ENHANCING home security through visual CRYPTOGRAPHY

Abstract

Home security systems in the recent times have gained greater importance due to increasing threat in the society. Biometrics deals with automated approaches of recognizing a user or verifying the user identity based on behavioral or physiological features. Visual cryptography is a scheme of secret sharing where a secret image is encrypted into shares which disclose no data independently about the original secret image. As the template of biometric are stored in centralized database due to the threats of security the template of biometric may be changed by attacker. If the template of biometric is changed then the authorized user will not be permitted to access the resource. To manage this problem the schemes of visual cryptography can be used to secure the face recognition. Visual cryptography offers huge ways for supporting such needs of security as well as additional authentication layer. To manage this problem the visual cryptography schemes can be used to secure digital biometric information privacy. In this approach the face or private image is dithered in two varied host images that is sheets and are stored in separate servers of data so as to assure that the original image can get extracted only by accessing both sheets together at a time and a single sheet will not be capable to show any data of private image. The main aim of the study is to propose an algorithm which is a combination of CVC and Siamese network. This research implements visual cryptography for face images in a biometric application. The Siamese network is essential to solve one shot learning by representation of learning feature that are compared to verification tasks. In this research face authentication helps in accomplishing robustness by locating face image from an n input image. This research explores the availability of using visual cryptography for securing the privacy to biometric data. The results of the proposed approach provide an accuracy of 93% which is found to be superior when compared with that of the approaches that are already in practice.

Introduction

Security and information privacy have been playing an essential part in the life of human beings. Information indicates essential perspectives of daily life and personal authentication has acquired huge attention as it is becoming an essential problem [1]. Various techniques have been employed for securing private and personal data involving steganographic, cryptographic and biometric techniques. Biometric technique offers automatic system for securing data and enabling the user's authentication based on biological and behavioral data such as iris, voice, face, palm veins and fingerprint. These approaches are supposed to be much accurate and secure for identification and authentication due to their greater accuracy compared to non-biometric techniques [19]. Continuous development in smart devices number and its related loads of connectivity has influenced mobile devices seamlessly provided anywhere around the world. Authentication is one of the major factor to manage this process securely [18]. The process where a user recognizes himself by sending the image to the system is known as authentication. The system authenticates her or his recognition by estimating F(a) and verify that it is relevant to the stored b value. Authentication is considered as a major safeguard against illegal access of device or any other sensitive application whether online or offline [4], [13]. In computer information systems authentication has been used by tradition based on something which one has for instance smart cards, magnetic strip cards or even keys or what one knows for instance passwords and usernames, secret codes or PIN codes [5].

The first factor which is used for authentication is face recognition. Initially the concerned person face will be stored in the database when she or he enters the system for the first time. Now if the user needs to acquire access to the system her or his face would be matched with earlier taken image which is stored in the database already and if the face matches then the user must be ready for next authentication [20]. The main purpose of multi factor authentication is to offer greater extent of assurance of individual recognition trying to use a resource such as computing device, physical location, database or a network. The mechanisms of authentication used for multi factor authentication must be independent of one another such that one factor access does not allow another factor access and one factor compromise does not influence the confidentiality or integrity of other factor [14]. The below Fig. 1 shows the authentication methods evolution from Single Factor Authentication to Multi Factor Authentication:

Multi factor authentication is a security system that needs greater than one authentication approach from independent types of credentials to ensure the identity of user for other transaction or login. Multifactor authentication integrates more than two independent credential factors namely what the user is (biometric verification) and what the user knows (password) and what the user has (token for security). The main aim of multi factor authentication is to create a layered location, database of network and computing device. If one factor is broken or compromised then the attacker has one more challenge to breach before breaking into target successfully. Multifactor authentication usually relies on 2FA (two factor authentication). Vendors are using the multifactor label increasingly to explain the scheme of authentication that needs greater than one credential of identity [10]. Two factor authentication enhances the access control system security where two factors are integrated to authenticate a user. The main purpose of two factor authentication is to add up the security of two factors. These factors involve passwords indicating something which the user knows or physical tokens namely smart cards indicating something which the user have. Additionally, the traits of biometrics are used indicating something which the user are [16]. These approaches need the biometric template storage on a database for the purpose of matching [6]. The techniques namely visual cryptography has been employed to offer BTP (biometric template protection) by distribution template storage to both the system administrator and user [17].

Establishing the individual recognition is very essential in many applications. Biometric based recognition technique is a convenient and reliable way for identification of an individual. It utilizes behavioral and physiological features of individuals and is becoming highly familiar compared to traditional knowledge or token-based techniques such as passwords, identification cards, etc. A biometric authentication system performs by obtaining raw biometric information from a subject that is face. This original raw data is stored in central database. Securing the digital biometric information privacy has become very essential. Visual cryptography can be used for securing this data [8]. A method for visual image encryption where the decryption is achieved without the requirement for whole mathematical algorithms is referred as visual cryptography. In this method a secret image encryption is carried out into n number of noise images randomly referred as sheets. A decryption is feasible only when at least k number of sheets is feasible and integrated using a logical operator. Visual cryptography has been used to secure the privacy of raw digital biometric data stored in central DB. An input image of biometrics is decomposed into 2 components such that the actual information can be recovered when both the components are feasible simultaneously and separate components which cannot be matched easily against the actual input biometric information, thereby de-recognizing the input data identity [12]. Visual cryptography is a technique of encryption for images that depends on computational vision or human for decryption. First a plain image is classified into numerous parts referred as shares. Every share has different composition of pixel from the actual image. When individually viewed every share exhibits the noise on a random basis [22]. The below Fig. 2 shows the Visual Cryptography Encryption and Decryption:

The scheme of visual cryptography removes complex computation issue in the process of decryption and the secret images can be restored by stacking operation. This property makes visual cryptography useful for low computation load need [2]. Visual cryptography has two essential characteristics. The first characteristic is its perfect secrecy and the second characteristic is its decryption approach which needs neither complicate decryption algorithm nor the use of PC. It uses only human visual system to recreate the actual image from stacked shares set. It also secures the integrity and privacy of biometric information as the actual image is recovered only when both the image shares are feasible simultaneously. It reduces the cost of storing biometric templates as only one share in the database. In the digital era several users will depend on biometrics concerning authorization and security of system to counterpart conventional passwords. Even though security, privacy, accuracy and usability concern still exist multi factor authentication becomes a system that ensures ease of use and security required for modern users while obtaining sensitive data access [21]. In this study multi factor biometric system based visual cryptography scheme for secure authentication in face has been proposed. Thus, it can be inferred that biometrics is one of the major layers to enable the future of multi factor authentication.