An Optimal Teaching and Learning based Optimization with Multi-Key Homomorphic Encryption for Image Security

Due to the drastic rise in multimedia content, digital images have become a major carrier of data. Generally, images are communicated or archived via wireless communication changes, and the significance of data security gets increased. In order to accomplish security, encryption is an effective technique which is used to encrypt the images using secret keys in such a way that it is not readable by the hacker. In this view, this study focuses on the design of Teaching and Learning based Optimization (TLBO) with Multi-Key Homomorphic Encryption (MHE) technique, called MHE-TLBO algorithm. The goal of the MHE-TLBO algorithm is to optimally select multiple keys using TLBO algorithm for encryption and decryption processes. In addition, the MHE-TLBO algorithm has derived a fitness function involving peak signal to noise ratio (PSNR) and thereby ensures the superior quality of the reconstructed image. For validating the security performance of the MHE-TLBO algorithm, a comprehensive result analysis is made and the simulation results ensured the betterment of the MHE-TLBO algorithm interms of different aspects.


Introduction
Recently, with the rapid popularization and promotion of digital communication and network technologies across the globe, digital video based digital images became a significant means for data transmission and storage in the computer network under the fields of military and civil [1,2]. But network security problems have been a significant factor which restricted and plagued the growth of network technologies [3,4]. Particularly, in the data resource of the government department and public, to comprehend the information security in the computer networks are the significant direction and content of the study fields in information security and network security. Amongst other, digital video and digital image became the significant contents of broadcast information in the network with the virtues of its convenience and intuitiveness [5,6]. Hence, the security protections of digital images have attained more interest from each party. Particularly, in the background of increasing serious network security situations over the past few years, data sharing and transmission depending on digital image frequently facing the problem of information tampering, theft, attack, and deletion that have created huge losses to the publishers/owners of digital image [7][8][9]. In the security method of digital data, encryption technique is a more popular method and technique. Encryption techniques can be encrypted by encrypting the new information. When the reliability and security of the encryption technique are relatively large, the security of digital data could be secured [10]. Hence, the study on digital image encryption methods and technology is a significant direction for digital image security. But, encryption system/encryption technology is based largely on the requirement of text encryption. Now, the most popular encryption scheme could not attain better results in the encryption quality and compatibility of digital image encryptions. Even though digital images could be treated as 2D datasets, cryptographic systems which straightforward utilize text encryption methods frequently facing the problem of inefficacy in decryption and encryption, low security, and low practicability [11,12]. Studying an encryption method/cryptographic system appropriate for digital image encryptions is a certain technique for protecting the security of digital images in network environments.
Khan et al. [13] proposed an effective method of making high nonlinear cryptographic substitution boxes as an alternative to algebraic/chaotic construction method. The PSO method is used in the construction of high nonlinear S-box, in the proposed method the first populations are arbitrarily generated, and the location vectors of particles are employed in producing S-box. A hyperchaotic image encryption scheme [14] is presented according to PSO and CA models. First of all, to increase the capability for resisting plaintext attack, the early condition of the hyperchaotic scheme is produced by the hash function values i.e., nearly associated with the plaintext images to be encrypted. Furthermore, the fitness of PSO is the relation coefficients among nearby pixels of images.
Farah et al. [15] proposed a novel image cipher on the basis of diffusion or confusion Shannon property. The presented approach is based on novel enhanced substitution boxes, has been performed using chaotic Jaya optimization approach for generating S-box based on their non-linearity scores. The aim of the optimization method is to contain a bijective matrix using higher non-linearity scores. [16] studied an image encryption method according to MOPSO method, 1D Logistic maps, and DNA encoding sequences. Initially, the primary objective of this work includes the subkey sequences elected with PSO method, hash values of the shuffle mark bit, and plaintext images. Generate arbitrary DNA mask image by DNA encoding and Logistic map. Later, utilize the block shuffling plaintext DNA encoding sequences for operating an encryption method.
In [17], hyperchaotic maps are enhanced by a multi-objective evolution optimization method. The DLS-MO method is employed for obtaining the optimum variables of an encryption factor and hyperchaotic map. Next, with an optimum parameter, a hyperchaotic map develops the secret key. This secret key is later employed for performing diffusion and permutation on a plain image for developing the encrypted image. In [18], a novel medicinal image encryption method integrating GSAPSO and MQC systems is presented for obtaining improved security performances. First, an enhanced MQC method is employed for generating is used key streams. Later, the cross operation and election of GA method are employed for processing the plaintext images. The optimum sequence created using SA method is applied for scrambling. While the PSO method is presented for the SA method. The early temperature is fixed based on the best fitness values of the primary population.
In [19], a hybrid image encryption technique was presented depends on chaos and GA. The encryption method includes 3 major phases: improvement, confusion, and diffusion stages with a GA method. Initially, Chen's chaotic map is employed in the confusion stage to make scrambled images with shuffling plain image pixels, and in diffusion phase, Logistic Sine map alters this pixel's gray level value. It creates few encrypted images that are deliberated as the primary population for the GA method. Later, with GA method, the encrypted image is enhanced.
This study focuses on the design of Teaching and Learning based Optimization (TLBO) with Multi-Key Homomorphic Encryption (MHE) technique, called MHE-TLBO algorithm. The goal of the MHE-TLBO algorithm is to optimally select multiple keys using TLBO algorithm for encryption and decryption processes. In addition, the MHE-TLBO algorithm has derived a fitness function involving peak signal to noise ratio (PSNR) and thereby ensures the superior quality of the reconstructed image. For validating the security performance of the MHE-TLBO algorithm, a comprehensive result analysis is made interms of different measures. security procedures. Semantically secured homomorphic public key encrypted techniques were focal cryptographic apparatus to any protected multiparty computation problems. The property of homomorphic has been valuable for building a secure model with high-security data recovery plan. These encryption structures are employed for performing tasks utilizing encoder data without significant the private key (with no decryption), for instance, the customer is a vital holder of the confidential key [20]. The homomorphic assessment technique produces as to account the polynomial many cipher images that are encrypted in N keys, composed of the connecting assessment key, and deliver the cipher images. In multiple encryptions were the only manner near altering on a unique message as to confused shape with carrying out encrypted to the number of times, also with implementing similar or distinctive technique processes. It could be demonstrated as cascade encryption, cascade ciphering, together with several encryptions. To examine model, encrypt as well as decrypt are carried out with several keys. The presented MHE considers 3 stages like multiple key generations, optimum key determination, and encrypt/decrypt approach [21].

Fig. 1. Block diagram of MHE-TLBO model
A key has been employed for encrypting/decrypting whatever data has been encrypted or decrypted. This technique was utilized to encoded as well as decoded keys and the compared images utilizing symmetric keys; both privacy and trust value security are provided. The private key (prk) and relative public key(puk) are in several key sets. The key equal was employed with asymmetric key technique; at this point, several keys K = {K1, K2, ..… Kn} was created to MHE. In specific cases, the keys are arbitrarily created utilizing a Random Number Generator. In order to select the optimum key in several keys, TLBO technique was implemented.
The TLBO technique was simulated in the knowledge transmission amongst the teacher as well as students from the educational periods [22]. Assume that 2 teachers, T1 and T2, teach a subject with similar data to an identical merit level learner from 2 dissimilar classes. Curves 1 and 2 demonstrate the marks reached as the learner taught by teacher T1 and T2 correspondingly. The normal distribution has been regarded as to obtained marks, however, it should be skewness. Normal distribution was demonstrated in Eq. (1).
where σ2 implies the variance, μ refers the mean and x is some value to that the normal distribution function was required. Curve 2 referred to optimum efficiency related to curve 1 and so it can be stated that T2 is higher than T1 with respect to teaching. An important difference amongst the efficiency was measured by calculating the mean amongst them (M2 for Curve-2 and M1 for Curve-1), for instance, an excellent teacher attains optimum mean to the outcomes of learner. The learner also learns to interface among themselves and it also supports improving its outcomes. In the above teaching procedure, the mathematical process was complete for optimizing the unconstrained non-linear continuous function, and so planning an effectual optimized manner called as TLBO technique. There are 2 essential levels from Where TF signifies the teaching issue that calculates the value of mean to be changed, and ri implies the arbitrary number from range of 0 and 1. The value of TF is one/two that is yet over a heuristic step and obvious in an arbitrary approach with similar probabilities as TF= round

Conclusion
This study has designed a new MHE-TLBO algorithm to ensure security in image transmission. The goal of the MHE-TLBO algorithm is to optimally select multiple keys using TLBO algorithm for encryption and decryption processes. In addition, the MHE-TLBO algorithm has derived a fitness function involving peak signal to noise ratio (PSNR) and thereby ensures the superior quality of the reconstructed image.