Abstract
A new asymmetric multiple information security system based on a wavelet transform and gyrator transform is proposed. In the proposed method, a set of four color images are allocated to each user. Each color image is a single-level 2-D discrete wavelet transformed to decompose into LL, HL, LH, and HH sub-bands. The LL sub-bands of four images are fused to obtain a single fused image as an input image, which is segregated into R, G, and B channels. Each channel is compressed by compressive sensing with measurement matrices and modulated by a measurement-matrices-based random phase mask. In similar fashion, n sets of modulated R, G, and B channels are individually multiplexed and then gyrator transformed. The phase of each encrypted channel is embedded into the corresponding channel of the host image to obtain a watermarked channel and its amplitude is used as a common decryption key. Each set has individual decryption keys and chaotic parameters as extremely sensitive decryption keys to ensure the nonlinearity of the system. Thus, it resists potential attacks. The proposed scheme significantly reduces the data volume to be processed, transmitted, and stored, and simplifies the keys to be distributed simultaneously. The retrieved images are devoid of cross-talk noise effects. A simple optoelectronic system can be employed to realize the proposed scheme. Numerical simulation results prove the feasibility of the strategy.
© 2018 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
With more and more threats to information communication systems, optical information security has attracted widespread interest due to their inherent properties of high-speed and multidimensional signal processing capabilities [1,2]. Since Réfrégier and Javidi proposed double random phase encoding (DRPE) [3], various methods for optical image encryption based on DRPE using different generalizations of the Fourier transform have been developed [4–13]. However, because of the inherent linearity property, DRPE-based encryption schemes are vulnerable to different types of attacks [14]. In order to remove the linearity, an asymmetric cryptosystem based on phase-truncated Fourier transforms has been proposed [15]. But it has been found susceptible to the specific attack based on an iterative amplitude-phase retrieval algorithm [16]. Several asymmetric encryption methods have been proposed to resist specific attack [17–19].
In recent years, multiple-image encryption (MIE) techniques particularly draw more and more attention due to its high application value in multiuser authentication, content distribution, enhancing the encryption capacity and the efficiency of secret information transmission. Many variations of MIE have been developed mainly to suppress the cross-talk noise effect and improve the data security [20–31].
Watermarking technique is an effective data security technique that protects the data by means of a watermark embedded within it for copyright protection purpose. Generally, the encryption and watermarking techniques are applied separately. If the watermark is encrypted prior to embedding into the host image, an unauthorized user cannot restitute the original watermark without using correct keys. Consequently, the joint encryption-watermarking technique can achieve a higher security level. A number of joint encryption-watermarking systems have been reported [32–37].
The newly introduced theory of compressive sensing (CS) [38] has attracted the interest of researchers. It is constructed on the inherent sparsity of images and can recover the compressed images with desirable quality from much fewer compressed data. Subsequently, various image encryption techniques based on compressive sensing have been studied [39–41].
Recently, an efficient double-image encryption system combining CS with the discrete fractional random transform (DFRT) has been designed. The measurement matrix in CS and the DFRT are constructed with the random circular matrix controlled by the 2D sine logistic modulation map. The images to be encrypted are represented in the discrete wavelet transform domain [42]. Further, a simultaneous image compression, fusion and encryption approach based on CS and chaos has been proposed to ensure the efficiency and security of image transmission [43]. A binary-tree encryption strategy has been put forwarded. In this approach, encryption units are regarded as nodes and plain images are input only into leaf nodes. This scheme is used to realize a secure authority management among the users sharing a cipher image [44]. Lately, a color image watermarking scheme based on CS in the gyrator transform domain has been presented. With the use of human visual characteristics, the significant blocks of the grayscale host image are chosen to formulate the appropriate reference image and the compressed color watermark is embedded into it to achieve camouflage property to some extent [45].
In this paper, for the first time to author’s knowledge, a new asymmetric multiple image security scheme using wavelet transform and gyrator transform is proposed. In this method, a set of four color images of each user are individually a single-level 2-D discrete wavelet transformed to split into their four sub-bands, corresponding sub-bands are fused to get a single fused image as an input image and then divided into and channels. Each channel is compressed by compressive sensing with measurement matrices (which are constructed by exploiting the circulant matrices and controlling the original row vectors of the circulant matrices with logistic map) and modulated by random phase mask generated by using measurement matrices. The modulated and channels of sets are separately multiplexed and then gyrator transformed. The phase of encrypted and channels are, respectively, embedded into and channels of host image to get corresponding watermarked channels. The amplitudes of encrypted and channels supply common decryption keys. The decrypted images are free from cross-talk noise effects. The security system can be implemented by using optoelectronic setup. Numerical simulation results show the validity and reliability of the proposed scheme.
Optical encryption systems evolved into multiple image encryption because, for the majority of data communications that take place today, several users must simultaneously share a common channel resource in a controlled and effective way. The usual multiplexed package is synthesized by superimposing individual encrypted images together. Digitally speaking, all the images are superimposed in one composite CCD frame, and each one of them can be independently reconstructed through a digital spatial filtering. Since no direct combination of the information of multiple hidden images is employed in the encryption process, each image can be perfectly retrieved without cross-talk caused by the existence of the other.
The proposed technique has three advantages compared with reported MIE systems. First, each set has initial value and bifurcation parameter as remarkably sensitive decryption keys as well as individual decryption key. Second, the security system has a common decryption key, which provides extra security layer. Third, the phase of encrypted image is embedded into host image to get watermarked image in order to achieve imperceptibility and robustness.
2. Theory
2.1 Compressive sensing
Compressive sensing (CS) is a mathematical paradigm which allows reconstructing the data from a substantially smaller number of measurements than those imposed by the Shannon–Nyquist theorem [38].
Suppose one-dimensional compressible signal (or image) with samples (or pixels) has a sparse representation under an arbitrary orthogonal basis matrix (or sparsifying operator)
where denotes the transform coefficient vector. It is an -sparse representation of signal (or image) projected on meaning, that has only nonzero entries while the remaining has zero entries.The sensing (or measurement) vector can be written as
where is a measurement matrix (or optical sensing operator). If satisfies the restricted isometry property (RIP) of order with isometry constant ,for all -sparse signals . Then original signal (or image) is reconstructed by solving norm minimization problem [46]. The estimated coefficients vector is the solution of the non-convex optimization program:In brief, a signal or image is transformed into its sparse form. The measurement matrix is exploited to compress the data. The smoothed norm algorithm is adopted to recover the signal.
2.2 Logistic map
The one-dimensional (1D) non-linear chaos function is a logistic map and its iterative form is expressed as
whereis called bifurcation parameter: and are iterative and initial values.The measurement matrix is constructed by circulant matrix [42]. First, the random sequence with length is produced by logistic map with initial value . The preceding elements are discarded to get new sequence. Second, the new sequence is used as the first row vector of the circulant matrix . The first row vector is circulated to construct its other row vectors as
where and . In order to reduce the relevance among the column vectors, the first element of vector is set as .2.3 Gyrator transform
The gyrator transform (GT), at transformation angle , of a two-dimensional function is given by [6]
where represents GT operator. and are the input and output coordinates, respectively. GT setup contains three generalized lenses with fixed distance z between them. Each generalized lens is an assembled set of two cylindrical lenses. The variation of the transformation parameter is achieved by the rotation of the cylindrical lenses. The optical GT system does not require axial movements.3. Proposed cryptosystem
The block diagram of proposed encryption method is shown in Fig. 1
. It consists of following steps:Step 1: A color image is a single-level 2-D discrete wavelet transformed to decompose into and sub-bands where subscripts and represent low and high frequency parts, respectively.
Each set contains four color images. Let sets of four color images be ,,, and The corresponding sub-bands are represented as
Step 2: The sets of four sub-bands and are fused as
Step 3: The fused image is split into and channels denoted as , and , respectively.
Step 4: The measurement matrices and with size of are constructed with corresponding parameters and by using Eq. (6).
Step 5: , and are, respectively, multiplied by and , and then modulated by corresponding phase masks , , and .
Step 6: The sets of compressed and channels are separately combined together as
Step 7: The combined and channels are, respectively, gyrator transformed at transformation angle and to obtain corresponding resultant images.
The phase and amplitude of the resultant images are given by
Equation (17) is used as the common asymmetric/decryption key.Step 8: The host image of size pixels is decomposed into and channels denoted as , and , respectively. , and (employed and channels of watermark image) are embedded into corresponding , and to obtain watermarked image as
where is a real weight factor.The individual asymmetric/decryption keys of and channels are generated asThe block diagram of proposed decryption method is shown in Fig. 2
. It consists of following steps:Step 1: The and channels of watermark image are retrieved as
Step 2: The and channels of watermark image are multiplied with corresponding second set of decryption keys and then gyrator transformed at transformation angle and.
Step 3: The obtained and channels are modulated by corresponding conjugate of phase masks and then multiplied by corresponding decryption keys.
Step 4: The and channels are decrypted by using algorithm with corresponding measurement matrices and .
Step 5: Finally, the fused image is decomposed into and which represent decrypted images as, , , and, respectively.
A single-level 2-D discrete wavelet is exploited to decompose each original image. The low-frequency part occupies most of energy. So low-frequency parts of four images are fused into a new image (as a set of four images). As the number of the original images increases, the level of discrete wavelet increases, which means that the number of pixels in the low frequency part decreases and corresponding energy occupies less. In other words, if the number of images increases (more than four), the qualities of decrypted images deteriorate.
Asymmetric cryptography, also known as public key cryptography, uses public (/encryption) and private (/decryption) keys to encrypt and decrypt data, respectively. The key distribution scheme is as follows:
- (1) If the sender (say Ali) wants to send information to multi-user, he first needs to register common decryption keys (as authentication keys) in information database of the multi-user.
- (2) Ali encrypts sets of images using the individual public keys of multi-user and common public keys and then generates common decryption keys. The encrypted image embeds into a host image and then sends the resulting watermarked image to multi-user.
- (3) User 1 exploits Ali’s common public and decryption keys to recover the ciphertext and then decrypts the set of images using the individual public and decryption keys.
- (4) User 2 exploits Ali’s common public and decryption keys to recover the ciphertext and then decrypts the set of images using the individual public and decryption keys, and so on.
The proposed encryption and decryption processes can be performed with the optoelectronic setup as shown in Figs. 3(a) and 3(b)
, respectively. In encryption system, the pre-processing operations are achieved digitally by means of a computer system to obtain the final compressed red channel, which is imported into the spatial light modulator and converted by the optical system of gyrator transform. The reference beam , similar to inline holography, is used to record the phase distribution of output image by a charge-coupled device camera. The post-processing operations are carried out digitally by using the computer system to get the watermarked red-channel . In the same way, and are obtained.In decryption system, the pre-processing operations are accomplished digitally by means of a computer system to get the red channel function , which is modulated on and converted by the optical system of inverse gyrator transform. The reference beam is used to record the output image by a camera and stored in the computer system to obtain decrypted image digitally. The output image is multiplied by corresponding conjugate of phase mask and decryption key. Subsequently, fused red channel is recovered by using algorithm with measurement matrix. Similarly, fused blue and green channels are recovered using the same processes. Finally recovered fused red, blue, and green channels are combined to produce fused color image, which is decomposed into four color images of the selected set.
4. Numerical results
Numerical simulations have been performed on a Matlab 9.0 (R2016a) platform to test the validly and security of the proposed technique. The four images of set I, set II, and set III are, respectively, shown in Figs. 4(a)–4(d), 4(e)–4(h), and 4(i)–4(l)
. The size of each image is pixels. The fused color image of set I, set II, and set III are, respectively, shown in Figs. 4(m)–4(o). The parameters of set I, set II, and set III are, respectively, and The compression ratio of column is taken as so the measurement matrix becomes pixels. The transformation angles of the GT for and channels are, respectively, and . The phase masks of set I, set II, and set III are, respectively, shown in Figs. 5(a)–5(c). The individual decryption phase keys of set I, set II, and set III are, respectively, demonstrated in Figs. 5(d)–5(f). The amplitude and phase of final encrypted images are depicted in Fig. 5(g) and Fig. 5(h), respectively. As the images are encoded into noise-like signals, no useful information can be detected. The watermarked host image is illustrated in Fig. 5(i). The size of each image displayed in Figs. 5(a)–5(i) is pixels.To weight the difference between the input image and output image, the correlation coefficient (CC) is given by
where and are output and input images, respectively. is the expected value operator. The CC has the value if the two images are fully correlated, if they are completely uncorrelated, and if they are totally anti-correlated. For simplicity, only the fused color image of set I has been studied.First, the security analysis of the proposed system has been investigated. The reconstructed images of individual images of set I without individual decryption keys (/asymmetric keys) are displayed in Figs. 6(a)–6(d)
. The CC values of and channels are, respectively, ( ), ( ), ( ), and ( ). The CC values of and channels are very low. The reconstructed images of individual images of set I without common decryption keys (/asymmetric keys) are displayed in Figs. 6(e)–6(h). The CC values of and channels are, respectively, ( ), ( ), ( ), and ( ). The CC values of and channels are very low. The retrieved images of individual images of set I without conjugate phase keys are presented in Figs. 6(i)–6(l). The CC values of and channels are, respectively, ( ), ( ), ( ), and ( ). The CC values of and channels are very low. It can be inferred that the decrypted images will not visually render any information about the input image without using decryption keys or conjugate phase keys.The initial values and of corresponding and channels fused color image of set I are changed by The CC values of and channels for individual images of set I as shown in Figs. 7(a)–7(d)
are, respectively, ( ), ( ), ( ), and ( ). The CC values of and channels are low. Thus, no information about the input image can be extracted. The bifurcation parameters and of corresponding and channels fused color image of set I are varied by The CC values of and channels for individual images of set I as demonstrated in Figs. 7(e)–7(h) are, respectively, ( ), ( ), ( ), and ( ). The CC values of and channels are very low. Hence, no information about the input image can be observed.The transformation angles and of multiplexed image are independently changed by The recovered individual images of set I are, respectively, depicted in Figs. 8(a)–8(d), 8(e)–8(h), and 8(i)–8(l)
. It can be observed that only noise-like images are obtained. The transformation angles and of multiplexed image are together varied by The extracted individual images of set I are, respectively, depicted in Figs. 8(m)–8(p). The corresponding CC values of and channels are ( ), ( ), ( ), and ( ). The CC values of and channels indicate that decrypted images will imperceptible. Figures 9(a)-9(c) show the decrypted fused color images of Set I, Set II, and Set III with all correct keys, respectively. The individual images Set I, Set II, and Set III are depicted in Figs. 9(d)-9(g), 9(h)-9(k), and 9(l)-9(o), respectively. The corresponding CC values are greater than 0.988, which is close to one. So, input images are decoded correctly with slight distortion.Second, the sensitivity analysis of security keys of fused color image of Set I of the proposed scheme has been tested. The CC values between the decrypted images and original images of and channels of fused color image of Set I are computed against the variation of initial value, bifurcation parameter and transformation angle and plotted in Fig. 10(a)-10(c)
, respectively. The curves of and channels clearly show that the CC values of corresponding channels attain one when the security keys are correct during decryption process whereas the curves of and channels decrease rapidly for slight change in CC values of corresponding channels. The parameters , and provide sensitive keys and thus enhance the security the proposed method.Finally, the attack analysis of the proposed scheme has been examined. In the proposed security system, different ciphertext of fused image has different decryption keys (/asymmetric keys) which mean that the chosen ciphertext attack [14], and chosen plaintext attack [15] will not work here. Moreover, the combined compressed channel is treated as a part of the input information; unauthorized users do not have enough constraints to set up the iterative amplitude-phase retrieval [15] even if they have obtained the correct transformation angle of corresponding channel as shown in Fig. 11
.5. Conclusion
A new nonlinear multiple image security method based on wavelet transform and compressive gyrator transform is put forwarded. The proposed technique has benefits of individual decryption keys, chaotic parameters as very sensitive decryption keys and common decryption keys, which enhance the nonlinear characteristics of the scheme. The phase of encrypted image is embedded into corresponding host image to obtain watermarked image so that imperceptibility and robustness can be maintained. The proposed algorithm not only reduces data volume but also simplify keys, which improve the efficiency of transmitting data and distributing keys. Thus, the proposed method is of great practical importance as the image fusion, compression, encryption, and watermarking are accomplished simultaneously. The security and viability are proved by the numerical simulation results.
Acknowledgments
The author is indebted to Abdul Aziz RA and Muhammad Sulayman RA for their inspiring supports.
References
1. A. Alfalou and C. Brosseau, “Optical image compression and encryption methods,” Adv. Opt. Photonics 1(3), 589–636 (2009). [CrossRef]
2. S. Liu, C. Guo, and J. T. Sheridan, “A review of optical image encryption techniques,” Opt. Laser Technol. 57, 327–342 (2014). [CrossRef]
3. P. Refregier and B. Javidi, “Optical image encryption based on input plane and Fourier plane random encoding,” Opt. Lett. 20(7), 767–769 (1995). [CrossRef] [PubMed]
4. G. Unnikrishnan, J. Joseph, and K. Singh, “Optical encryption by double-random phase encoding in the fractional Fourier domain,” Opt. Lett. 25(12), 887–889 (2000). [CrossRef] [PubMed]
5. G. Situ and J. Zhang, “Double random-phase encoding in the Fresnel domain,” Opt. Lett. 29(14), 1584–1586 (2004). [CrossRef] [PubMed]
6. J. A. Rodrigo, T. Alieva, and M. L. Calvo, “Experimental implementation of the gyrator transform,” J. Opt. Soc. Am. A 24(10), 3135–3139 (2007). [CrossRef] [PubMed]
7. Z. Liu, M. A. Ahmad, and S. Liu, “Image sharing scheme based on combination theory,” Opt. Commun. 281(21), 5322–5325 (2008). [CrossRef]
8. M. R. Abuturab, “Color image security system using double random-structured phase encoding in gyrator transform domain,” Appl. Opt. 51(15), 3006–3016 (2012). [CrossRef] [PubMed]
9. J. F. Barrera, A. Mira, and R. Torroba, “Optical encryption and QR codes: secure and noise-free information retrieval,” Opt. Express 21(5), 5373–5378 (2013). [CrossRef] [PubMed]
10. W. Liu, Z. Liu, and S. Liu, “Asymmetric cryptosystem using random binary phase modulation based on mixture retrieval type of Yang-Gu algorithm,” Opt. Lett. 38(10), 1651–1653 (2013). [CrossRef] [PubMed]
11. W. Chen and X. Chen, “Object authentication in computational ghost imaging with the realizations less than 5% of Nyquist limit,” Opt. Lett. 38(4), 546–548 (2013). [CrossRef] [PubMed]
12. M. Aldossari, A. Alfalou, and C. Brosseau, “Simultaneous compression and encryption of closely resembling images: application to video sequences and polarimetric images,” Opt. Express 22(19), 22349–22368 (2014). [CrossRef] [PubMed]
13. Z. Liu, C. Shen, J. Tan, and S. Liu, “A recovery method of double random phase encoding system with a parallel phase retrieval,” IEEE Photonics J. 8(1), 7801807 (2016). [CrossRef]
14. Y. Frauel, A. Castro, T. J. Naughton, and B. Javidi, “Resistance of the double random phase encryption against various attacks,” Opt. Express 15(16), 10253–10265 (2007). [CrossRef] [PubMed]
15. W. Qin and X. Peng, “Asymmetric cryptosystem based on phase-truncated Fourier transforms,” Opt. Lett. 35(2), 118–120 (2010). [CrossRef] [PubMed]
16. X. Wang and D. Zhao, “A special attack on the asymmetric cryptosystem based on phase-truncated Fourier transforms,” Opt. Commun. 285(6), 1078–1081 (2012). [CrossRef]
17. X. Wang and D. Zhao, “Security enhancement of a phase-truncation based image encryption algorithm,” Appl. Opt. 50(36), 6645–6651 (2011). [CrossRef] [PubMed]
18. M. R. Abuturab, “Color information cryptosystem based on optical superposition principle and phase-truncated gyrator transform,” Appl. Opt. 51(33), 7994–8002 (2012). [CrossRef] [PubMed]
19. M. R. Abuturab, “Security enhancement of color image cryptosystem by optical interference principle and spiral phase encoding,” Appl. Opt. 52(8), 1555–1563 (2013). [CrossRef] [PubMed]
20. G. Situ and J. Zhang, “Multiple-image encryption by wavelength multiplexing,” Opt. Lett. 30(11), 1306–1308 (2005). [CrossRef] [PubMed]
21. Y. Shi, G. Situ, and J. Zhang, “Multiple-image hiding in the Fresnel domain,” Opt. Lett. 32(13), 1914–1916 (2007). [CrossRef] [PubMed]
22. D. Amaya, M. Tebaldi, R. Torroba, and N. Bolognini, “Wavelength multiplexing encryption using joint transform correlator architecture,” Appl. Opt. 48(11), 2099–2104 (2009). [CrossRef] [PubMed]
23. A. Alfalou and C. Brosseau, “Exploiting root-mean-square time-frequency structure for multiple-image optical compression and encryption,” Opt. Lett. 35(11), 1914–1916 (2010). [CrossRef] [PubMed]
24. H. T. Chang, H.-E. Hwang, C.-L. Lee, and M. T. Lee, “Wavelength multiplexing multiple-image encryption using cascaded phase-only masks in the Fresnel transform domain,” Appl. Opt. 50(5), 710–716 (2011). [CrossRef] [PubMed]
25. A. Alfalou, C. Brosseau, N. Abdallah, and M. Jridi, “Assessing the performance of a method of simultaneous compression and encryption of multiple images and its resistance against various attacks,” Opt. Express 21(7), 8025–8043 (2013). [CrossRef] [PubMed]
26. S. Liansheng, X. Meiting, and T. Ailing, “Multiple-image encryption based on phase mask multiplexing in fractional Fourier transform domain,” Opt. Lett. 38(11), 1996–1998 (2013). [CrossRef] [PubMed]
27. M. R. Abuturab, “Generalized Arnold map-based optical multiple color-image encoding in gyrator transform domain,” Opt. Commun. 343, 157–171 (2015). [CrossRef]
28. M. R. Abuturab, “Fully phase multiple information encoding based on superposition of two beams and Fresnel-transform domain,” Opt. Commun. 356, 306–324 (2015). [CrossRef]
29. S. Liansheng, Z. Bei, N. Xiaojuan, and T. Ailing, “Optical multiple-image encryption based on the chaotic structured phase masks under the illumination of a vortex beam in the gyrator domain,” Opt. Express 24(1), 499–515 (2016). [CrossRef] [PubMed]
30. Q. Wang, A. Alfalou, and C. Brosseau, “Security enhanced multiple-image authentication based on cascaded optical interference and sparse phase mixed encoding,” Opt. Commun. 372, 144–154 (2016). [CrossRef]
31. M. R. Abuturab, “Multiple color-image authentication system using HSI color space and QR decomposition in gyrator domains,” J. Mod. Opt. 63(11), 1035–1050 (2016). [CrossRef]
32. Z. Liu, L. Xu, Q. Guo, C. Lin, and S. Liu, “Image watermarking by using phase retrieval algorithm in gyrator transform domain,” Opt. Commun. 283(24), 4923–4927 (2010). [CrossRef]
33. N. Singh and A. Sinha, “Digital image watermarking using gyrator transform and chaotic maps,” Optik (Stuttg.) 121(15), 1427–1437 (2010). [CrossRef]
34. Z. Shao, Y. Duan, G. Coatrieux, J. Wu, J. Meng, and H. Shu, “Combining double random phase encoding for color image watermarking in quaternion gyrator domain,” Opt. Commun. 343, 56–65 (2015). [CrossRef]
35. A. K. Yadav, S. Vashisth, H. Singh, and K. Singh, “A phase-image watermarking scheme in gyrator domain using devil’s vortex Fresnel lens as a phase mask,” Opt. Commun. 344, 172–180 (2015). [CrossRef]
36. M. R. Abuturab, “Group multiple-image encoding and watermarking using coupled logistic maps and gyrator wavelet transform,” J. Opt. Soc. Am. A 32(10), 1811–1820 (2015). [CrossRef] [PubMed]
37. M. R. Abuturab, “Multiple color-image fusion and watermarking based on optical interference and wavelet transform,” Opt. Lasers Eng. 89, 47–58 (2017). [CrossRef]
38. D. L. Donoho, “Compressed sensing,” IEEE Trans. Inf. Theory 52(4), 1289–1306 (2006). [CrossRef]
39. N. Zhou, A. Zhang, F. Zheng, and L. Gong, “Novel image compression-encryption hybrid algorithm based on key-controlled measurement matrix in compressive sensing,” Opt. Laser Technol. 62, 152–160 (2014). [CrossRef]
40. B. Deepan, C. Quan, Y. Wang, and C. J. Tay, “Multiple-image encryption by space multiplexing based on compressive sensing and the double-random phase-encoding technique,” Appl. Opt. 53(20), 4539–4547 (2014). [CrossRef] [PubMed]
41. J. Yi and G. Tan, “Optical compression and encryption system combining multiple measurement matrices with fractional Fourier transform,” Appl. Opt. 54(36), 10650–10658 (2015). [CrossRef] [PubMed]
42. N. Zhou, J. Yang, C. Tan, S. Pan, and Z. Zhou, “Double-image encryption scheme combining DWT-based compressive sensing with discrete fractional random transform,” Opt. Commun. 354, 112–121 (2015). [CrossRef]
43. X. Liu, W. Mei, and H. Du, “Simultaneous image compression, fusion and encryption algorithm based on compressive sensing and chaos,” Opt. Commun. 366, 22–32 (2016). [CrossRef]
44. J. Yi and G. Tan, “Binary-tree encryption strategy for optical multiple-image encryption,” Appl. Opt. 55(20), 5280–5291 (2016). [CrossRef] [PubMed]
45. L. Sui, B. Zhou, Z. Wang, and A. Tian, “An optical color image watermarking scheme by using compressive sensing with human visual characteristics in gyrator domain,” Opt. Lasers Eng. 92, 85–93 (2017). [CrossRef]
46. H. Mohimani, M. Babaie-Zadeh, and C. Jutten, “A fast approach for over complete sparse decomposition based on smoothed l0 norm,” IEEE Trans. Signal Process. 57(1), 289–301 (2009). [CrossRef]