Biometric-Based Optical Systems for Security and Authentication

Abstract

In a digital world, biometric authentication is becoming more and more popular for reliable automatic recognition of people, which is widely being deployed in optical information security-related systems. The adoption of biometrics into optical security-based applications and fields has been adding excellent security due to their distinctive attribute that gains from optics. In this chapter, we present an optical nonlinear cryptosystem for image encryption using biometric keys generated from fingerprint hologram for security and authentication. In order to generate biometric keys, we implemented an optoelectronics experiment setup using digital holography for capturing the fingerprint hologram, storing, and then numerically reconstructing it. The reconstructed features of the fingerprint object offer very appealing attributes from the perspective of data encryption such as uniqueness, randomness, and discriminability. Fingerprint biometric features are kept inside interference patterns optically, which are also protected with experimental parameters. If both pieces of information are provided to be known to the person at the decryption stage, as a result, it keeps maintaining user specificity in order to access system information. Furthermore, we exploit the utility of the biometric key in designing an optical cryptosystem for encrypting the information which offers a solution to the distribution of keys with heightened security.

1. Introduction

Biometrics refers to a unique, measurable, biological trait or attribute of a human being that is used to validate the identity of a person [1, 2]. The use of biometrics in a system relies on automated recognition methods based on a person’s physiological or behavioral features, whose functionality works on the conforming exact identity of an individual compared to traditional authentication methods such as passwords, tokens, and PINs (Personal Identification Numbers) [3, 4, 5]. A number of applications are implemented to automate authentication methods by the use of biometric traits for access control, commercial, phone, government, and forensic [1, 2, 3, 4, 5, 6]. In general, the biometric trait comprises physiological or behavioral features [5]. The physiological traits, which use fingerprints, retina, iris, facial images, and hand geometry, are physical characteristics computed at a specific point in time, while behavioral biometric traits commonly list in particular, signature, gait, voice recordings, and keystroke rhythms, make attention to the mode some action is accomplished by every individual [1, 4].

Due to recent technological advances in real-world applications, an automated identification system has been implemented in many security-related systems using biometric traits for reliable and trusted authentication [1, 2, 3, 4, 5, 6, 7]. Moreover, different kinds of threats, challenges, and privacy issues are growing concerns in today’s modern world, and biometric technology is used to ensure secure and safe circumstances [4, 5, 6, 7]. The development of optics-based biometric systems has brought tremendous growth in data security and authentication in recent times [6, 7]. More excitingly, most of the current optical encryption systems are now being incorporated with the biometric features of a person [7]. These scenarios, however, increase the security level for information protection through person identification against unauthorized access [4].

Optical systems are extensively involved in the field of information security due to high speed, parallel processing, and exploitation of multidimensional data such as wavelength, frequency, and polarization [7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66]. Optics-related cryptosystems, such as optical, compression, encryption, photon counting, and authentication, are extensively developed to secure sensitive data or images during transmission and reception through the digital medium [6, 7]. In the field of optics, image encryption was brought into existence since the introduction of the double random phase encoding (DRPE) scheme. The DRPE method converts input information into white stationary noise by the involvement of two random phase masks (RPMs) at the input and the Fourier plane, respectively [8]. Due to the linear and symmetric nature of the DRPE system, these RPMs for encryption and decryption processes are similar to security keys [8, 9, 10, 11, 12, 13]. In order to take advantage of optical encryption, Qin et al. reported the optical cryptosystem for image encryption based on phase-truncated Fourier transforms (PTFTs) [14, 15]. The PTFT operation is used to truncate the Fourier spectrum of the image into the phase and the amplitude distributions. The PTFT encryption scheme uses two RPMs for encryption, while two phase-only masks are obtained as the decryption keys for the decryption process. The main advantages of PTFT over the DRPE are that keys for encryption and decryption processes are distributed due to nonlinear operation. These RPMs act as the main security component. From the cryptanalysis point of view, it is noticed that the RPM-based encryption system suffers from the problem of key management, distribution, and authentication and is found to be insecure against various types of attacks [12, 13, 16, 17, 18, 19]. Furthermore, by adding security to optical encryption using unique features of human beings, optical security using biometrics facilitates a secure and reliable way for information processing. Many types of optical systems using biometrics are implemented [24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40], which offer a wide range of features such as keys management, higher security, and user authentication.

From the recent research in the optical encryption domain, it is observed that optical cryptosystems are perceived to be unsuited to implement traditional cryptography because of being unable to address the key distribution issue in terms of public keys and private keys [33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53]. A nonlinear encryption system in the Fresnel domain using the optical phase-retrieval algorithm is developed as reported in [39], which fulfills the criteria of asymmetric cryptography agreement. The phase-retrieval algorithm-based method has also been studied on the development of nonlinear cryptosystems that show progress in the generation of public and private keys in the encryption system, which is demanded with authentication. Zhao et al. presented an optical nonlinear cryptosystem using a fingerprint combined with a phase-retrieval algorithm and public key cryptography [40]. In this scheme, the fingerprint features of a person are associated with encryption and decryption operations that help to decrypt the information in the authenticated way at the receiver and also solve the issue of the public-private keys [40, 41, 42]. One of the approaches to the implementation of optical information authentication systems is carried out by combining a median-filtering-based phase-retrieval algorithm [51]. Moreover, optical image encryption techniques using digital holography are applied for authentication and security [32, 33, 34, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65]. This method includes an optical process for recording a hologram by the involvement of a charge-coupled device (CCD), and the captured hologram is known as a digital hologram, which is further numerically reconstructed in a computer [32, 33, 34, 57]. The reconstructed information provides additional information in terms of the amplitude, and the phase of the object. Thus, the reconstructed features are explored in the design of optical information processing systems for security and authentication.

In this chapter, we describe our proposed optical nonlinear cryptosystem for image encryption using biometric keys based on an optical phase-retrieval algorithm and phase-truncated Fourier transform for security and authentication, which offer the solution of key distribution with improved security. In this direction, we implement the optoelectronics experiment based on digital holography for recording the fingerprint hologram, which is digitally reconstructed to obtain keys information in terms of the amplitude and the phase. The merit is that the digital recording and reconstruction process of the fingerprint hologram using holography makes it possible for transmission and reception over a communication medium. In addition, the fingerprint hologram is protected by a reconstruction parameter which also enables verification and authentication approach in the proposed encryption/decryption processes. First, we introduce the idea of the optoelectronic experimental process for recording the fingerprint hologram, and its numerical reconstruction. Next, we analyze the features of the reconstructed fingerprint image by performing the statistical test that makes its usage as an encryption key for the image [32]. Furthermore, we explore the utility of the biometric key for optical cryptosystem for image encryption based on the phase-retrieval algorithm and the phase-truncated Fourier transforms (PTFT) scheme. The system uses keys for encrypting the information using the public keys or encryption keys that can only be truly recovered using the private keys or decryption keys, while the involvement of biometric keys maintains the authenticity of the user throughout the process. Our work is the first attempt to develop an optical cryptosystem using the phase-retrieval algorithm and PTFT combined image encryption system utilizing biometric keys from fingerprint hologram. Finally, we demonstrate the security performance and robustness of our cryptosystem. This chapter is structured as follows: Section 2 introduces the optoelectronics setup using digital holography for biometric keys generation and analyzes the keys features demonstrating its utility for image encryption. Section 3 introduces the cryptography perspective using biometrics. Section 4 presents an implementation of optical encryption process. Section 5 investigates experimental results to present enhanced security with user authentication and management of keys. Finally, the conclusion presents the significant contributions, discusses the challenges of the work, and suggests future research directions.

2. Optoelectronic experimental setup for biometric keys

In this section, the method of fingerprint hologram recording has been presented using digital holography in order to capture both phase and amplitude distributions [54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65]. In order to perform fingerprint imaging, an optoelectronics experiment using a digital holographic technique is implemented, as shown in Figure 1.

Figure 1.

Optoelectronic experimental setup: BSs: beam splitters, Ms: mirrors, SF: spatial filtering.

2.1 Fingerprint database

From the biometric perspective, fingerprints contain essential patterns like arches, ridges, and whorls on the surface of a finger which are unique to the person. In our study, we use samples of the fingerprint from the fingerprint verification competition (FVC) database, as reported in [66]. The ‘FVC’ term signifies a fingerprint verification competition. This database has eight sets of fingerprint impressions of 100 users which are captured and collected using different sensor-based technologies. Features of the fingerprint images from the dataset have been extracted with pixels 300 × 480 and a resolution of 512 dots per inch (dpi). From an imaging perspective, the fingerprint is employed in a digital holographic-based setup as shown in Figure 1 and its detailed process is explained in Section 2.2.

2.2 Recording process of the biometric hologram

In order to capture fingerprint image hologram using the experimental setup as shown in Figure 1, the optical beam emerging from the He-Ne laser is initially collimated through spatial filtering (SF) operation and then separated into the object arm and the reference arm with the use of a beam splitter (BS₁), which is directed with the help of mirrors (Ms) and finally combined at the BS₂. The light that passes from the fingerprint object is known as the object beam, while another beam O(x,y) that comes from the reference arm is denoted as the reference beam U(x,y). Several techniques for the acquisition of fingerprints are widely investigated by researchers in security and optical imaging-related applications. In our optical configuration, we use fingerprint images of a person from a public biometric dataset as reported in [66]. In order to perform transmission imaging, fingerprint features are displayed on the transparent sheet of size ‘1 cm × 1 cm’ which shows high contrast features for recording as per the detailed procedure reported in our previous work [32, 33, 57]. In the object arm, the fingerprint images of size ‘1 cm × 1 cm’ are employed at a distance‘d’ from the charge-coupled device (CCD) device [32, 33, 57]. As a result, this optical configuration makes it practically more suitable for optical imaging as well as information processing using digital holographic techniques in comparison to conventional fingerprint acquisition methods as described in [57]. In the process of recording, when light rays from a He-Ne laser strike the fingerprint object it causes diffraction, scattering, and absorption. This phenomenon carries signifying information about the object’s amplitude and shape, which further interfered with the reference beam that is recorded by the CCD camera. The recorded interference pattern contains the complete information on the fingerprint object shown in Figure 2, which is stored in the computer and coined as a digital hologram. This process can be mathematically expressed as:

Figure 2.

Recorded fingerprint hologram.

$$\begin{array}{l}\mathrm{H}\left(\mathrm{x},\mathrm{y}\right)={\left|\mathrm{O}\left(\mathrm{x},\mathrm{y}\right)+\mathrm{U}\left(\mathrm{x},\mathrm{y}\right)\right|}^2\\ =\mathrm{O}\left(\mathrm{x},\mathrm{y}\right){\mathrm{O}}^{\ast }\left(\mathrm{x},\mathrm{y}\right)+\mathrm{U}\left(\mathrm{x},\mathrm{y}\right){\mathrm{U}}^{\ast }\left(\mathrm{x},\mathrm{y}\right)\\ +\mathrm{O}\left(\mathrm{x},\mathrm{y}\right){\mathrm{U}}^{\ast }\left(\mathrm{x},\mathrm{y}\right)+\mathrm{U}\left(\mathrm{x},\mathrm{y}\right){\mathrm{O}}^{\ast }\left(\mathrm{x},\mathrm{y}\right)\end{array}$$E1

As given in Eq. (1), the ‘$\mathrm{H}\left(\mathrm{x},\mathrm{y}\right)$’ is termed as the digital hologram and ‘_*’ shows the complex conjugate operation. The recording parameters for the fingerprint hologram are given as wavelength λ = 632.8 nm, the distance d = 0.29 m, and pixel sizes 4.65 μm × 4.65 μm of the CCD (Lumenera’s Infinity2, 1360 × 1024 pixels).

2.3 Reconstruction process of the fingerprint hologram

In order to reconstruct the fingerprint features from the recorded hologram as shown in Figure 2, the reconstruction processes using the Fresnel-Kirchhoff integral are employed numerically, which makes it free from the zero order term in separating the real and the virtual images [32, 33, 54, 55, 56, 57] as:

$$D\left(v,u\right)=\frac{i}{\lambda }\underset{-\infty }{\overset{\infty }{\iint }}H\left(x,y\right)E_R\left(x,y\right)\frac{exp\left(-i\frac{2\pi }{\lambda }{\mathrm{\rho }}^{\prime }\right)}{{\mathrm{\rho }}^{\prime }}\mathit{dx}\mathit{dy}$$E2

$${\rho }^{\prime }=\sqrt{{\left(x-v^{\prime }\right)}^2+{\left(y-u^{\prime }\right)}^2+d^2}$$E3

where D(v, u) shows the reconstructed object, E_R refers to the plane wave, and ρ’ represents the distance between the recording plane (x, y) and the reconstruction plane (v’, u’), respectively. It can be noticed that the process of reconstruction is completely carried out digitally. From the reconstructed object as demonstrated in Figure 3, the real fingerprint image of size 146 × 146 pixels is extracted in our analysis. The reconstructed fingerprint object results in a complex field which is further separated in terms of the intensity and phase distribution:

Figure 3.

Reconstructed features: (a) amplitude distribution and (b) phase distribution.

$$\mathit{D}\left(\mathit{v},\mathit{u}\right)=\mathit{A}\left(\mathit{v},\mathit{u}\right).\mathit{exp}\left(\mathit{i\varphi }\left(\mathit{v},\mathit{u}\right)\right)$$E4

$$\mathit{A}\left(\mathit{v},\mathit{u}\right)=Re\left[\mathit{D}{\left(\mathit{v},\mathit{u}\right)}^2\right]+Im\left[\mathit{D}{\left(\mathit{v},\mathit{u}\right)}^2\right]$$E5

$$\varnothing \left(\mathit{v},\mathit{u}\right)=\mathit{arctan}\frac{Im\left[\mathit{D}\left(\mathit{v},\mathit{u}\right)\right]}{Re\left[\mathit{D}\left(\mathit{v},\mathit{u}\right)\right]}$$E6

The reconstructed object clearly reproduces the fingerprint pattern as shown in Figure 3a while the phase obtained from the same hologram is represented in Figure 3b which shows significant information on the fingerprint pattern as well as some variation of thickness over the field of view. Therefore, both pieces of information clearly exhibit the significant features of the fingerprint biometric.

Furthermore, it is evident that the obtained phase is utilized to construct a phase mask (PM) with phases uniformly distributed in the region [0, 2π] as:

$$\mathit{\text{Phase Mask}}=exp\left(\mathrm{i}2\mathrm{\pi }\varnothing \left(v,u\right)\right)$$E7

The generated phase mask is full of speckle patterns and randomness, as shown in Figure 4. This is also unique to the person because of the associated fingerprint biometrics.

Figure 4.

Generated fingerprint phase mask (PM).

Our previous works have shown that the biometric keys from fingerprint hologram are a promising candidate for image encryption. A detailed description of the biometric key characteristics, such as uniqueness, randomness, and robustness, is recently reported in our published work [32, 33, 34, 57]. Motivated by the utility of biometric keys, the authors presented a cryptosystem for image encryption and decryption.

3. Basic cryptography in the perspective of biometric authentication

In the domain of data security, the public key cryptographic technique is used to secure the information using the public keys (K_Public) and the private keys (K_Private) [53]. The cryptographic process between Alice and Bob is shown in Figure 5, which can be explained as:

1. In the cryptographic process, the public key is utilized as the key for encryption while the private key is used only for decryption.

2. To transfer the ciphertext to Alice, Bob applies the use of a public key for encrypting the input data.

3. At the decryption stage, Alice involves the use of private keys to get back the input information from the ciphertext using the decryption process. Moreover, Alice cannot decrypt the unspecified ciphertext because of the nonavailability of their private keys.

Figure 5.

Basic public key cryptography for encryption: P = plaintext, E = encrypted or ciphertext, and f = processing algorithm.

The purpose of the cryptographic process is to convert plaintext information into ciphertext. Moreover, its security strength depends on keys and processing algorithms which are not linked with the user’s identity [53]. In general, security systems use token, ID, and password to authenticate the person but still suffer from several issues and limitations with regard to information security as well as insufficient database to prevent unauthorized access [5]. In order to implement a biometric-based authentication approach, the biometrics of a person must be first registered into the system that works only for an authorized person and it would not work if a person is not registered. For this purpose, a fingerprint hologram of a person using a digital holographic process is utilized, as shown in Figure 6. The digital process for recording, as well as retrieval of the keys, makes it safe, secure, and accessible for encryption and decryption processes. In addition, the use of experimental parameters results in additional security for the system [32, 33, 34, 57].

Figure 6.

Cryptography perspective using biometrics.

In view of cryptography using a biometric perspective, Alice uses her biometric keys retrieved from fingerprint hologram for encoding the plaintext information, and Bob confirms Alice’s biometric keys by matching them to the registered database so that it can assure the ciphertext coded by Alice [33]. In this way, this strategy satisfies the criteria of asymmetric cryptography with authentication. The rules of the system can be elaborated as:

1. If Bob wishes to transmit the input information to Alice. In this case, Bob first needs to register his fingerprint biometric details using the process of digital holography.

2. In the next step, Bob employs the encryption algorithm for converting the information into the ciphertext using the public key by including biometric keys.

3. In order to decode, Alice needs the simultaneous presence of the private keys and the availability of Bob’s fingerprint hologram to obtain biometric keys.

Hence, the main contribution of this chapter is to present an optical nonlinear cryptosystem by involving biometric authentication using a fingerprint hologram. The authors evaluated measures of the optical cryptosystem process in achieving keys in terms of public and private keys for encryption and decryption with higher levels of security.

4. Optical cryptosystem using biometric keys

This section presents algorithms for encoding of the image by the use of biometric keys.

4.1 Encryption process

To encrypt the input information using the optical cryptosystem, a preprocessing layer of the phase-retrieval algorithm using biometric keys generated from fingerprint hologram has been included, and then the PTFT scheme is employed. The flow diagram of the proposed encryption system is shown in Figure 7.

Figure 7.

Flow diagram of the proposed encryption scheme.

In order to do this, the input image (I) is initially encoded by involving the biometric keys from the fingerprint hologram. This encoding applies constraints as fingerprint amplitude by replacing the amplitude in the Fourier domain, while the Fourier phase is kept unchanged [31]. This process implements iteratively back and forth between the object domain and the Fourier domain. The iteration number is decided between the input image (I) and the retrieved image ($I^{\prime }$) by measuring the correlation coefficient (CC) as:

$$\mathit{CC}=\frac{{\sum }_{x=1}^M{\sum }_{y=1}^N\left(I\left(x,y\right)-\overline{I}\right)\left(I^{\prime }\left(x,y\right)-\overline{I^{\prime }}\right)}{\sqrt{{\sum }_{x=1}^M{\sum }_{y=1}^N{\left(I\left(x,y\right)-\overline{I}\right)}^2}\sqrt{{\sum }_{x=1}^M{\sum }_{y=1}^N{\left(I^{\prime }\left(x,y\right)-\overline{I^{\prime }}\right)}^2}}$$E8

As shown in Eq. (8), the average values of the input image ($I\left(x,y\right)$) and the retrieved image ($I^{\prime }\left(x,y\right)$) are represented as $\overline{I}$ and $\overline{I^{\prime }}$, respectively. The domain $\left(x,y\right)$ represents information in the image plane, while M and N show the row and column of the image. The CC values are plotted with the number of iterations that illustrate the error between the decrypted image and the input image continuously keeps going down as the number of iterations increases. This process facilitates the retrieval of better-quality images, as shown in Figure 8.

Figure 8

(a) Graph between the number of iterations and correlation coefficient (CC) and (b) retrieved image

From the retrieved image as shown in Figure 8, when the CC values reach the desired level (≥ 0.998) then the iteration process is stopped and its output (ψ_k) is combined with fingerprint phase (${\varphi }_{\mathit{\text{fingerprint}}})$ (as)

$${\psi }_k\otimes {\varphi }_{\mathit{\text{fingerprint}}}=ϴ$$E9

Here, $\otimes$ shows the multiplication operator. This resultant information (ϴ) is further distributed into two parts:

(a) Binary key (B): For binary key (B), the resultant information $\left(ϴ\right)$ is coded using the following mathematical identity:

$$B=\left\{\begin{array}{c}0,ϴ<0\\ 1,ϴ\ge 0\end{array}\right.$$E10

If $ϴ$ is greater than or equal to zero then it is set to be 1 while if $ϴ$ is less than zero, it is denoted by zero. This result is coined as a binary key and kept as the private key, which is protected using the pixel scrambling operation to make it safe for transmission [33].

(b) C = abs(ϴ): The absolute information is just an intensity distribution that looks like a speckle, in which the fingerprint biometric features are deeply hidden. Moreover, to obtain the ciphertext (E), the information ‘C’ is processed using the two encryption keys ($\mathit{RPM}1$ and $\mathit{RPM}2$) at the input plane and the Fourier plane using the PTFT operation, as shown in Figure 9. This process is expressed as:

Figure 9.

Phase-truncated Fourier transform (PTFT) scheme for encryption.

$$E1=\mathit{PT}\left\{\mathit{\text{FT}}\left[C·\mathit{RPM}1\right]\right\}$$E11

$$E=\mathit{PT}\left\{\mathit{IFT}\left[E1·\mathit{RPM}2\right]\right\}$$E12

From the process, the two decryption keys ($D1$ and $D2$) are obtained as:

$$D2=\mathit{PR}\left\{\mathit{\text{FT}}\left[C·\mathit{RPM}1\right]\right\}$$E13

$$D1=\mathit{PR}\left\{\mathit{IFT}\left[E1·\mathit{RPM}2\right]\right\}$$E14

From the reported process, the three keys are obtained during encryption processes, which are kept as the private keys to decode the information. As a result, our system involves the use of public keys to encode the input data that can only be decoded using the private keys while the fingerprint keys corroborate the user specificity throughout the optical encryption and decryption processes.

4.2 Decryption process

In order to retrieve the information, the fingerprint hologram and the reconstruction parameters are provided to be known to the user, which further enables the process to recover the original information. This is only possible if both the provided information is correct. Therefore, this processing step has the capability to confirm the authenticity of the person. First, the decryption process is performed to retrieve the information (C) from the ciphertext (E), as shown in Figure 10.

Figure 10.

Phase-truncated Fourier transform (PTFT) scheme for decryption.

$$E1=\mathit{PT}\left\{\mathit{\text{FT}}\left[E·D1\right]\right\}$$E15

$$C=\mathit{PT}\left\{\mathit{IFT}\left[E1·D2\right]\right\}$$E16

For decryption, the initially encoded information (ϴ) is first obtained from the ciphertext (C) by applying the binary key and the scrambling key. As shown in Figure 11, the biometric keys as the phase $\left({\Phi }_{\mathit{\text{Fingerprint}}}\right)$ data and the magnitude ($P)$ are involved to obtain the input image ($I$). These decryption steps are mathematically given by

Figure 11.

Flow diagram of the decryption process.

$$ϴ=C\otimes B=\mathit{abs}\left(ϴ\right)\otimes B$$E17

$${\psi }_k=ϴ/{\Phi }_{\mathit{\text{Fingerprint}}}$$E18

$$I=\mathit{IFT}\left(\left|P\right|exp\left(i{\psi }_k\right)\right)$$E19

where, the term ‘abs’ represents the absolute value of a matrix.

Figure 12 shows the implementation of the optical experimental setup for decrypting the information using electronic devices such as spatial light modulators (SLMs) and CCD, which are controlled by a personal computer. At the beginning of the process, the combined data of the encrypted information (E) with the private key (D1) are shown in the SLM₁ device and then illuminated by the He-Ne laser beam to carry out optical Fourier transform (FT). This result combined with the second private key (D2) is shown in the SLM₂ device, which is further optical FT. In the last step, the obtained information is involved digitally using the binary key (B) and biometric keys. By performing optical FT, the CCD captures the decrypted information.

Figure 12.

Optical setup for decryption: CCD: charge-coupled device, SLM: spatial light modulator, d: focal length of lenses, and SF: spatial filtering.

5. Results and discussion

This section evaluates the performance of our system by performing a number of computer simulations on a MATLAB platform. The obtained results of the system validate the effectiveness of our scheme that exhibits higher security with keys management and distribution.

5.1 Input data

In this section, the computer simulation results of our system are presented. In our experiments, the size of all images employed is 146 × 146 pixels. The input image to be encrypted and the generated biometric keys from the fingerprint hologram are shown in Figure 13a–c. The RPM keys for the PTFT scheme are shown in Figure 13d–e. Our system obtains ciphertext, as shown in Figure 13f.

Figure 13.

(a) Input image. (b) Fingerprint AM key. (c) Fingerprint phase key. (d) RPM1 key. (e) RPM2 key. (f) Ciphertext (E).

5.2 Decryption results

To evaluate our system, a series of decryption experiments are carried out. In the first experiments, we wished to recover the input information against unauthorized attempts using the possible key combinations in Figure 14a–e. Simulation results indicate that Figure 14a shows the truly decrypted input information using all keys in the correct order with authentication. Figure 14b shows the recovered image using the private keys (D1 and D2) in the wrong positions. Figure 14c represents the recovered information when no keys are employed for decryption. Figure 14d displays the obtained image by applying any arbitrarily generated binary key. Figure 14e represents the decrypted noisy data when the biometric phase key is wrongly employed. In addition, to investigate the quality of the recovered data, the mean-square error (MSE) measures for Figure 14a–e are evaluated as 1.5740 × 10⁻⁴, 0.2132, 0.1313, 0.1163, and 0.2609, respectively. Moreover, the CC parameter between the input image as shown in Figure 13a, and the retrieved image as displayed in Figure 14a is calculated as a value of 0.998.

Figure 14.

Decryption results using (a) all correct keys, (b) keys ($D1$ and $D2$) in wrong positions, (c) no keys, (d) wrong binary key, and (e) different fingerprint phase keys.

5.3 Effect of quantization

In the next step of experiments to make digital simulations closer to the true physical process, we evaluated the influence of the quantization at different quantization levels in the process [33]. For this purpose, the encrypted data shown in the SLM device and the retrieved information were captured by a CCD camera that is quantized at different levels of quantization such as 5 bits, 8 bits, 10 bits, 12 bits, and 16 bits, respectively. In this context, the obtained images are displayed in Figure 15. To examine the effect of quantization, the CC and MSE values are calculated. As shown in Table 1, the results illustrate the high accuracy of our system.

Figure 15.

Decrypted image for different quantization levels: (a) 5 bits, (b) 8 bits, (c) 10 bits, (d) 12 bits, and (e) 16 bits.

Quantization level	MSE value	CC value
5 bits	0.2276	0.3525
8bits	0.0654	0.6527
10 bits	0.0240	0.8812
12 bits	0.0045	0.9793
16 bits	2.6235 × 10^−4	0.9968

Table 1.

Showing performance at different quantization levels.

5.4 Robustness of the binary key

In this section, we evaluated the system performance against the binary key. As explained in Section 3.2, the binary key is produced as one of the private keys that are secured by pixel scrambling operation. Using this approach, if the attacker knows the total number of pixels present in the key, this is not sufficient to know the exact distribution of zeros or ones. Keeping in view this point, an attempt is performed using the true binary values of the key and its wrong distribution. We have illustrated results as shown in Figure 16a and b by plotting MSE and CC values between the original image and decrypted images. Experimental results indicate that the true input image is recovered if the binary key is correct while other retrieved images using the wrong binary key represent noisy information.

Figure 16.

Graph to illustrate robustness of the binary key (a) mean-square error (MSE) curve and (b) correlation coefficient (CC) curve.

5.5 Security analysis and discussion

In this section, we evaluated the system performance against iterative phase-retrieval algorithms as reported by researchers [16, 19]. In our work, the fingerprint object information about the amplitude and the phase keys is exploited for encrypting the input information. In this context, the inclusions of the biometric keys make the encryption process relevant to be user-specific. This implication breaks the linearity and enhances the nonlinearity and complexity of the encryption process [32, 33]. In order to access the image using the attack algorithm, the inherent noise goes on boosting at each iterative step. Thus, our system provides resistance against the attacks. In order to prove this point, cryptanalysis was conducted against the attacks such as the special attack and known-plaintext attack (KPA). To illustrate the attack process the special attack is used to break the cryptosystem with the two encryption keys RPMs and the ciphertext, which are allowed to be known to the attacker. This attack is based on a two-step iterative algorithm as mentioned in [16, 19]. Using known resources as shown in Figure 13, the attacker retrieved the information as shown in Figure 17a. Moreover, in a similar manner, our system performed cryptanalysis for the KPA whose results are shown in Figure 17b. Thus, our systems indicate the robustness of the proposed scheme against both the special attack and the KPA.

Figure 17.

Attack results: (a) special attack. (b) Known-plaintext attack (KPA).

5.6 Comparison with other schemes

Finally, we present a comprehensive comparative performance of our system with the recently published schemes [20, 26, 27, 30, 31, 40, 43, 44, 45, 46]. Alarifi et al. [20] introduced an optical PTFT-based asymmetric encryption algorithm for biometric template protection using a cancellable approach. Takeda et al. [26] reported a smart card holder authentication based on the DRPE scheme, in which the encryption key as Fourier phase data of the fingerprint is involved, which gives rise to issues of wrong authentication due to positional variation of fingerprint at the enrollment and verification stages. Saini et al. [27] worked on optical security using a DRPE system that uses encryption keys linked to the biometrics of a user which offers a solution to keys distribution. Tashima et al. [30] presented an improved DRPE security by avoiding the known-plaintext attack. Mehra et al. [43] reported recently an asymmetric system for encrypting the fingerprint image based on quick response (QR) decomposition in the domain of gyrator wavelet transform. Castro et al. [45] proposed an encryption scheme for medical images based on fingerprint authentication. Souza et al. [44] reported an optical encryption technique in which passwords and tokens were included as multifactor for authentication. Chang et al. [46] developed an asymmetric encryption scheme using optical scanning cryptography by combining elliptic curve cryptography, which also helps to achieve keys management. In comparison with the reported encryption algorithms for information using biometrics as reported in [20, 26, 27, 30, 31, 40, 43, 44, 45, 46], our scheme uses the biometric keys obtained from the fingerprint hologram, which is protected by experimental parameters, which help to enable verification and authentication at decryption stage. The digital approach of biometric key generation is safe and secure for encryption and decryption processes. Based on the results obtained from our system, it can resist several types of potential attacks because of its complexity, nonlinearity, and robustness in comparison to other reported cryptosystems. The obtained outcomes in Table 2 demonstrated that the security performance measures of our system are superior and reliable as compared to those mentioned in the previously published work. Our system has a simple experimental implementation that has included the involvement of optoelectronics components and devices. Thus, our system is efficient in terms of security including the distribution of keys with information authentication.

Authors	Methods	Security performance measures
Alarifi et al. [20]	Optical PTFT Asymmetric Cryptosystem-Based Secure and Efficient Cancelable Biometric Recognition System.	Probability of false distribution, equal error rate, false rejection ratio, false acceptance ratio, correlation analysis
Takeda et al. [26]	Encoding plaintext by Fourier transform hologram in double random phase encoding using fingerprint keys	Equal error rate, false rejection ratio, false acceptance ratio
Saini et al. [27]	Biometrics-based key management of double random phase encoding scheme using error control codes	Equal error rate; attack-resistant,; key management
Tashima et al. [30]	Known-plaintext attack on double random phase encoding using fingerprint as key and a method for avoiding the attack	Sum-squared error,; phase-only correlation,; known-plaintext attack
Zhu et al. [31]	Computational ghost imaging encryption based on fingerprint phase mask	Mean-square error, correlation coefficient,; robust against iterative algorithm attacks
Zhao et al. [40]	Image encryption using fingerprint as key based on phase-retrieval algorithm and public key cryptography	Mean-square error, correlation coefficient,; robust against iterative algorithm attacks, fingerprint authentication
Mehra et al. [43]	Fingerprint image encryption using phase-retrieval algorithm in gyrator wavelet transform domain using QR decomposition	Brute force attack, known-plaintext attack, and special attack
Castro et al. [44]	A Medical Image Encryption Scheme for Secure Fingerprint-Based Authenticated Transmission	Mean-square error, peak signal-to-noise ratio, running efficiency, authentication
Souza et al. [45]	Improving biometrics authentication with a multifactor approach based on optical interference and chaotic maps	mean-square error, correlation coefficient, entropy, authentication
Chang et al. [46]	Asymmetric cryptosystem based on optical scanning cryptography and elliptic curve algorithm	Key management, authentication, robust against attacks
Our work [32, 33, 57]	Biometric-based optical systems for security and authentication	Equal error rate, false rejection ratio, false acceptance ratio, mean-square error, correlation coefficient; Known-plaintext attack, special attack, key management, authentication

Table 2.

Comparative study of our work with recent optical cryptosystems.

6. Conclusions and future work

In this chapter, we have described the optical nonlinear cryptosystem using fingerprint biometric keys based on phase retrieval and the PTFT scheme for image encryption. It also showed how fingerprint biometrics can be captured based on optical implementation using digital holography in a practical manner without inconvenience to the person. Our system has the salient features that the capability of biometric key retrieval from fingerprint hologram is led to authenticate the person who possesses the ciphertext for decrypting the information. In addition, our system meets the criteria of asymmetric encryption approach which help to provide a solution for the key management and distribution in the encryption and decryption processes. This system has simple implementation either numerically or optically. As a result, we could mention clearly that our system has validity and robustness against unauthorized attempts and well-known attacks.

In future work, a study about the use of the optoelectronic system to record a hologram of the real fingerprint pattern of a person will be performed in order to explore the scientific potential of the emerging field and make clear the validity of our cryptosystem by evaluating security performance.

Acknowledgments

This work was supported by the National Natural Science Foundation of China (61875129 and 62061136005).

Conflict of interest

The authors declare no conflict of interest.