phase modulated to obtain the decryption reference light wave G rd (u,v); Next, I ◇ (x, y) and G rd (u,v) and perform a Fresnel inverse diffraction operation on the result with a wavelength of λ and a diffraction distance of Z to obtain the light wave containing the decryption object. The spectrum distribution is then filtered by constructing an optical information mask (OIM) to obtain the decrypted optical wave. Among them, IFrt λ，Z (·) is the Fresnel inverse diffraction with wavelength λ and diffraction distance Z; (iii) Decrypted watermark image reconstruction based on double random phase decoding in Gyrator transform domain: The first finger vein quantum conjugate phase mask Second finger vein quantum conjugation phase mask As the decryption phase key, Gyrator transforms the angle μ i (i=0,1,2) is used as the decryption digital key to decrypt the object light wave. The decrypted watermark image W is obtained by using the double random phase decoding operation based on the Gyrator transform domain. D (x,y):

7. The color image encryption method based on finger vein key and off-axis digital holography according to claim 1, characterized in that: The specific process of the decryption step J3 is as follows: (i) Decryption data extraction based on wavelet transform: Decryption watermark image W is extracted using decryption data extraction technology based on wavelet transform. D Decrypted data in (x,y) Among them, LL D To decrypt the watermark image W D Low-frequency information in the horizontal and vertical directions in (x,y); (ii) Decryption result acquisition based on DNA decoding: First, the decrypted data is obtained in the same manner as sub-step (i) in encryption step S2. Finger vein quantum DNA decryption mask VQDM, finger vein quantum DNA decryption index VQDI perform DNA encoding operation to obtain binary encoding result VQDI D 、VQDM D ; Then VQDI D As scrambling index, Perform the inverse scrambling operation to get the inverse scrambling result Among them, ISO(·) represents the inverse scrambling operation. VQDM D Perform an XOR operation to obtain the XOR result Then, XOR the results separately Perform DNA reverse encoding operation to obtain the red channel decryption result Green channel decryption results Blue channel decryption results Finally, the channel merging technology is used to Perform the merge operation to obtain the final decryption result

Images