Color medical image encryption method based on iris key and digital holography

Abstract

The invention discloses a color medical image encryption method based on an iris secret key and digital holography, and the method comprises the encryption steps: firstly generating an iris chaotic structure light phase mask through employing an iris feature extraction technology, a chaotic matrix generation program, and a phase modulation technology; then, an FRFCM algorithm, a grating modulation technology, a Fresnel domain double random phase coding technology based on chaotic structured light illumination and a digital holographic coding technology are utilized to encrypt the original color medical plaintext image, so that three amplitude type holographic ciphertext graphs are generated; and a decryption step: firstly, carrying out iris authentication on the decryption user and the encryption user, if the authentication is successful, decrypting the ciphertext hologram by using iris high-order mask data of the encryption user so as to generate a correct decryption result, and otherwise, stopping decryption if the authentication fails. The color medical image encryption method provided by the invention has the advantages of high security, good iris key robustness, high digital key sensitivity, small calculation burden, large key space and high decryption quality.

Description

Technical Field

[0001] The present invention relates to the technical field of information security, and particularly to a color medical image encryption method based on iris key and digital holography. Background Art

[0002] With the rapid development of information technology and medical conditions, the risk of color medical images being stolen or tampered with by illegal attackers is increasing day by day. To protect the sensitive information in color medical images, domestic and foreign research scholars have conducted extensive research on color image encryption technologies based on the spatial domain, transform domain, and optical methods. Among them, the optical method has attracted extensive attention from domestic and foreign research groups due to its characteristics of fast encryption speed, high security, high parallelism, rich key types, and the ability to process large-capacity images. The most representative one is the double random phase encoding (DRPE) technology based on the 4f optical system proposed by Refregier and Javidi in 1995. Subsequently, color image encryption methods based on advanced optical technologies such as interference, diffraction imaging, single-pixel imaging, and computer-generated holography have been gradually proposed by various research groups. For example, Chen et al. proposed a color image encryption method based on the Arnold transform and the interference principle. Qin et al. proposed a color image encryption method based on diffraction imaging. Qu et al. proposed a color image encryption method based on the Arnold transform and single-pixel imaging. Liu et al. proposed a color image encryption method based on chaos theory and computer-generated holography. Later, for medical color images, El-Shafai et al. proposed an asymmetric medical color image encryption method based on DRPE in the fractional Fourier transform domain. In addition, Alqahtani et al. also proposed a color medical image encryption method based on the Jigsaw transform and the fractional Fourier transform.

[0003] However, in the above-mentioned optical encryption methods for color images or medical images, there is generally a problem of high key complexity. In particular, random mask keys are difficult to memorize or carry. At the same time, there is a lack of a reliable correlation between the key mechanism and the user identity. Once the key is lost or stolen, illegal attackers can take advantage of this vulnerability to illegally obtain sensitive data in the system, thus seriously threatening information security. Biometric keys, especially iris feature keys, are effective means to solve the above problems because of their unique advantages of high authentication accuracy, non-contact acquisition, and portability. Therefore, research groups at home and abroad have begun to try to introduce iris feature keys into optical color image encryption technology in recent years and have made useful research. For example, Abuturab et al. proposed a color image encryption method based on iris keys and unequal spectral decomposition (USD). Later, Abuturab et al. also proposed a color image encryption method based on iris keys, chaotic mapping, and QR decomposition. However, at present, there are few reports on medical color image encryption methods based on iris feature keys.

[0004] On the other hand, the iris keys in the above-mentioned optical encryption methods all have the problem of poor robustness. Specifically, when there is a slight deviation in the acquisition distance during the decryption acquisition of the iris key of a legitimate user, or there is interference from wearing glasses, or there is a slight tilt in the fixation point, or there is a slight light attack, the security system will misjudge this decrypted iris as the wrong iris of an illegal attacker, resulting in incorrect decryption. This seriously affects the decryption experience of legitimate decryption users. In addition, most of the existing optical encryption technologies have problems of large computational load and inconvenient storage of ciphertexts. To solve the above problems, the present invention discloses a color medical image encryption method based on iris keys and digital holography, aiming to achieve color medical image encryption with strong encryption security, small computational burden, easy storage of ciphertexts, and high robustness of iris keys. Summary of the Invention

[0005] The object of the present invention is to provide a color medical image encryption method based on iris keys and digital holography. The present invention uses the high-order information of the iris of the encrypted user as the encryption key and uses the fuzzy C-means clustering algorithm (FRFCM), grating modulation technology, double random phase encoding technology under chaotic structured light illumination, and digital holographic encoding technology to encrypt the color medical image, having the advantages of a large key space, high sensitivity of digital keys, fast encryption speed, small computational burden, and strong robustness of iris keys. To achieve the above object of the invention, the technical solution provided by the present invention is as follows: A color medical image encryption method based on iris keys and digital holography, characterized in that: it includes an encryption step:

[0006] S1: Encryption key generation based on iris high-order data acquisition: First, use the iris acquisition device to collect the original iris image Z(x,y) of the encrypted user; then use iris image processing technology and iris feature extraction technology based on fully convolutional networks to obtain the high-order mask information Z of the two encrypted iris images of the encrypted user F (x,y) and Z M (x, y), and use image hash value extraction technology to obtain Z F (x,y),Z M The decimal hash values ​​SH1 and SH2 of (x, y) are obtained and stored in the encryption database; finally, the chaotic matrix generation program and the structured light phase mask generation algorithm are used to generate the iris chaotic structured light phase masks ICSPM1 and ICSPM2 required in the encryption step S3 for SH1 and SH2 respectively.

[0007] S2: Plaintext image reconstruction based on FRFCM algorithm and grating modulation technology: First, the foreground area P of the color plaintext image P(x, y) to be encrypted is extracted using the FRFCM algorithm and image processing technology. ROI (x,y); then P ROI (x, y) performs image reshaping operation to obtain the encrypted reshaped image P of size m×m SHA (x,y); then P SHA (x, y) performs image channel separation operation to obtain P SHA The red component data P of (x,y) SR (x, y), green component data P SG (x, y), blue component data P SB (x, y); finally, the grating modulation technology is used to SR (x,y),P SG (x,y),P SB (x, y) performs a fusion operation to obtain an encrypted fusion result G(x, y).

[0008] S3: Image encryption based on chaotic structured light illumination and digital holographic coding: First, ICSPM1, ICSPM2 and double random phase coding encryption technology are used to perform encryption operations on G(x, y) to obtain the encrypted light wave L o (η,ξ); then using the reference light wave L r (η,ξ) and phase-shift digital holographic coding technology to convert L o (η,ξ) is encoded into three real amplitude holograms I containing color plaintext information ν (ν=1,2,3) and output it as the final ciphertext.

[0009] The encryption step is now complete.

[0010] Decryption steps:

[0011] J1: Obtaining the decryption key based on iris feature authentication: First, use an iris acquisition device to collect the original iris image of the decryption user Subsequently, use iris image processing technology and iris feature extraction technology based on a fully convolutional network to obtain the high-order mask information of two iris images of the decryption user and Then, calculate the iris correlation CoI corresponding to the decryption user and the encryption user. If CoI is greater than the preset threshold δ, the authentication is successful. The system will output SH1 and SH2 stored in the encrypted database and perform the same operations as the encryption step S1 to generate the iris chaotic structured light phase masks ICSPM1 and ICSPM2. Subsequently, perform a conjugation operation on ICSPM1 and ICSPM2 to generate the iris chaotic structured light conjugate phase mask required for the next decryption step and Otherwise, the authentication fails and the decryption is aborted

[0012] J2: Image decryption based on digital holographic decoding and chaotic structured light illumination: First, use the digital holographic decoding algorithm to perform a decoding operation on three holographic ciphertext holograms to obtain the decrypted object wave Subsequently, use and as the first decryption phase key and the second decryption phase key respectively, and use the double random phase decoding algorithm to perform a decryption operation on to obtain the decrypted fusion result

[0013] J3: Decrypted image reconstruction based on spectral filtering and image reshaping technology: First, use spectral filtering technology to perform an image decomposition operation on to obtain the reshaped decrypted image red component decryption result green component decryption result blue component decryption result Subsequently, perform an image channel merging operation on to obtain the reshaped decrypted image Finally, perform an image reshaping operation and an image zero-padding operation on to obtain the final decryption result

[0014] Thus, the decryption steps are completed

[0015] The above-mentioned color medical image encryption method based on iris key and digital holography, wherein the encryption step S1 is generated based on the encryption key obtained from iris high-order data, and includes the following steps:

[0016] (i) Acquisition and preprocessing of encrypted iris images: First, use an iris acquisition device to collect the original iris image Z(x, y) of the encrypted user. Subsequently, perform grayscale conversion, localization, enhancement, and stretching operations on Z(x, y) in sequence to obtain the encrypted iris normalized image Z N (x, y).

[0017] (ii) Extraction of high-order features of encrypted iris based on fully convolutional network: Input Z N (x, y) into a fully convolutional network composed of an iris feature extraction network (FeatNet) and an iris region recognition network (MaskNet) to obtain the high-order mask information Z F (x, y) and Z M (x, y):

[0018]

[0019] Among them, the structure of FeatNet includes a 3×7 convolutional layer (C1), a 3×5 convolutional layer (C2), two 3×3 convolutional layers (C3, C4), two pooling layers (P1, P2), and three activation functions (T1, T2, T3). The structure of MaskNet includes four 3×3 convolutional layers (C1, C2, C3, C4), three 1×1 convolutional layers (C2_S, C3_S, C4_S), two 2×2 pooling layers (P1, P2), and a 4×4 pooling layer (P3).

[0020] (iii) Generation of encrypted iris chaotic structured light phase masks: First, calculate the hash values of Z F (x, y) and Z M (x, y) respectively, and then convert them into decimal data to obtain the conversion results SH1 and SH2, and store them in the encrypted database. Subsequently, use SH1 and SH2 as the chaotic initial values respectively, where the general term SH r(n) is:

[0021]

[0022] Among them, SH 1(n) = SH n and n = 1, 2. Secondly, arrange the chaotic sequence into a two-dimensional matrix to generate the iris chaotic matrix ICM n (n = 1, 2). Immediately afterwards, the encrypted user constructs a Fresnel zone plate FZP and a radial Hilbert mask RHM using a set of simple digital keys:

[0023]

[0024] Among them, 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, p is the topological charge number of the radial Hilbert mask, is the spatial azimuth angle of the radial Hilbert mask, j is the imaginary unit, and arg(·) is the phase extraction operation. Finally, perform phase modulation operations on ICM n (n = 1, 2), FZP, and RHM to generate the iris chaotic structured light phase mask ICSPM n (n = 1, 2): ICSPM n = exp(j(ICM n × FZP × RHM))

[0025] The above-mentioned color medical image encryption method based on iris key and digital holography, the encryption step S2 is the reshaping of the plaintext image based on the FRFCM algorithm and grating modulation technology, including the following steps:

[0026] (i) Extraction of foreground information of the plaintext image based on the FRFCM algorithm: First, perform graying and morphological reconstruction operations on the original color plaintext image P(x, y) to obtain the morphological reconstruction result P Δ (x, y). Subsequently, perform the FRFCM algorithm with the number of clusters c, the fuzzification parameter s, and the minimum error threshold τ on P Δ (x, y), so as to obtain the segmentation result P SEG (x, y) of P(x, y):

[0027] P SEG (x, y) = FRFCM c,s,τ (P Δ (x, y))

[0028] = FRFCM c,s,τ (Mcr(Gry(P(x, y)))

[0029] Among them, Mcr(·), Gry(·), and FRFCM c,s,τ respectively represent the morphological reconstruction operation, the graying operation, and the FRFCM algorithm with the number of clusters c, the fuzzification parameter s, and the minimum error threshold τ. Subsequently, perform a clustering operation on P SEG (x, y) to obtain the mask MASK of the color plaintext image P(x, y):

[0030]

[0031] Finally, perform a pixel-by-pixel multiplication operation through MASK and P(x, y) to extract the foreground region P ROI (x, y) of the plaintext image:

[0032] P ROI(x, y) = MASK × P(x, y)

[0033] (ii) Plaintext image reshaping based on grating modulation: First, use pixel extraction technology to extract each pixel x1, x2,..., x ROI in P(x, y), where N = P × Q, and P and Q are the horizontal and vertical pixel numbers of P(x, y) respectively. Subsequently, use image reshaping technology to rearrange x1, x2,..., x N (N = P × Q) into an encrypted reshaped image P ROI (x, y) with a size of N . Subsequently, perform an image channel separation operation on P (x, y) to obtain the red component data P SHA (x, y), green component data P SHA (x, y), and blue component data P SHA (x, y) of P(x, y). Finally, perform a grating modulation operation on P SR (x, y), P SG (x, y), and P SB (x, y) to obtain the encrypted fusion result G(x, y): SR (x, y), P SG (x, y), P SB (x, y) to obtain the encrypted fusion result G(x, y):

[0034]

[0035] where K n (x, y) is the grating coefficient, and its mathematical expression is:

[0036]

[0037] where f n is the grating spatial frequency, θ xn is the angle between the grating line direction and the horizontal direction, and θ yn is the angle between the grating bar direction and the vertical direction, and n = 1, 2, 3.

[0038] The above-mentioned color medical image encryption method based on iris key and digital holography, wherein the encryption step S3 is based on image encryption by chaotic structured light illumination, and includes the following steps:

[0039] (i) Double random phase encoding based on chaotic structured light illumination: First, perform a Fresnel diffraction transform on the encrypted fusion result G(x,y) with a wavelength of λ and a diffraction distance of Z1. Subsequently, multiply the transformed result by the first iris chaotic structured light phase mask ICSPH1. Immediately after that, perform a Fresnel diffraction transform on the multiplied result with a wavelength of λ and a diffraction distance of Z2. Then multiply this transformed result by the second iris chaotic structured light phase mask ICSPH2. Finally, perform a Fresnel diffraction transform on this multiplied result again with a wavelength of λ and a diffraction distance of Z3 to obtain the encrypted object wave L o (η,ξ):

[0040]

[0041] Among them, FrT λ,Z (·) represents the Fresnel diffraction transform with a wavelength of λ and a diffraction distance of Z.

[0042] (ii) Generation of holographic ciphertext based on phase-shifting digital holographic encoding: First, introduce an axial plane reference wave L o (η,ξ) that is coaxial with the encrypted object wave L r (η,ξ) and perpendicular to the recording plane. Subsequently, change the phase shift amount of the reference wave three times and make L o (η,ξ) and L r (η,ξ) interfere on the recording plane to generate the holographic ciphertext I ν (x,y)(ν = 1,2,3):

[0043]

[0044] Among them, |·| represents the modulo operation with a wavelength, and L r (η,ξ) * , L o (η,ξ) * respectively represent the conjugate of the reference wave L r (η,ξ) and the encrypted object wave L o (η,ξ).

[0045] The above-mentioned color medical image encryption method based on iris key and digital holography, the decryption step J1 for obtaining the decryption key based on iris feature authentication includes the following steps:

[0046] (i) Obtaining the decrypted original iris feature image: First, use an iris acquisition device to collect the original iris image of the decryption user Subsequently, perform grayscale conversion, localization, enhancement, and stretching operations on in sequence to obtain the decrypted iris normalized image

[0047] (ii) Iris high-order information authentication: First, input into the fully convolutional network composed of the iris feature extraction network (FeatNet) and the iris region recognition network (MaskNet) to obtain the decrypted iris image high-order mask information and Subsequently, use the decrypted iris image high-order mask information and the Z stored in the encrypted database F (x,y) and Z M (x,y) to calculate the iris correlation of interest COI between the decrypted user and the encrypted user:

[0048]

[0049] where Z F (i,j), Z M (i,j) are respectively represented as Z F (x,y), Z M (x,y) at the pixel value of (i,j), m and n respectively represent the horizontal pixel number and vertical pixel number of the iris image high-order mask information, and xor is the exclusive or operation. Finally, perform iris high-order information authentication by judging whether COI is greater than the preset threshold δ. If COI is greater than the preset threshold δ, the authentication is successful, and the SH1 and SH2 stored in the encrypted database are output as decrypted data and enter the next decryption step; otherwise, the authentication fails and the decryption terminates.

[0050] (iii) Iris chaotic structured light conjugate phase mask generation: Use the SH1 and SH2 obtained in the previous decryption step as the chaotic initial values respectively, and generate the iris chaotic matrices ICM1 and ICM2 using the same recurrence formula as in sub-step (iii) of the encryption step S1. Subsequently, the decrypted user needs to input a set of digital keys to construct the decrypted Fresnel zone plate and the decrypted radial Hilbert mask After that, use the phase modulation technology to perform phase modulation operations on ICM1, ICM2, and and perform a conjugation operation on the modulation result to obtain the iris chaotic structured light conjugate phase mask

[0051]

[0052] where conj(·) is the conjugation operation.

[0053] The above-mentioned color medical image encryption method based on iris key and digital holography, the decryption step J2 is based on image decryption of digital holographic decoding and chaotic structured light illumination, and includes the following steps:

[0054] (i) Generation of decrypted object wave based on digital holographic decoding: Use the digital holographic decoding algorithm to perform decoding operations on the holographic ciphertext I ν (x,y)(ν = 1,2,3) to obtain the decrypted object wave

[0055]

[0056] (ii) Double random phase decoding based on chaotic structured light illumination: First, perform an inverse Fresnel diffraction transform on the decrypted object wave with a wavelength of λ and a diffraction distance of and multiply the result of this transform with the conjugate phase mask of the second iris chaotic structured light Then, perform another inverse Fresnel diffraction transform on the product result with a wavelength of λ and a diffraction distance of Next, multiply the result of this transform with the conjugate phase mask of the first iris chaotic structured light Finally, perform an inverse Fresnel diffraction transform on the product result with a wavelength of λ and a diffraction distance of to obtain the decrypted fusion result

[0057]

[0058] where represents the inverse Fresnel diffraction with a wavelength of λ and a diffraction distance of

[0059] The above-mentioned color medical image encryption method based on iris key and digital holography, the decryption step J3 is based on decryption image reconstruction of spectral filtering and image reshaping technology, and includes the following steps:

[0060] (i) Generation of decrypted reshaped image based on spectral filtering: By constructing a triple filter, perform spectral filtering operations on the decrypted fusion result to obtain the red component decryption result the green component decryption result and the blue component decryption result of the reshaped decrypted image

[0061] where FT(·) and IFT(·) respectively represent the Fourier transform operation and the inverse Fourier transform operation, and FIL χ is the triple filter. Then for​ Perform an image channel merging operation to obtain a reshaped decrypted image

[0062] (ii) Generation of the decryption result based on the image reshaping technique: First, extract each pixel in Subsequently, use the image reshaping technique to reshape into a decrypted foreground area with a size of P×Q Finally, perform zero-padding operation on to generate the final decryption result

[0063] The beneficial effects of the present invention are as follows: (1) The iris feature key in the present invention is convenient to carry, the digital key is easy to remember, and there is no random phase mask key, thus effectively reducing the complexity of the key; (2) The present invention uses the iris authentication algorithm based on the iris high-order feature to authenticate the iris features of the encrypted user and the decrypted user, significantly increasing the security and robustness of the iris key; (3) The present invention uses the image reshaping technique based on the FRFCM algorithm and grating modulation to convert the original color plaintext image into a grayscale image with a smaller size, thus effectively reducing the encrypted data volume and improving the encryption efficiency of the encryption algorithm. (4) The present invention uses the chaotic matrix generation program, Fresnel zone plate, and radial Hilbert mask to generate the iris chaotic structured light phase mask used in the encryption step, which can effectively increase the key space and the sensitivity of the digital key. (5) The present invention uses the digital holographic coding technique to finally encrypt the color plaintext image into three amplitude-type ciphertext images, which are easy to store and transmit. Brief Description of the Drawings

[0064] Figure 1 is a flowchart of the encryption step in a color medical image encryption method based on iris key and digital holography.

[0065] Figure 2 is a flowchart of the encryption sub-step S1 in a color medical image encryption method based on iris key and digital holography.

[0066] Figure 3 is a schematic diagram of the iris feature extraction network (FeatNet) and the iris region recognition network (MaskNet) in the encryption sub-step S1 of a color medical image encryption method based on iris key and digital holography.

[0067] Figure 4 is a flowchart of the encryption sub-step S2 in a color medical image encryption method based on iris key and digital holography.

[0068] Figure 5It is a flowchart of the encryption sub-step S3 in a color medical image encryption method based on iris key and digital holography.

[0069] Figure 6 It is a schematic diagram of an optical device for assisting in describing the encryption sub-step S3, where P is a polarizer, OL is an objective lens, PH is a pinhole, CL is a collimating lens, BS is a beam splitting prism, M is a mirror, and L is a lens.

[0070] Figure 7 It is a flowchart of the decryption step in a color medical image encryption method based on iris key and digital holography.

[0071] Figure 8 It is a flowchart of the decryption sub-step J1 in a color medical image encryption method based on iris key and digital holography.

[0072] Figure 9 It is a flowchart of the decryption sub-step J2 in a color medical image encryption method based on iris key and digital holography.

[0073] Figure 10 It is a flowchart of the decryption sub-step J3 in a color medical image encryption method based on iris key and digital holography.

[0074] Figure 11 (a) is the color medical plaintext image ("Brain"), (b) is the iris image of the encryption user, (c1)-(c2) are the iris chaotic structured light phase masks (ICSPM1, ICSPM2) generated according to (b), and (d) is the segmentation result P SEG (x,y) of the color plaintext image, (e) is the foreground region P ROI (x,y) of the color plaintext image, (f) is the reshaping result P SHA (x,y) of the color plaintext image, (g1)-(g3) are three different gratings for grating modulation respectively, (h) is the encryption fusion result G(x,y), and (i1)-(i2) are the encrypted object light waves L o (η,ξ) amplitude distribution A(x,y) and phase distribution (j1)-(j3) are the holographic ciphertexts I1(x,y), I2(x,y), and I3(x,y) respectively.

[0075] Figure 12 (a1)-(a3) are the decryption results of the red, green, and blue channels of the color medical image ("Brain") respectively when all keys are correct, and (a4) is the final decryption result of the color medical image ("Brain").

[0076] Figure 13 It is the decryption result when the iris key is incorrect.

[0077] Figure 14 (a) Robustness test results when the iris key is subjected to the acquisition distance deviation attack, (b) Robustness test results when the iris key is subjected to glasses interference, (c) Robustness test results when the iris key is subjected to the fixation point attack, (d) Robustness test results when the iris key is subjected to the illumination attack.

[0078] Figure 15 (a) Sensitivity curve of the first diffraction key Z1, (b) Sensitivity curve of the second diffraction key Z2, (c) Sensitivity curve of the third diffraction key Z3, (d) Sensitivity curve of the focal length key f, (e) Sensitivity curve of the topological charge number key p, (f) Sensitivity curve of the wavelength key λ, (g) Multivariate function relationship between the average correlation coefficient between the decryption result and the original plaintext image and the first diffraction distance key deviation ΔZ1, the second diffraction distance key deviation ΔZ2, the third diffraction distance key deviation ΔZ3, the focal length key deviation Δf, the topological charge number key deviation Δp, and the wavelength key deviation Δλ when the iris key is correct. Detailed implementation manner

[0079] The present invention will be further described below in conjunction with the embodiments and the drawings, but it is not used as a basis for limiting the present invention.

[0080] Embodiment: A color medical image encryption method based on an iris key and digital holography, including the following Figure 1 encryption steps shown:

[0081] S1: Generation of an encryption key based on the acquisition of iris high-order data: First, use an iris acquisition device to acquire the original iris image Z(x,y) of the encrypted user; subsequently, use iris image processing technology and iris feature extraction technology based on a fully convolutional network to obtain two encrypted iris image high-order mask information Z F (x,y) and Z M (x,y) of the encrypted user, and use image hash value extraction technology to respectively obtain the decimal hash values SH1 and SH2 of Z F (x,y) and Z M (x,y) and store them in the encryption database; finally, use the chaotic matrix generation program and the structured light phase mask generation algorithm for SH1 and SH2 respectively to generate the iris chaotic structured light phase masks ICSPM1 and ICSPM2 required in the encryption step S3. The flowchart is as Figure 2 shown and includes the following sub-steps:

[0082] (i) Acquisition and preprocessing of encrypted iris images: First, use an iris acquisition device to collect the original iris image Z(x,y) of the encrypted user. Subsequently, perform grayscale conversion, localization, enhancement, and stretching operations on Z(x,y) in sequence to obtain the encrypted iris normalized image Z N (x,y).

[0083] (ii) Extraction of high-order features of encrypted iris based on fully convolutional network: Input Z N (x,y) into the fully convolutional network composed of an iris feature extraction network (FeatNet) and an iris region recognition network (MaskNet) as shown to obtain the high-order mask information Z Figure 3 (x,y) and Z F (x,y): M (x,y):

[0084]

[0085] Among them, the structure of FeatNet includes a 3×7 convolutional layer (C1), a 3×5 convolutional layer (C2), two 3×3 convolutional layers (C3, C4), two pooling layers (P1, P2), and three activation functions (T1, T2, T3). The structure of MaskNet includes four 3×3 convolutional layers (C1, C2, C3, C4), three 1×1 convolutional layers (C2_S, C3_S, C4_S), two 2×2 pooling layers (P1, P2), and a 4×4 pooling layer (P3).

[0086] (iii) Generation of encrypted iris chaotic structured light phase mask: First, calculate the hash values of Z F (x,y) and Z M (x,y) respectively, and then convert them into decimal data to obtain the conversion results SH1 and SH2, and store them in the encrypted database. Subsequently, use SH1 and SH2 as the chaotic initial values respectively, where the general term SH r(n) is:

[0087]

[0088] Among them, SH 1(n) = SH n and n = 1, 2. Secondly, arrange the chaotic sequence into a two-dimensional matrix to generate the iris chaotic matrix ICM n (n = 1, 2). Immediately afterwards, the encrypted user constructs a Fresnel zone plate FZP and a radial Hilbert mask RHM using a set of simple digital keys:

[0089]

[0090] Among them, 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, p is the topological charge number of the radial Hilbert mask, is the spatial azimuth angle of the radial Hilbert mask, j is the imaginary unit, and arg(·) is the phase extraction operation. Finally, perform phase modulation operations on ICM n (n = 1, 2), FZP, and RHM to generate the iris chaotic structured light phase mask ICSPM n (n = 1, 2): ICSPM n = exp(j(ICM n ×FZP×RHM))

[0091] S2: Plaintext image reshaping based on the FRFCM algorithm and grating modulation technology: First, use the FRFCM algorithm and image processing technology to extract the foreground region P of the color plaintext image P(x, y) to be encrypted ROI (x, y); Subsequently, perform an image reshaping operation on P ROI (x, y) to obtain the encrypted reshaped image P of size m×m SHA (x, y); Immediately afterwards, perform an image channel separation operation on P SHA (x, y) to obtain the red component data P of P SHA (x, y), the green component data P SR (x, y), and the blue component data P SG (x, y); Finally, use the grating modulation technology to perform a fusion operation on P SB (x, y), P SR (x, y), and P SG (x, y) to obtain the encrypted fusion result G(x, y). Its flowchart is as SB shown, including the following sub-steps: Figure 4 shown, including the following sub-steps:

[0092] (i) Extraction of foreground information of the plaintext image based on the FRFCM algorithm: First, perform grayscale conversion and morphological reconstruction operations on the original color plaintext image P(x, y) to obtain the morphological reconstruction result P Δ (x, y). Subsequently, perform the FRFCM algorithm with the number of clusters being c, the fuzzification parameter being s, and the minimum error threshold being τ on P Δ (x, y) to obtain the segmentation result P of P(x, y) SEG (x, y):

[0093] P SEG (x, y) = FRFCM c,s,τ (P Δ (x, y))

[0094] = FRFCM c,s,τ(Mcr(Gry(P(x,y)))

[0095] Among them, Mcr(·), Gry(·), and FRFCM c,s,τ respectively represent the morphological reconstruction operation, the grayscale operation, and the FRFCM algorithm with the number of clusters being c, the fuzzification parameter being s, and the minimum error threshold being τ. Subsequently, perform a clustering operation on P SEG (x,y) to obtain the color plaintext image P(x,y) mask MASK:

[0096]

[0097] Finally, perform a pixel-by-pixel multiplication operation through MASK and P(x,y) to extract the foreground region P of the plaintext image ROI (x,y):

[0098] P ROI (x,y) = MASK × P(x,y)

[0099] (ii) Reshaping of the plaintext image based on raster modulation: First, use the pixel extraction technique to extract each pixel x1, x2,..., x ROI in P(x,y) (N = P × Q), where P and Q are the horizontal and vertical pixel numbers of P N (x,y) respectively. Subsequently, use the image reshaping technique to rearrange x1, x2,..., x ROI (N = P × Q) into the encrypted reshaped image P N (x,y) with the size of . Subsequently, perform an image channel separation operation on P SHA (x,y) to obtain the red component data P SHA (x,y), the green component data P SHA (x,y), and the blue component data P SR (x,y), P SG (x,y), P SB (x,y). Finally, perform a raster modulation operation on P SR (x,y), P SG (x,y), P SB (x,y) to obtain the encrypted fusion result G(x,y):

[0100]

[0101] Among them, K n (x,y) is the raster coefficient, and its mathematical expression is:

[0102]

[0103] Among them, fn is the grating spatial frequency, θ xn is the angle between the grating line direction and the horizontal direction, θ yn is the angle between the grating line direction and the vertical direction, n = 1, 2, 3.

[0104] S3: Image encryption based on chaotic structured light illumination and digital holographic coding: First, use the ICSPM1, ICSPM2, and double random phase encoding encryption techniques to perform an encryption operation on G(x, y) to obtain the encrypted object light wave L o (η, ξ); Subsequently, use the reference light wave L r (η, ξ) and the phase-shifting digital holographic coding technique to encode L o (η, ξ) into three real-amplitude holograms I ν (ν = 1, 2, 3) containing color plaintext information, and output them as the final ciphertext. The flow chart and optical path schematic diagram are respectively as Figure 5 and Figure 6 shown, including the following sub-steps:

[0105] (i) Double random phase encoding based on chaotic structured light illumination: First, perform a Fresnel diffraction transform on the encrypted fusion result G(x, y) with a wavelength of λ and a diffraction distance of Z1. Subsequently, multiply this transformation result by the first iris chaotic structured light phase mask ICSPH1. Immediately afterwards, perform a Fresnel diffraction transform on the multiplication result with a wavelength of λ and a diffraction distance of Z2. Immediately afterwards, multiply this transformation result by the second iris chaotic structured light phase mask ICSPH2. Finally, perform a Fresnel diffraction transform on this multiplication result again with a wavelength of λ and a diffraction distance of Z3 to obtain the encrypted object light wave L o (η, ξ):

[0106]

[0107] where, FrT λ,Z (·) represents the Fresnel diffraction transform with a wavelength of λ and a diffraction distance of Z.

[0108] (ii) Holographic ciphertext generation based on phase-shifting digital holographic coding: First, introduce an axial plane reference light wave L o (η, ξ) coaxial with the encrypted object light wave L r (η, ξ) and perpendicular to the recording plane. Subsequently, by changing the phase shift amount of the reference light wave three times and making L o (η, ξ) and L r (η, ξ) interfere on the recording plane, thereby generating the holographic ciphertext I ν (x, y) (ν = 1, 2, 3):

[0109]

[0110] Among them, |·| represents the modulo operation for the wavelength, L r (η, ξ) * , L o (η, ξ) * respectively represent the reference light wave L r (η, ξ) and the conjugate of the encrypted object light wave L o (η, ξ).

[0111] So far, the encryption step is completed.

[0112] A color medical image encryption method based on iris key and digital holography includes the Figure 7 shown decryption steps:

[0113] J1: Obtaining the decryption key based on iris feature authentication: First, use the iris acquisition device to collect the original iris image of the decryption user Subsequently, use iris image processing technology and iris feature extraction technology based on the fully convolutional network to obtain the high-order mask information of two iris images of the decryption user and Immediately calculate the iris correlation CoI corresponding to the decryption user and the encryption user. If CoI is greater than the preset threshold δ, the authentication is successful. The system will output SH1 and SH2 stored in the encrypted database and perform the same operation as the encryption step S1 to generate the iris chaotic structured light phase masks ICSPM1 and ICSPM2. Subsequently, perform a conjugation operation on ICSPM1 and ICSPM2 to generate the iris chaotic structured light conjugate phase masks required for the next decryption step and Otherwise, the decryption is aborted if the authentication fails. The flowchart is as Figure 8 shown and includes the following sub-steps:

[0114] (i) Obtaining the decrypted iris original feature image: First, use the iris acquisition device to collect the original iris image of the decryption user Subsequently, perform grayscale conversion, localization, enhancement, and stretching operations on in sequence to obtain the decrypted iris normalized image

[0115] (ii) Iris high-order information authentication: First, input into the fully convolutional network composed of the iris feature extraction network (FeatNet) and the iris region recognition network (MaskNet) to obtain the high-order mask information of the decrypted iris image and Subsequently, use the high-order mask information of the decrypted iris image and Z stored in the encrypted database F(x, y) and Z M (x, y) calculates the iris correlation of interest (COI) between the decryption user and the encryption user:

[0116]

[0117] Where Z F (i, j), Z M (i, j) are respectively expressed as Z F (x, y), Z M (x, y) at the pixel value at (i, j), m and n respectively represent the horizontal and vertical pixel numbers of the high-order mask information of the iris image, and xor is the exclusive OR operation. Finally, iris high-order information authentication is performed by determining whether the COI is greater than the preset threshold δ. If the COI is greater than the preset threshold δ, the authentication is successful, and the stored SH1 and SH2 in the encrypted database are output as decryption data and enter the next decryption step; otherwise, the authentication fails and the decryption terminates.

[0118] (iii) Generation of iris chaotic structured light conjugate phase mask: Take the SH1 and SH2 obtained in the previous decryption step as the chaotic initial values respectively, and use the same recurrence formula as the sub-step (iii) in the encryption step S1 to generate the iris chaotic matrices ICM1 and ICM2. Subsequently, the decryption user needs to input a set of digital keys to construct the decryption Fresnel zone plate and the decryption radial Hilbert mask After that, use the phase modulation technology to perform phase modulation operations on ICM1, ICM2, and and perform a conjugation operation on the modulation result to obtain the iris chaotic structured light conjugate phase mask

[0119]

[0120] where conj(·) is the conjugation operation.

[0121] J2: Image decryption based on digital holographic decoding and chaotic structured light illumination: First, use the digital holographic decoding algorithm to perform decoding operations on three holographic ciphertext holograms to obtain the decrypted object wave Subsequently, and are respectively used as the first decryption phase key and the second decryption phase key, and the double random phase decoding algorithm is used to perform decryption operations on to obtain the decrypted fusion result Its flowchart is as Figure 9 shown, including the following sub-steps:

[0122] (i) Generation of the decrypted object wave based on digital holographic decoding: Use the digital holographic decoding algorithm to perform decoding operations on the holographic ciphertext I ν (x,y)(ν = 1,2,3) to obtain the decrypted object wave

[0123]

[0124] (ii) Double random phase decoding based on chaotic structured light illumination: First, perform an inverse Fresnel diffraction transform on the decrypted object wave with a wavelength of λ and a diffraction distance of and multiply the result of this transform with the conjugate phase mask of the second iris chaotic structured light . Subsequently, perform another inverse Fresnel diffraction transform on the product result with a wavelength of λ and a diffraction distance of . Immediately after that, multiply the result of this transform with the conjugate phase mask of the first iris chaotic structured light . Finally, perform an inverse Fresnel diffraction transform on the product result of this time with a wavelength of λ and a diffraction distance of to obtain the decrypted fusion result

[0125]

[0126] where represents the inverse Fresnel diffraction with a wavelength of λ and a diffraction distance of .

[0127] J3: Decrypted image reconstruction based on spectral filtering and image reshaping technology: First, use spectral filtering technology to perform image decomposition operations on to obtain the decrypted result of the red component of the reshaped decrypted image the decrypted result of the green component the decrypted result of the blue component Subsequently, perform image channel merging operations on to obtain the reshaped decrypted image Finally, perform image reshaping operations and image zero-padding operations on to obtain the final decrypted result The flowchart is as shown in Figure 10 and includes the following steps:

[0128] (i) Generation of the decrypted reshaped image based on spectral filtering: By constructing a triple filter, perform spectral filtering operations on the decrypted fusion result to obtain the decrypted result of the red component of the reshaped decrypted image the decrypted result of the green component Decryption result of the blue component

[0129] wherein, FT(·) and IFT(·) respectively represent the Fourier transform operation and the inverse Fourier transform operation, and FIL χ is a triple filter. Subsequently, perform an image channel merging operation to obtain the reshaped decrypted image

[0130] (ii) Generation of the decryption result based on the image reshaping technique: First, extract each pixel in Subsequently, use the image reshaping technique to reshape into a decrypted foreground region with a size of P×Q Finally, perform a zero-padding operation to generate the final decryption result

[0131] The content of the present invention will be further explained below with reference to the accompanying drawings:

[0132] First, select a color medical image ("Brain") with 256×256 pixels as shown in Figure 11 (a) as the plaintext image P(x,y) to be encrypted, and perform an encryption operation on it according to Figure 1 the encryption steps shown. First, use an iris image acquisition device to collect the original iris image Z(x,y) of the encrypted user (as shown in Figure 11 (b)), and then perform grayscale conversion, localization, enhancement, and stretching operations on Z(x,y) in sequence and input the processing result into the iris feature extraction network (FeatNet) and the iris region recognition network (MaskNet) as shown in Figure 3 . Immediately afterwards, use the image hash value extraction technology to respectively obtain the decimal hash values (SH1 = 34, SH2 = 43) of the output results of FeatNet and MaskNet. Next, take SH1 and SH2 as the initial chaos values respectively and use the chaos matrix generation program to generate the iris chaos matrices ICM1 and ICM2. Subsequently, set the light wave wavelength λ = 632.8 nm, the focal length f of the Fresnel zone plate = 40 mm, and the topological charge number p = 6 to construct the Fresnel zone plate FZP and the radial Hilbert mask RHM. Then, perform a phase modulation operation on ICM1, ICM2, FZP, and RHM to generate the first iris chaos structured light phase mask ICSPM1 and the second iris chaos structured light phase mask ICSPM2 as shown in (c1)-(c2).

[0133] Secondly, P(x,y) is successively subjected to grayscale conversion and morphological reconstruction operations, and a FRFCM segmentation operation with the number of clusters c = 3, fuzzification parameter m = 2, and minimum error threshold τ = 10 is performed on the reconstruction result to obtain the segmentation result P -5 as shown in Figure 11 (d), namely P SEG (x,y). Then, a clustering operation is performed on P SEG (x,y) to obtain the mask image MASK of P(x,y). Subsequently, a pixel-by-pixel multiplication operation is performed on MASK and P(x,y) to obtain the foreground region P Figure 11 of P(x,y) as shown in ROI (e), namely P ROI (x,y). Then, an image reshaping operation is performed on P Figure 11 (x,y) to obtain the encrypted reshaped image P SHA (x,y) with 152×152 pixels as shown in Figure 11 (f). Immediately afterwards, the encrypted fusion result G(x,y) is obtained by performing a grating modulation operation on P SHA (x,y) using three different gratings as shown in Figure 11 (g1)-(g3). Next, a Fresnel diffraction transform operation with wavelength λ = 632.8 nm and diffraction distance Z = 100 mm is performed on G(x,y), and the result of this operation is multiplied by the first iris chaotic structured light phase mask ICSPM1. Subsequently, a Fresnel diffraction transform operation with wavelength λ = 632.8 nm and diffraction distance Z = 120 mm is performed on the product result, and the result of this operation is multiplied by the second iris chaotic structured light phase mask ICSPM2. Then, a Fresnel diffraction transform operation with λ = 632.8 nm and diffraction distance Z = 150 mm is performed on the product result again to obtain the amplitude distribution A(x,y) and the phase distribution which are respectively the encrypted object light wave L Figure 11 as shown in o (i1) and (i2), namely L o (η,ξ). Finally, an axial plane reference light wave L r (η,ξ) coaxial with the encrypted object light wave L r (η,ξ) and perpendicular to the recording plane is introduced. Subsequently, by changing the phase shift amount of L o (η,ξ) three times r and making L Figure 11 (η,ξ) and L (η,ξ) interfere on the recording plane, three holographic ciphertexts (I1(x,y), I2(x,y), I3(x,y)) as shown in (j1-j3) are generated. Thus, the encryption step is completed.

[0134] Next, according to Figure 7 the decryption steps shown below, the decryption key required decrypts into the user's iris key and digital keys (wavelength key, three diffraction distance keys, focal length key of the Fresnel zone plate, topological charge number key). When all the keys are correct, the decryption results of the red, green, and blue channels of the color medical image ("Brain") are respectively as shown in Figure 12 (a1)-(a3), and the final decryption result is as shown in Figure 12 (b). In addition, in order to evaluate the decryption quality, the correlation coefficient (CC) and average correlation coefficient (Average CC) are used to evaluate the degree of correlation between the decryption results of each channel and the final decryption result and the original color plaintext image, that is:

[0135]

[0136] where P υ (x,y) and respectively represent the gray values of the original plaintext image and the decryption result at (x,y) when the channel υ (υ = R, G, B), and respectively represent the average pixel gray values of the original plaintext image and the decryption result when the channel υ (υ = R, G, B). It can be seen from Figure 12 that when all the keys are correct, the correlation coefficients between the decryption results of the red, green, and blue channels of the color medical image ("Brain") and the original color plaintext image are all greater than 0.995. Therefore, it can be proved that when all the keys are correct, the original color plaintext image can be correctly reconstructed with high quality. Next, the effectiveness and security of the iris key in the present invention are investigated. The applicant randomly collected more than 500 iris images from different attackers and used them as incorrect decryption iris keys. Figure 13 The decryption results are given when the decryption iris key of the user is incorrect. It should be noted that Figure 13 the incorrect iris images shown in are randomly selected, and the decryption results corresponding to the other incorrect decryption iris keys are similar to the decryption results shown in Figure 13 . It can be seen from Figure 13 that even in the extreme case where the illegal attacker has a twin relationship with the encrypted user, the iris correlation degree between its iris key and the iris key of the encrypted user is only 0.463, which is significantly lower than the system preset threshold δ = 0.75. In addition, the correlation degrees between the iris keys of other illegal attackers and the iris key of the encrypted user are also lower than the preset threshold. Therefore, it can be proved that the iris key in the present invention has strong security.

[0137] Next, the robustness of the iris key in the present invention is investigated, including the robustness against acquisition distance deviation, glasses interference, fixation point tilt, and anti-light attack. First, the proportion of the iris in the image is negatively correlated with the acquisition distance, that is, as the acquisition distance increases, the proportion of the iris region in the image decreases accordingly. To evaluate the robustness of the iris key against acquisition distance deviation, the applicant tests the ability of the iris key to resist acquisition distance deviation attacks by using the iris keys of correctly decrypted users collected at different acquisition distances. The test results are as shown in Figure 14 (a). As can be seen from the authentication results in Figure 14 (a), as the proportion of the iris region in the image decreases (i.e., the acquisition distance increases), the correlation degree between the decrypted iris key and the iris key with a proportion of 100% shows a downward trend. However, even when the iris proportion drops to 5%, its correlation degree is still higher than the preset threshold (0.75), thus enabling correct decryption. Therefore, it can be proved that the iris key in the present invention has significant robustness against acquisition distance deviation attacks. Next, the ability of the iris key to resist glasses interference attacks is tested by interfering with the iris keys of correctly decrypted users. The test results are as shown in Figure 14 (b). As can be seen from Figure 14 (b), although the correlation degree between the decrypted user wearing glasses and the iris without glasses decreases, it is still higher than the preset threshold (0.75). Therefore, it can be proved that the iris key in the present invention has significant robustness against glasses interference attacks. Secondly, the ability of the iris key to resist fixation point tilt attacks is tested by using the iris keys of correctly decrypted users collected under different fixation point tilt conditions. The test results are as shown in Figure 14 (c). As can be seen from Figure 14 (c), although the correlation degree between the iris key with a tilted fixation point and the iris key without fixation point offset decreases, it is still higher than the preset threshold (0.75). Therefore, it can be proved that the iris key in the present invention has significant robustness against fixation point tilt attacks. Finally, the applicant tests the ability of the iris key to resist light attacks by changing the light offset value. The test results are as shown in Figure 15 (c), where the decrypted iris after being subjected to the light with a light offset value of ω can be expressed as:

[0138] where E is a matrix of all 1s. As can be seen from Figure 14 (c), even for the decrypted iris image subjected to a light offset value of ω = ±40 and the decrypted iris image without being subjected to light attack, the correlation degree is still greater than the preset threshold (0.75). Therefore, it can be proved that the iris key in the present invention has significant robustness against light attacks.

[0139] Finally, the sensitivity of the digital key in the present invention is investigated, where the average correlation coefficient (Average CC) is used to evaluate the correlation degree between the final decryption result and the original color plaintext image. Figure 15 (a)-(f) respectively show the sensitivity curves of the first diffraction key Z1, the second diffraction key Z2, the third diffraction key Z3, the focal length key f, the topological charge number key p, and the wavelength key λ. In addition, when the deviation of the first diffraction distance key ΔZ1 = 1 mm, the deviation of the second diffraction distance key ΔZ2 = 1 mm, the deviation of the third diffraction distance key ΔZ3 = 1 mm, the deviation of the focal length key Δf = 1 mm, the deviation of the topological charge number key Δp = 1%, and the deviation of the wavelength key Δλ = 1 nm, the decryption results are respectively marked in Figure 15 (a)-(f). On the other hand, Figure 15 (g) shows the multivariate function relationship between the average correlation coefficient between the decryption result and the original plaintext image and the deviation of the first diffraction distance key ΔZ1, the deviation of the second diffraction distance key ΔZ2, the deviation of the third diffraction distance key ΔZ3, the deviation of the focal length key Δf, the deviation of the topological charge number key Δp, and the deviation of the wavelength key Δλ when the iris key is correct. It can be seen from Figure 15 that when there is a slight deviation in any one of the digital keys, the average correlation coefficient between the decryption result and the original color plaintext image drops sharply to nearly 0. At this time, the decryption result shows an obvious noise distribution. Only when the deviations of all digital keys are 0 can the original color plaintext image be correctly restored, which proves that the digital key in the present invention has high sensitivity.

[0140] The above is the description of the specific implementation of the present invention, rather than the 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. Therefore, all equivalent technical solutions should be included in the patent protection scope of the present invention.

Claims

1. A color medical image encryption method based on iris key and digital holography, characterized in that: The encryption steps include: S1: Encryption key generation based on iris high-order data acquisition: First, use the iris acquisition device to collect the original iris image Z(x,y) of the encrypted user; then use iris image processing technology and iris feature extraction technology based on fully convolutional networks to obtain the high-order mask information Z of the two encrypted iris images of the encrypted user F (x,y) and Z M (x, y), and use image hash value extraction technology to obtain Z F (x,y),Z M The decimal hash values ​​SH1 and SH2 of (x, y) are obtained and stored in the encryption database; finally, the chaotic matrix generation program and the structured light phase mask generation algorithm are used to generate the iris chaotic structured light phase masks ICSPM1 and ICSPM2 required in the encryption step S3 for SH1 and SH2 respectively; S2: Plaintext image reconstruction based on FRFCM algorithm and grating modulation technology: First, the foreground area P of the color plaintext image P(x, y) to be encrypted is extracted using the FRFCM algorithm and image processing technology. ROI (x,y); then P ROI (x, y) performs image reshaping operation to obtain the encrypted reshaped image P of size m×m SHA (x,y); then P SHA (x, y) performs image channel separation operation to obtain P SHA The red component data P of (x,y) SR (x, y), green component data P SG (x, y), blue component data P SB (x, y); finally, the grating modulation technology is used to SR (x,y),P SG (x,y),P SB (x, y) performs a fusion operation to obtain an encrypted fusion result G(x, y); S3: Image encryption based on chaotic structured light illumination and digital holographic coding: First, ICSPM1, ICSPM2 and double random phase coding encryption technology are used to perform encryption operations on G(x, y) to obtain the encrypted light wave L o (η,ξ); Then, using the reference light wave L r (η,ξ) and phase-shift digital holographic coding technology to convert L o (η,ξ) is encoded into three real amplitude holograms I containing color plaintext information ν (ν=1,2,3), and output it as the final ciphertext; The encryption step is now complete; Decryption steps: J1: Decryption key acquisition based on iris feature authentication: First, use the iris acquisition device to collect and decrypt the user's original iris image Then, iris image processing technology and iris feature extraction technology based on fully convolutional networks are used to obtain high-order mask information of the two iris images of the decrypted user. and Next, the iris correlation CoI corresponding to the decryption user and the encryption user is calculated. If CoI is greater than the preset threshold δ, the authentication is successful. The system will output SH1 and SH2 stored in the encryption database, and perform the same operation as the encryption step S1 to generate the iris chaotic structured light phase masks ICSPM1 and ICSPM2. Then, the conjugation operation is performed on ICSPM1 and ICSPM2 to generate the iris chaotic structured light conjugate phase mask required for the next decryption step. and Otherwise, if the authentication fails, the decryption is terminated; J2: Image decryption based on digital holographic decoding and chaotic structured light illumination: First, the digital holographic decoding algorithm is used to perform decoding operations on the three holographic ciphertext holograms to obtain the decrypted object light wave Then will and As the first decryption phase key and the second decryption phase key respectively, and using the double random phase decoding algorithm to Perform decryption operation to obtain decryption fusion result J3: Decrypted image reconstruction based on spectral filtering and image reconstruction technology: First, spectral filtering technology is used to Perform image decomposition to obtain the reshaped decrypted image The red component decryption result Green component decryption result Blue component decryption result Then Perform the image channel merging operation to obtain the reshaped decrypted image Last pair Perform image reshaping and image zero padding operations to obtain the final decryption result The decryption step is now completed.

2. The color medical image encryption method based on iris key and digital holography according to claim 1, characterized in that: The specific process of the encryption step S1 is as follows: (i) Encrypted iris image acquisition and preprocessing: First, the original iris image Z(x, y) of the encrypted user is acquired using an iris acquisition device. Then, grayscale, positioning, enhancement, and stretching operations are performed on Z(x, y) in sequence to obtain the encrypted iris normalized image Z N (x,y); (ii) Encrypted iris high-order feature extraction based on fully convolutional network: Z N (x, y) is input into the fully convolutional network composed of the iris feature extraction network (FeatNet) and the iris region recognition network (MaskNet) to obtain the encrypted iris image high-order mask information Z F (x,y) and Z M (x,y): Among them, the structure of FeatNet includes a 3×7 convolution layer (C1), a 3×5 convolution layer (C2), two 3×3 convolution layers (C3, C4), two pooling layers (P1, P2), and three activation functions (T1, T2, T3); the structure of MaskNet includes four 3×3 convolution layers (C1, C2, C3, C4), three 1×1 convolution layers (C2_S, C3_S, C4_S), two 2×2 pooling layers (P1, P2), and a 4×4 pooling layer (P3); (iii) Encrypted iris chaotic structured light phase mask generation: First, calculate Z F (x,y) and Z M The hash value of (x, y) is then converted into decimal data to obtain the conversion results SH1 and SH2, which are stored in the encrypted database. Then SH1 and SH2 are used as the initial values ​​of chaos, respectively. The general term SH of the chaotic sequence is r(n) for: Among them, SH 1(n) =SH n And n = 1, 2; secondly, the chaotic sequence is arranged into a two-dimensional matrix to generate the iris chaotic matrix ICM n (n=1,2); Next, the encryption user uses a simple set of digital keys to construct the Fresnel zone plate FZP and radial Hilbert mask RHM: 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, j is the imaginary unit, arg(·) is the phase operation; finally, ICM n (n=1,2), FZP, RHM perform phase modulation operation to generate iris chaotic structured light phase mask ICSPM n (n=1,2): ICSPM n =exp(j(ICM n ×FZP×RHM))。 3. The color medical image encryption method based on iris key and digital holography according to claim 1, characterized in that: The specific process of the encryption step S2 is as follows: (i) Extraction of foreground information of plaintext image based on FRFCM algorithm: First, grayscale and morphological reconstruction operations are performed on the original color plaintext image P(x, y) to obtain the morphological reconstruction result P Δ (x,y); then P Δ (x, y) executes the FRFCM algorithm with the number of clusters c, the fuzzification parameter s, and the minimum error threshold τ, thus obtaining the segmentation result P of P(x, y) SEG (x,y): P SEG (x,y)=FRFCM c,s,τ (P Δ (x,y)) =FRFCM c,s,τ (Mcr(Gry(P(x,y))) Among them, Mcr(·), Gry(·), FRFCM c,s,τ They are respectively represented as morphological reconstruction operation and grayscale operation, as well as the FRFCM algorithm with the number of clusters c, fuzzification parameter s, and minimum error threshold τ; then P SEG (x, y) performs clustering operation to obtain the color plaintext image P(x, y) mask MASK: Finally, the foreground area P of the plaintext image is extracted by performing pixel-by-pixel multiplication on MASK and P(x,y). ROI (x,y): P ROI (x,y)=MASK×P(x,y) (ii) Plaintext image reconstruction based on grating modulation: First, pixel extraction technology is used to extract P ROI Each pixel x1,x2,...,x in (x,y) N (N=P×Q), where P and Q are P ROI The horizontal and vertical pixel numbers of (x,y); then the image reconstruction technique is used to convert x1,x2,...,x N (N = P × Q) is rearranged into a size of The encrypted reshaped image P SHA (x,y); then, for P SHA (x, y) performs image channel separation operation to obtain P SHA The red component data P of (x,y) SR (x, y), green component data P SG (x, y), blue component data P SB (x,y); Finally, for P SR (x,y),P SG (x,y),P SB (x, y) performs grating modulation operation to obtain the encrypted fusion result G(x, y): Among them, K n (x, y) is the grating coefficient, and its mathematical expression is: Among them, f n is the grating spatial frequency, θ xn is the angle between the grating line direction and the horizontal direction, θ yn is the angle between the grating line direction and the vertical direction, n=1,2,3.

4. The color medical image encryption method based on iris key and digital holography according to claim 1, characterized in that: The specific process of the encryption step S3 is as follows: (i) Double random phase encoding based on chaotic structured light illumination: First, a Fresnel diffraction transform with a wavelength of λ and a diffraction distance of Z1 is performed on the encrypted fusion result G(x, y), and then the transform result is multiplied with the first iris chaotic structured light phase mask ICSPH1. Then, a Fresnel diffraction transform with a wavelength of λ and a diffraction distance of Z2 is performed on the multiplication result, and then the transform result is multiplied with the second iris chaotic structured light phase mask ICSPH2. Finally, a Fresnel diffraction transform with a wavelength of λ and a diffraction distance of Z3 is performed on the multiplication result to obtain the encrypted object light wave L o (η,ξ): Among them, FrT λ,Z (·) represents the Fresnel diffraction transformation with wavelength λ and diffraction distance Z; (ii) Holographic ciphertext generation based on phase-shift digital holographic coding: First, we introduce the encryption light wave L o (η,ξ) is the axial plane reference light wave L that is coaxial and perpendicular to the recording plane r (η,ξ), and then by changing the phase shift of the reference light wave three times And make L o (η,ξ) and L r (η,ξ) interfere on the recording plane, thus generating the holographic ciphertext I ν (x,y)(ν=1,2,3): Among them, |·| represents the wavelength modulus operation, L r (η,ξ) * , L o (η,ξ) * They are represented as reference light wave L r (η,ξ) and the encrypted light wave L o The conjugate of (η,ξ).

5. The color medical image encryption method based on iris key and digital holography according to claim 1, characterized in that: The specific process of the decryption step J1 is as follows: (i) Decrypted iris original feature image acquisition: First, use the iris acquisition device to collect and decrypt the user's original iris image Then Grayscale, positioning, enhancement, and stretching operations are performed in sequence to obtain the decrypted iris normalized image. (ii) Iris high-level information authentication: First, Input into the fully convolutional network composed of the iris feature extraction network (FeatNet) and the iris region recognition network (MaskNet) to obtain the high-order mask information of the decrypted iris image and Then, the high-order mask information of the decrypted iris image and the Z stored in the encrypted database are used to F (x,y) and Z M (x,y) Calculate the iris association COI between the decrypted user and the encrypted user: in Z F (i,j), Z M (i,j) are represented as Z F (x,y), Z M The pixel value of (x, y) at (i, j), m and n represent the horizontal pixel number and vertical pixel number of the high-order mask information of the iris image, respectively, and xor is an exclusive OR operation. Finally, the iris high-order information authentication is performed by judging whether COI is greater than the preset threshold δ. If COI is greater than the preset threshold δ, the authentication is successful, and SH1 and SH2 stored in the encryption database are output as decrypted data and enter the next decryption step; Otherwise, the authentication fails and the decryption is terminated; (iii) Generation of iris chaotic structured light conjugate phase mask: SH1 and SH2 obtained in the previous decryption step are used as the chaotic initial values, and the iris chaotic matrices ICM1 ​​and ICM2 are generated using the same recursive formula as sub-step (iii) in the encryption step S1; then the decryption user needs to enter a set of digital keys to construct the decrypted Fresnel zone plate and decrypted radial Hilbert mask Then, phase modulation technology is used to modulate ICM1, ICM2, as well as Perform phase modulation operation, and perform conjugation operation on the modulation result to obtain the iris chaotic structure light conjugate phase mask where conj(·) is the conjugation operation.

6. The color medical image encryption method based on iris key and digital holography according to claim 1, characterized in that: The specific process of the decryption step J2 is as follows: (i) Decryption light wave generation based on digital holographic decoding: The holographic ciphertext I is decoded using the digital holographic decoding algorithm. ν (x, y)(ν=1,2,3) performs the decoding operation to obtain the decrypted light wave (ii) Dual random phase decoding based on chaotic structured light illumination: First, the decryption object light wave The wavelength is λ and the diffraction distance is The inverse Fresnel diffraction transform is used to transform the result of the transformation and the conjugate phase mask of the second iris chaotic structure light. Multiply, and then perform the multiplication again. The wavelength is λ, and the diffraction distance is The inverse Fresnel diffraction transform is then combined with the first iris chaotic structure light conjugate phase mask Multiply, and finally perform a wavelength of λ on the product result, and the diffraction distance is The inverse Fresnel diffraction transform is used to obtain the decrypted fusion result in Expressed as wavelength λ, the diffraction distance is Inverse Fresnel diffraction.

7. The color medical image encryption method based on iris key and digital holography according to claim 1, characterized in that: The specific process of the decryption step J3 is as follows: (i) Decrypted and reshaped image generation based on spectral filtering: Decrypted fusion results are fused by constructing a triple filter Perform spectral filtering to obtain the reconstructed decrypted image The red component decryption result Green component decryption result Blue component decryption result Among them, FT(·) and IFT(·) represent the Fourier transform operation and the inverse Fourier transform operation respectively. χ is a triple filter. Perform the image channel merging operation to obtain the reshaped decrypted image (ii) Decryption result generation based on image reconstruction technology: First, extract Every pixel in Then, using image reconstruction technology, Reshape into a decrypted foreground region of size P×Q Last pair Perform zero padding to generate the final decrypted result

Images