An image edge based approach for image password encryption

Transcription

1 SECURITY AND COMMUNICATION NETWORKS Security Comm. Networks 2016; 9: Published online 16 January 2017 in Wiley Online Library (wileyonlinelibrary.com) RESEARCH ARTICLE An image edge based approach for image password encryption N. K. Sreelaja 1 * and N. K. Sreeja 2 1 Sri Krishna College of Engineering and Technology, Coimbatore, India 2 PSG College of Technology, Coimbatore, India ABSTRACT Authentication plays a major role in ensuring the security of the system by allowing only the authorized user. The traditional authentication system of using text-based passwords has many flaws in the aspects of usability and security issues. Hence, graphical passwords, which consist of clicking or dragging activities on the pictures rather than typing textual characters, were introduced to overcome this problem. However, it is found that these graphical passwords are susceptible to brute force and dictionary attack methods. To overcome this drawback, images can be used as password replacing alphanumeric and graphical passwords. The challenge lies in encrypting the image password and storing it in the database. An image edge password encryption (IEPE) algorithm is proposed to encrypt the image passwords based on its edge values. Experimental results are shown to prove that encrypted image passwords using IEPE algorithm requires a less internal storage space when compared with the existing password encryption techniques. It is also shown that IEPE algorithm scales better when compared with the existing text-based and graphical password authentication methods. Also, IEPE algorithm is shown to resist various password cryptanalytic attacks. Copyright 2017 John Wiley & Sons, Ltd. KEYWORDS authentication; image edge; image password encryption; picture password; security analysis *Correspondence N. K. Sreelaja, Sri Krishna College of Engineering and Technology, Coimbatore, India. sreelajank@gmail.com 1. INTRODUCTION Secure computer systems permit legitimate users to gain access to the system by proving their identity. The users prove their identity using credentials such as username and password. Authentication plays a major role in ensuring the security of the system by verifying the identity of the user and allowing the authorized user to gain access to the system. Out of the several authentication schemes, the most commonly used one is the username and the password. In current practice, alphanumeric passwords are the most widely used mechanism to authenticate users in most of the online applications. According to the studies reported in [1,2], most people find it difficult to remember these passwords. Adams et al. [1] have stated that the short passwords chosen are easy to remember but can be easily guessed or broken. However, the usage of alphanumeric passwords is beset with the trouble of password cracking. One method of cracking text passwords is by tricking people into revealing their password by claiming to be the administrator who phones a user and asks for the user s logon credentials. Also, the alphanumeric password of a person can be found using shoulder surfing. Another form of cracking these passwords is guessing a user s password. Graphical authentication schemes have been proposed as a possible alternative to traditional alphanumeric password techniques, motivated particularly by the fact that humans can remember pictures better than text [3]. However, these methods are also susceptible to password attacks. Although biometrics passwords are considered to be the most secure method of authentication, the drawback of these methods is that it is time-consuming and costlier [4]. Deploying such systems for online applications may be very complex and not suitable. To overcome the drawbacks in the existing authentication methods, it has been proposed to use an image as a password instead of a text password. Because the size of an image is larger than that of a text, the challenge lies in encrypting and storing the image password. Some of the existing hash algorithms in literature for password encryption include MD5, SHA-0, SHA-1, SHA-2, and SHA-3 Copyright 2017 John Wiley & Sons, Ltd. 5733

2 An image edge based approach for image password encryption N. K. Sreelaja and N. K. Sreeja algorithms. The drawback of MD5, SHA-0, and SHA-1 algorithms is that they are susceptible to collision attacks [5]. It is shown that SHA-2 and SHA-3 algorithms require higher internal storage. To overcome these drawbacks in the existing password encryption methods and to reduce the internal storage, a novel algorithm called image edge password encryption (IEPE) algorithm is proposed to encrypt the image password. The design of IEPE algorithm is such that tremendous importance is given to pre-image resistance, collision resistance, and resistance against brute force attacks. The advantage of this method is that there lies no difficulty in remembering the password because the users can store the image password in an external safe storage such as a USB or smart card and upload it for authentication. It is also shown that IEPE algorithm overcomes the drawbacks in the existing password encryption techniques and requires less internal storage when compared with the existing hash algorithms for password encryption. The paper is organized as follows. Section 2 explains the related work. Section 3 explains the model of the system. Section 4 explains image encoding. Section 5 explains IEPE algorithm. Section 6 describes a case study. Section 7 discusses the computational complexity. Section 8 discusses the experimental results. Section 9 compares the existing methods with the proposed method. Section 10 discusses the security analysis. Section 11 presents the conclusion. 2. RELATED WORK Wiedenbeck et al. [6] conducted a user study, in which one group of participants were asked to use alphanumerical password, while the other group was asked to use the graphical password. The result showed that graphical password took fewer attempts for the user than alphanumerical passwords. However, graphical password users had more difficulties learning the password and took more time to input their passwords than the alphanumerical users. In respect to storage and communication, graphical passwords require much more storage space than text-based passwords. Tens of thousands of pictures may have to be maintained in a centralized database. Network transfer delay is also a concern for graphical passwords, especially for recognition-based techniques in which a large number of pictures may need to be displayed for each round of verification [3]. Soumyadeb et al. [7] have proposed the technique of using multiple image passwords for authentication. However, it is stated in their work that the users recorded the passwords using screen captures or written descriptions. Thus, the passwords can be cracked if the recorded note is revealed. Because the images are displayed to the users, this method requires more storage space to store the images. Hong et al. [8] proposed a scheme called Pict-O-Lock for the purpose of picture memorability. The drawback of this approach is that the process is significantly timeconsuming and tedious. Also, the password cannot be guessed but can be broken by brute force method, and the users have to memorize both picture objects and their codes and are more difficult than text-based password. Jermyn et al. [9] proposed a scheme, known as Draw- A-Secret (DAS). The drawback of DAS approach is that the password can be guessed. Sobrado and Birget [4] developed a graphical password technique that deals with the shoulder surfing problem. The main drawback of these algorithms is that the logging in process can be slow. The password can be attacked using brute force search method. Passface [10] is a technique developed by Real User Corporation [10]. The possible password attacks in this method include the dictionary attack, brute force search, password guess, and shoulder surfing. Syukri et al. [11] have proposed a system where authentication is conducted by having the user draw their signature using a mouse. The possible password attacks for this method include the dictionary attack, password guess, and shoulder surfing. Hashing algorithms for password encryption such as SHA-3 algorithm [12] uses the sponge construction in which the message blocks are XORed into the initial bits of the state, which is then invertibly permuted. In the version used in SHA-3, the state array consists of a maximum of 5 5 array of 64-bit words, 1600 bits in total. This algorithm is resistant to cryptanalytic attacks. Another hashing algorithm for password encryption is SHA-2 algorithm [13]. In this approach, the state array consists of a maximum of 8 1 array of 64-bit words, 512 bits in total. This algorithm is resistant to cryptanalytic attacks. However, the internal storage in these algorithms is large. Hashing algorithms such as SHA-0 [14] and SHA-1 [15] cryptographic algorithms are also used to encrypt passwords. In this approach, the internal state size is 160 and the block size is 512. However, the drawback of SHA-0 and SHA-1 algorithms is that both cryptographic algorithms are susceptible to collision attacks [5]. 3. SYSTEM ARCHITECTURE The process of encrypting an image password is explained in two phases. The first phase explains the registration of the image password for a user and the second phase explains the authentication of a user using the image password Registration In the registration phase, the user enters the image password. To encrypt the image password, the edges of the image password are obtained. The edges of the image are 5734 Security Comm. Networks 2016; 9: John Wiley & Sons, Ltd.

3 N. K. Sreelaja and N. K. Sreeja An image edge based approach for image password encryption passed to IEPE algorithm. During encryption, a salt value is generated for the user and given to IEPE algorithm to encrypt the image password. The encrypted image password value along with the salt value for the user is stored in the database. Figure 1 shows the model of the registration system Authentication In the authentication phase, the user enters the username and the image password. The edge values of the image password are obtained and passed to IEPE algorithm. The salt value corresponding to the user is retrieved and passed to the IEPE algorithm. The image password for the user is encrypted and checked for a match with the stored image password value of the user. If a match occurs, the user is authenticated and access is granted. Figure 2 shows the model of an authentication system. 4. CONVERSION OF AN IMAGE INTO EDGE POINTS The edge points of the image are obtained using a Sobel operator and are represented in the form of a binary matrix. The binary matrix takes the values 0 and 1. The edge points are denoted by the value 1. Figure 3 shows the representation of the edge points of an image as a binary matrix. 5. ENCRYPTION OF IMAGE PASSWORD USING IEPE ALGORITHM A novel algorithm called the IEPE algorithm is proposed for encrypting an image password. According to this approach, the edges of an image are obtained as described in Section 4 and passed to IEPE algorithm. Consider the edges of an image represented in the form of a binary matrix M as shown in equation (1), where M ij denotes the values of the binary matrix, 1 < i < m, 1< j < n. m denotes the number of rows, and n denotes the number of columns of matrix M. The number of rows and columns denotes the size of the image. 2 3 M11 M12 M13 M1n M21 M22 M23 M2n M ¼ M31 M32 M33 M3n 5 Mm1 Mm2 Mm3 Mmn The element M ij of the matrix takes the value 0 or 1. The elements in the binary matrix having a value 1 denote an edge point. The binary matrix is traversed, and the edge points are encoded. To encode the edge point, the row position and the column position of the edge point are concatenated denoting the encoded value as shown in equation (2). (1) Figure 1. Model of registration system. IEPE, image edge password encryption. Security Comm. Networks 2016; 9: John Wiley & Sons, Ltd. 5735

4 An image edge based approach for image password encryption N. K. Sreelaja and N. K. Sreeja Figure 2. Model of user authentication system. IEPE, image edge password encryption. encodedimg ¼ Rowencval 1 &Rowencval 2 & &Rowencval m &m&n (4) Figure 4 shows the pseudocode to encode an image. The encryption of the encoded image is carried out by dividing the encoded image into blocks based on a randomly chosen grouping value. The number of blocks is found as shown in equation (5). No : of blocks ¼ lenðencodedimgþ=grouping value (5) If M ij 1; encoded value M ij ¼ i&j (2) The encoded value of the edge points in each row is concatenated denoting row encoded value as shown in equation (3). Rowencval i ¼ Figure 3. Edge value representation. encoded valueðm i1 Þ&encoded value ðm i2 Þ& &encoded valueðm in Þ (3) where i=1, 2,, m. The row encoded value of all rows in the binary matrix is concatenated denoting image encoded value. Finally, the number of rows and columns in the binary matrix is concatenated at the end of the image encoded value denoting the encoded image as shown in equation (4). The encryption key is chosen as follows. The minimum row encoded value is found. Two random numbers are chosen and added to the number of rows and number of columns of the binary matrix. The newly generated row number (NRn) and column number (NCn) is concatenated at the end of the minimum row encoded value denoting the encryption key for the first block. For each successive block, the row number and column number of the previous blocks are incremented with the chosen random numbers and concatenated at the end of the minimum encoded value forming the encryption key for each block. NR n ð1þ ¼ m þ Rnd1 (6) NC n ðþ¼n 1 þ Rnd2 KeyðBlock k NR n ðkþ ¼ NR n ðk 1ÞþRnd1 NC n ðþ¼nc k n ðk 1ÞþRnd2 (7) Þ ¼ minðrowencvalþ&nr n ðkþ&nc n ðþ(8) k where k =1, 2, 3,, No: of blocks Security Comm. Networks 2016; 9: John Wiley & Sons, Ltd.

5 N. K. Sreelaja and N. K. Sreeja An image edge based approach for image password encryption Figure 4. Pseudocode to encode edge values of the image password. An XOR operation is performed between the digits of the encoded image in each block and the encryption key as shown in equation (9). V k ¼ Block k ðencodedimgþxor KeyðBlock k Þ; where k ¼ 1; 2; 3; ; No : of blocks (9) Block k (encodedimg) denotes the digits in each block of the encoded image. A salt value is added to the XORed value V K in each block to form a new value (x k ) as shown in equation (10). x k ¼ V k þ salt value (10) This value is then passed to a one-way function f(x) = x 2 + x. The resulting value from each block is concatenated, which denotes the encrypted image password as shown in equation (11). EncðBlock k Þ ¼ x 2 k þ x k (11) Encryptedpassimage ¼ EncðBlock 1 Þ&EncðBlock 2 Þ (12) & &Enc ðblock No: of Blocks Þ Figure 5 shows the pseudocode for IEPE algorithm. 6. CASE STUDY This section explains the process of encryption of an image password. Consider the image password chosen by the user as shown in Figure 6. IEPE algorithm is invoked to encrypt the image password. The edges of the image are obtained as described in Section 4 and passed to IEPE algorithm. The edge values of the image in Figure 6 are denoted in the form of a binary matrix having values of 0s and 1s as shown in Table I. The binary matrix having a value 1 denotes the edge point. A salt value for the user is passed to IEPE algorithm. According to IEPE algorithm, each edge point in the image is encoded by concatenating its corresponding row position and column position. It is seen from Table I that the binary matrix has a value 1 in the sixth row and sixth column. The encoded value of this edge point is 66. Similarly, the entire matrix is traversed, and the edge points in each row are encoded. The number of rows and columns in the image is 21 and 15, respectively. The encoded image is obtained by concatenating the encoded values in each row along with the number of rows and columns of the image. Thus, the resulting encoded image of the image password is The values in the encoded image are divided into blocks based on the grouping value. The grouping value is chosen as 4. The number of blocks is found by dividing the length of the encoded image by the grouping value. In this case, the length of the encoded image is 30 and the number of blocks for the encoded image is 8. The encryption key is found as follows. The minimum value among the row encoded values of the image password is found. Here, the minimum encoded value is found to be 66. Two random numbers 3 and 2 are chosen and added to the number of rows and columns of the image, respectively. Thus, the newly generated row and column numbers are 24 and 17, respectively. The newly generated row number and column numbers are concatenated to the minimum encoded value forming the encryption key for the first block. To generate the encryption Security Comm. Networks 2016; 9: John Wiley & Sons, Ltd. 5737

6 An image edge based approach for image password encryption N. K. Sreelaja and N. K. Sreeja Figure 5. Pseudocode for image edge password encryption (IEPE) algorithm. Table I. Encoded edge point values in an image password. Edge point values of the image password Encoded value Figure 6. Image password. key for the successive blocks, the newly generated row number and column number of the previous block are added to the chosen random numbers and concatenated with the minimum encoded value as shown in Table II. An XOR operation is performed between the digits in each block of the encoded image and the encryption key of the corresponding block. The salt value of the user is added to the XORed value in each block. The resulting value X K in each block is passed to a one-way function f(x) = x 2 + x. The resulting f(x) value in each block is concatenated to form the encrypted image password value. Table II shows the encryption of the image password shown in Figure Security Comm. Networks 2016; 9: John Wiley & Sons, Ltd.

7 N. K. Sreelaja and N. K. Sreeja An image edge based approach for image password encryption Table II. Image password encryption. Encoded image (A) Encryption key (B) A XOR B (VK) Salt value (D) XK = VK + D XK2 + X K Encrypted image password COMPUTATIONAL COMPLEXITY The binary matrix denoting the edge points of the image password is traversed to encode the edge points in an image. Consider an image password of size m*n. Thus, the total number of comparisons is O(mn). The encryption key is found by choosing the minimum row encoded value. Thus, the number of comparisons to be made is O(m). The XORed value of the encoded image and the encryption key, which is added to a salt value, is passed to a oneway function to obtain an encrypted image password. The one-way function involves a multiplication and an addition operation for one block. Thus, the computational complexity of the system is given as O(mn) +O(m) + (2 * No: of blocks). 8. EXPERIMENTAL RESULTS The experiment has been conducted for encrypting several image passwords of various sizes. Figure 7 shows the different image passwords and the encrypted image passwords using IEPE algorithm for different grouping values. It is shown that even a slight variation in images leads to different encrypted image passwords. It is seen from Figure 7 that Image1 and Image2 differ by a single bit leading to different encrypted image password. Figure 8 shows the time taken for image password encryption. It is seen from Figure 8 that the time taken to encrypt bigger image passwords is meager. 9. COMPARISON WITH PASSWORD ENCRYPTION TECHNIQUES IEPE algorithm is compared with existing cryptographic hash algorithms as well as picture password techniques Comparison with cryptographic hash algorithms resistant to attacks A comparison of IEPE algorithm is compared with SHA-3 and SHA-2 algorithms Comparison of IEPE algorithm versus SHA-3 algorithm According to SHA-3 algorithm [12], it uses the sponge construction in which the message blocks are XORed into the initial bits of the state, which is then invertibly permuted. In the version used in SHA-3, the state array consists of a maximum of 5 5 array of 64-bit words, 1600 bits in total. This algorithm is resistant to cryptanalytic attacks. In IEPE algorithm, four values, namely, a grouping value, two random numbers to increment the number of rows and columns, and the salt value are stored. Thus, the maximum size of the array to store the values is 4 * 1. Assuming the maximum size of the value stored in each Security Comm. Networks 2016; 9: John Wiley & Sons, Ltd. 5739

8 An image edge based approach for image password encryption N. K. Sreelaja and N. K. Sreeja Figure 7. Encrypted image passwords. element of the array is 15 digits, the maximum number of bits to be stored is 480, which is comparatively less when compared with the internal storage in SHA-3 algorithm. Also, IEPE algorithm is resistant to cryptanalytic attacks Comparison of IEPE algorithm versus SHA-2 algorithm According to SHA-2 algorithm [13], the state array consists of a maximum of 8 1 array of 64-bit words, 512 bits in total. This algorithm is resistant to cryptanalytic attacks. In IEPE algorithm, four values, namely, a grouping value, two random numbers to increment the number of rows and columns and the salt value are stored. Thus, the maximum size of the array to store the values is 4 * 1. Assuming the maximum size of the value stored in each element of the array is 15 digits, the maximum number of bits to be stored is 480, which is comparatively less when compared with the internal storage in SHA-2 algorithm. Also, IEPE algorithm is resistant to cryptanalytic attacks Comparison with cryptographic hash algorithms susceptible to attacks A comparison of IEPE algorithm is compared with SHA-0 and SHA-1 algorithms Security Comm. Networks 2016; 9: John Wiley & Sons, Ltd.

9 N. K. Sreelaja and N. K. Sreeja An image edge based approach for image password encryption the image cannot be guessed or broken by brute force attack as shown in Section Figure 8. Time taken to encrypt image password Comparison of IEPE algorithm versus SHA-0 and SHA-1 algorithms According to SHA-0 [14] and SHA-1 [15] cryptographic algorithms, the internal state size is 160 and the block size is 512. However, both cryptographic algorithms are susceptible to collision attacks [5]. In IEPE algorithm, the maximum number of bits to be stored is 480, which is comparatively less when compared with the internal storage in SHA-0 and SHA-1 algorithms. Also, IEPE algorithm is resistant to cryptanalytic attacks as discussed in Section Comparison of IEPE algorithm versus picture password techniques The IEPE algorithm is compared with the picture password techniques Comparison of IEPE algorithm versus Pict-O-Lock method Hong et al. [8] proposed a scheme called Pict-O-Lock for the purpose of picture memorability. Hong et al. [8] allowed users to choose their own words to associate with each pass-object variant. However, this significantly extends the process of password registration. To arrange the pictures systematically, Hong et al. [8] used a grid-based picture arrangement and each time the login process began, the images displayed on the screen are generated randomly by the program. To protect against brute force attacks, Hong et al. [8] used many decoy images in their scheme. To prevent shoulder surfing attacks, this scheme requires several verification processes. Apparently, this process is significantly time-consuming and tedious, therefore might not be a choice for users. In Pict-O-Lock graphical password method, which is a recognition technique, the password cannot be guessed but can be broken by brute force method. Also, the users have to memorize both picture objects and their codes and are more difficult than text-based password. Unlike Pict-O-Lock method, IEPE algorithm is not time-consuming because the image password is uploaded thereby reducing the need for the graphical database storage on the server side and the traffic loads without transferring the images through network. Also, in IEPE technique, Comparison of IEPE algorithm versus Draw-A-Secret method Jermyn et al. [9] proposed a scheme, known as DAS. This scheme is based on a two-dimensional grid where the users have to draw something to represent their password. Each of the grid coordinates from the drawn pictures is stored in the order of the drawing. To be authenticated, the user needs to redraw the picture again. If the drawing lines are at the same grid coordinates with the proper sequence, then the user is authenticated. There are some advantages when using a grid as the background for the drawing. First, the users can draw a password as long as they wish. Second, grid-based techniques also lessen the need for the graphical database storage on the server side and reduce the traffic loads without transferring the images through network. Furthermore, the full password space for a grid-based scheme is much better than traditional textual passwords. However, in DAS graphical password method, which is a recall technique, the password cannot be broken by brute force method but can be guessed. In IEPE algorithm, the image password is uploaded during the login process and cannot be guessed as in the case of DAS graphical password method. Also, this method reduces the need for graphical-based storage on the server side and reduces the traffic loads without transferring the images through network Comparison of IEPE algorithm versus graphical password technique The graphical password schemes existing in literature include recognition-based and recall-based techniques. In recognition-based techniques, the picture passwords are displayed while logging in and the user has to select the pictures in a sequence. In recall-based techniques, a single image is presented to the user while logging into the system and the user has to click on particular positions in the image, which would be chosen as the password. Users may find this process long and tedious. Because most users are not familiar with the graphical passwords, they often find graphical passwords less convenient than text-based passwords [10]. The major design issue for recall-based methods is the reliability and accuracy of user input recognition. In this type of method, the error tolerances have to be set carefully; overly high tolerances may lead to many false positives, while overly low tolerances may lead to many false negatives. In addition, the more error tolerant the program, the more vulnerable it is to attacks [10]. There are several graphical password techniques available in literature such as Triagle, Movable Frame, Intersection, Blonder [2], VisKeySFR [3], Passlogix [16 18], and Passpoints [19]. However, it is shown that all these schemes are vulnerable to brute force, guessing, and shoulder surfing attack [3]. Sobrado and Birget [4] developed a graphical password technique that deals with the shoulder surfing problem. In Security Comm. Networks 2016; 9: John Wiley & Sons, Ltd. 5741

10 An image edge based approach for image password encryption N. K. Sreelaja and N. K. Sreeja the first scheme, the system will display a number of pass objects (preselected by the user) among many other objects. To be authenticated, a user needs to recognize pass objects and click inside the convex hull formed by all the pass objects. In order to make the password hard to guess, Sobrado and Birget [4] suggested using 1000 objects, which make the display very crowded and the objects almost indistinguishable, but using fewer objects may lead to a smaller password space, because the resulting convex hull can be large. In their second algorithm, a user moves a frame (and the objects within it) until the pass object on the frame lines up with the other two pass objects. The authors also suggest repeating the process a few more times to minimize the likelihood of logging in by randomly clicking or rotating. The main drawback of these algorithms is that the logging in process can be slow. Also, it can be hard to remember when large numbers of objects are involved. The possible password attack methods in this approach are brute force search and guess. In IEPE algorithm, the image password is uploaded during the login process and the password cannot be guessed or broken by brute force method. The scheme is not vulnerable to shoulder surfing attack because the pictures are not chosen. Also, this method reduces the need for storing images on the server side and reduces the traffic loads without transferring the images through network Comparison of IEPE algorithm versus Passface technique Passface [10] is a technique developed by Real User Corporation [10]. According to this approach, the users recognize and pick the preregistered pictures. This method of authentication takes longer than text-based password. The possible password attacks include the dictionary attack, brute force search, password guess, and shoulder surfing. In IEPE algorithm, the image password can neither be guessed because it is an image nor can be broken by brute force method. Also, the scheme is not vulnerable to shoulder surfing and dictionary attack because the pictures are not chosen Comparison of IEPE algorithm versus signature technique Syukri et al. [11] have proposed a system where authentication is conducted by having the user draw their signature using a mouse, and this method requires a reliable signature recognition program. The possible password attacks for this method include the dictionary attack, password guess, and shoulder surfing. In IEPE algorithm, the image password cannot be guessed because it is an image. Also, the scheme is not vulnerable to shoulder surfing and dictionary attack because the pictures are not chosen Comparison of IEPE algorithm versus authentication schemes using picture, object, and pseudoword recognition Weinshall and Kirkpatrick [20] sketched several authentication schemes, such as picture recognition, object recognition, and pseudoword recognition and conducted a number of user studies. In the picture recognition study, a user is trained to recognize a large set of images ( images) selected from a database of images. After 1 to 3 months, users in their study were able to recognize over 90% of the images in the training set. This study showed that pictures are the most effective among the three schemes tested. Pseudocodes can also be used, but require proper setting and training. IEPE algorithm reduces the need for graphical-based storage on the server side and reduces the traffic loads without transferring the images through network. 10. SECURITY ANALYSIS The IEPE algorithm is studied to show that it is resistant to various password cryptanalytic attacks Pre-image resistance Pre-image resistance means from all pre-specified outputs, it is computationally infeasible to find any input that hashes to that output. That is, to find any pre-image x 0 such that h(x 0 )=y when given any y for which a corresponding input is not known. Consider the binary matrix denoting the edge values of an image password shown in Figure 9. The IEPE algorithm is invoked to encrypt the image password. The grouping value is chosen as 3. Table III shows the encryption of the image password. Figure 9. Edge values of an image password. Table III. Encryption of the image password. Grouping value 3 Random numbers 3, 2 Encoded image (A) Encryption key (B) A XOR B (V K ) Salt Value (D) X K = V K + D Encrypted image password (y) Security Comm. Networks 2016; 9: John Wiley & Sons, Ltd.

11 N. K. Sreelaja and N. K. Sreeja An image edge based approach for image password encryption To show that IEPE algorithm is resistant to pre-image attack, we assume that an attempt is made to find any pre-image x 0 such that h(x 0 )=y. In this case, y denotes the encrypted image password. Hence, y = In order to construct an image x 0 from y, the hacker has to group the values in the encrypted image password for each block. If the encrypted image password value for each block is found, the value of X K must be found. Because a one-way function is used to obtain the encrypted image password value, finding the inverse of the function is impossible. Hence, it is difficult for the hacker to find the value of X K. To find the encryption key, the minimum encoded row value from the image password has to be found. Also, the random number chosen to add to the number of rows and columns to generate the encryption key has to be found. Hence, it is shown that it is not possible to construct an image from the encrypted image password y for which the input image is not known. Thus, it is shown that pre-image resistance attack is not possible because it is difficult to construct the exact image as that of input password image from the encrypted password. Also, it is explained in Section 10.2 that even a minor change in the image will result in a different encrypted password Second pre-image resistance According to the second pre-image resistance, given an input m 1, it should be difficult to find another input m 2 such that m 1 m 2 and hash(m 1 ) = hash(m 2 ). This property is sometimes referred to as weak collision resistance, and functions that lack this property are vulnerable to second pre-image attacks. Consider the binary matrix denoting the edge points of an input image password m1 as shown in Figure 10. The encrypted image password is as shown in Table IV. It is shown that IEPE is not susceptible to second preimage resistance attack by showing that it is not possible to construct two different images having the same encrypted password value because the number of rows and columns are concatenated at the end of the encoding image. Also, the number of rows and columns are concatenated to the end of the minimum row encoded value denoting the encryption key. Consider the edge point values of an image password shown in Figure 11. Table IV. Encryption of the image password m1. Grouping value 3 Random numbers 3, 2 Encoded image (A) Encryption key (B) A XOR B (V K ) Salt value (D) X K = V K + D Encrypted image password Figure 11. Edge point of an image password m2. It is seen from Figures 10 and 11 that the edge values in both the images are in the same position. However, the encrypted password value for the images shown in Figures 10 and 11 is and , respectively (Tables IV and V). Because the number of rows and columns are concatenated at the end of the encoded image, the encoded value varies for every image and it is not possible for two different images to have the same encrypted password value. Thus, it is shown that the IEPE algorithm is resistant to second pre-image attack Collision resistance According to this method, it should be difficult to find two different messages m 1 and m 2 such that hash(m 1 ) = hash(m 2 ). Such a pair is called a cryptographic hash collision. This property is sometimes referred to as strong collision resistance. This is applicable to IEPE algorithm. It has been shown in Section 10.2 that it is not possible to find two different image passwords having the same encrypted value. Figure 12 shows an image password Table V. Encryption of the image password m2. Grouping value 3 Figure 10. Edge point of an image password m1. Random numbers 2, 3 Encoded image (A) Encryption key (B) A XOR B (V K ) Salt value (D) X K = V K + D Encrypted image password Security Comm. Networks 2016; 9: John Wiley & Sons, Ltd. 5743

12 An image edge based approach for image password encryption N. K. Sreelaja and N. K. Sreeja Table VII. Encryption of the image password m4. Grouping value 3 Random numbers 2, 3 Encoded image (A) Encryption key (B) A XOR B (V K ) Salt value (D) X K = V K + D Encrypted image password Dictionary attacks Figure 12. Edge point of an image password m3. whose edge points are similar to image passwords in Figures 10 and 11. Because the number of rows and number of columns are different, the encrypted image value for the image in Figure 12 and the images in Figures 10 and 11 is different (Tables IV VI). The image passwords shown in Figures 13 and 11 have the same encryption key. However, it is seen that both the image passwords are different and does not have the same encrypted value, as shown in Tables VII and V respectively. The image passwords are stored in the password table in an encrypted format using IEPE algorithm. It is shown in Section 10.1 that it is impossible for the attacker to find the image from the encrypted password. Also, it is not possible to use brute force attacks because a slight variation in an image gives an entirely different password as shown in Figure 7. Table VI. Encryption of the image password m3. Grouping value 3 Random numbers 4, 3 Encoded image (A) Encryption key (B) A XOR B (V K ) Salt value (D) X K = V K + D Encrypted image password Dictionary attacks can be carried out only in alphanumeric passwords. Because IEPE algorithm uses image as a password and not alphanumeric characters, it is not susceptible to dictionary attack Shoulder surfing According to IEPE algorithm, shoulder surfing attack method cannot be used because the attacker cannot spy the user s movement to obtain the password because the user does not press the keys in the keyboard. Also, he cannot guess the password by listening to the number of keys pressed because the user enters the image password by uploading the image Key loggers The key loggers are software programs installed in the system where the keys pressed by the user are stored in a log file and sent to the attacker. A keylogger is something that records keystrokes made on a computer. It captures every key pressed on the keyboard and stores it down in a file or memory bank that can be viewed by the person performing the monitoring in real time, or at a later date. Some of the applications have a graphical keyboard where the users can enter the characters they want by clicking the mouse on it. This is especially useful for numeric personal identification numbers. However, it is again vulnerable to some of the key loggers that take screenshots as and when the data are entered. This risk can be mitigated with the use of a multifactor verification device such as entering passwords using a smart card. Because IEPE uses an image password, the key logger attack fails in this system because the text is not typed in the system. 11. CONCLUSIONS Figure 13. Edge point of an image password m4. An IEPE algorithm is proposed to encrypt the image passwords. This method increases the usability of the password because the user does not find it difficult to remember image passwords. It is shown that the encrypted image password requires less storage because it is in a text 5744 Security Comm. Networks 2016; 9: John Wiley & Sons, Ltd.

13 N. K. Sreelaja and N. K. Sreeja An image edge based approach for image password encryption format. It is shown to be better when compared with picture password techniques. It is also shown that the internal storage is less when compared with the hash algorithms for password encryption. Also, it is shown that it is resistant to password cryptanalytic attacks such as pre-image, second pre-image, and collision-resistant attacks. It is also resistant to brute force and dictionary attacks. It is also proved that the encrypted image password varies a lot even for a very slight variation in the image. REFERENCES 1. Adams A, Sasse MA. Users are not the enemy: why users compromise computer security mechanisms and how to take remedial measures. Communications of the ACM 1999; 42: Blonder G, Graphical password, In Lucent Technologies, Inc., Murray Hill, NJ, United States Patent , SFR IT Engineering, EN/pocketpc/viskey/, Accessed on January Sobrado, L, Birget, J. Graphical passwords, The Rutgers Scholar, An Electronic Bulletin of Undergraduate Research, Rutgers University, New Jersey, Vol. 4, (2002), sobrbirg/sobrbirg.htm 5. Stevens, M. Attacks on hash functions and applications, Ph.D Thesis, June Wiedenbeck S, Waters J, Birget JC, Brodskiy A, Memon N. PassPoints: design and longitudinal evaluation of a graphical password system. International Journal of Human-Computer Studies 2005; 63: Chowdhury, S, Poet, R, Mackenzie, L (2013) A comprehensive study of the usability of multiple graphical passwords. In:Kotzé, Paula, Marsden, Gary, Lindgaard, Gitte, Wesson, Janet and Winckler, Marco (eds.) Human-Computer Interaction INTERACT 2013: 14th IFIP TC 13 International Conference, Cape Town, South Africa, September 2 6, 2013, Proceedings, Part III. Series: Lecture Notes in Computer Science (8119). Springer, pp ISBN Hong D, Man S, Hawes B, Mathews M, A password scheme strongly resistant to spyware, In Proceedings of International conference on security and management, Las Vergas, NV, Jermyn I, Mayer A, Monrose F. Reiter MK, Rubin AD, The design and analysis of graphical passwords, In Proceedings of the 8th USENIX Security Symposium, Real User Corporation, Passfaces TM, http//:www. realuser.com, Accessed on January Syukri AF, Okamoto E, Mambo M. A user identification system using signature written with mouse. In Third Australasian Conference on Information Security and Privacy (ACISP). Springer-Verlag Lecture Notes in Computer Science (1438): London, UK, 1998; Keccak implementation overview version keccak.noekeon.org/keccak-implementation-3.2.pdf- SHA-3, NIST, Descriptions of SHA-256, SHA-384, and SHA- 512, 2001 MAY, < sha pdf>. 14. SHA-0, National Security Agency, United States, SHA-1, National Security Agency, United States, Paulson LD. Taking a graphical approach to the password. Computer 2002; 35: Passlogix, Accessed on February a96582/overview.htm# Suo X, Zhu Y, Owen GS, Graphical passwords: a survey, Department of Computer Science, Georgia State University, Weinshall D, Kirkpatrick S. Passwords you ll never forget, but can t recall. In Proceedings of Conference on Human Factors in Computing Systems(CHI). ACM: Vienna, Austria, 2004; Security Comm. Networks 2016; 9: John Wiley & Sons, Ltd. 5745