Research Article Open Access
Secure Electronic Voting Application Based on Face Recognition and Ciphering
Zinah J Mohammed Ameen*
Computer Engineering Department, University of Technology, Baghdad, Iraq
*Corresponding author: Zinah J Mohammed Ameen, Computer Engineering Department, University of Technology, Baghdad, Iraq, E-mail: @
Received: July 26, 2018; Accepted: August 02, 2018; Published: August 10, 2018
Citation: Zinah JMA (2018) Secure Electronic Voting Application Based on Face Recognition and Ciphering. J Comp Sci Appl Inform Technol. 3(2): 1-11. DOI: 10.15226/2474-9257/3/2/00131
Abstract
Voting is the fundamental right for every nation. An Electronic Voting (E-Voting) system is a voting system in which the election process is notated, saved, stored and processed digitally, that makes the voting management task better than traditional paper based method. However, security remains the bottleneck of each system. Therefore, the need of designing a secure e-voting system is very important. This paper introduced a secure e-voting web application. Depending on secure voter authentication using biometrics besides steganography and cryptosystem. A local neighborhood intensity pattern of feature descriptor method is used to identify the allowable and authorized voters. RSA public key algorithm is proposed to encrypt a voter choice, then based on a secret key dedicated to each voter with the input voter instant image the encrypted data with RSA algorithm is encrypted again before casting to the server using spread spectrum steganography technique.

Keywords: LNIP recognition face; Web application; SSIS steganography;
Introduction
Voting theory began formally in 18th century and many proposals for voting systems have been made over since. There have been several studies on using electronic technologies to improve the process of voting. When designing an e-voting system, it is essential to consider ways in which the voting tasks are performed electronically without sacrificing voter privacy or introducing opportunities for fraud [1]. Integrity of the voting process is an important issue in the integrity of the democracy itself. Therefore, the whole system of election must be secure and robust against a variety of fraudulent behaviors [2]. There is no measurement for agreeable security standard, because the standard depends on type of the information. An agreeable security level is always an adjustment between usability and strength of the security method [3]. In the recent years, automated method of recognizing a person is going to be more popular by using biometric system. A biometric has shown an increasing interest in many fields such as surveillances, human computer interface and security identification. Biometric recognition is an automated method of recognizing an individual by means of comparing the feature vector derived from the behavioral and the physiological distinctiveness such as finger print, iris, face recognition etc. [4]. Face detection and recognition is becoming increasingly important. There are multiple cues available and can be used as features. Facial features are extracted and compared using support vector machine classification algorithm [5]. Paper layout is organized as follow section 2 briefly discusses the previous similar work, section summarizes the secure e-voting system, section 3 and 5 explains the proposed system design and implementation. Finally conclusion is discussed in section 5.
Related work
Firas Hazzaa, et al. [1] proposed an e-voting system based on fingerprint. This system is implemented using C#. Neha Gandhi introduced an online voting system based on fingerprint then encrypted the result using SHA 256 with MD5 [2]. Patil RH, et al. [6] proposed a secure e-voting system using face recognition and dactyl gram method. A. Geetha, et al. [7] implemented a face recognition algorithm based on local derivate tetra pattern. Matma Jain, et al. [8] introduced an adaptive circular queue image steganography with RSA cryptosystem.
Secure E-voting System
E-voting application is a web based public voting system deployed to make election process easy to use and speed competent [6]. Objective behind the development of this system is to simplify the process of organizing elections and make it easy for voters to vote remotely from their home computers or smart devices while taking into consideration better security, anonymity and providing auditioning capabilities [3]. Problems like prohibited unauthorized election, incorrect use of cryptography, vulnerabilities to network threat. E-voting system is more susceptible to attacks such that voting results can be easily manipulated, hence may result in fraud. To avoid that this paper proposed the use of biometric face recognition in order to authenticate the eligible voters then voters’ choice would be encrypted twice first using cryptography methodology then with steganography technique.
Facial Recognition
Biometrics is becoming a primary component of a personal identification solution, since biometric identifiers cannot be participated or misplaced, and they represent any individual’s identity. Biometric recognition refers to the use of iris, fingerprint, face palm and speech characteristics called biometric identifiers [1]. In recent years, face recognition plays an important role in intensive research. With the current discerned world security situation, governments as well as private sector require reliable methods to accurately identify individuals without overlay contravene on rights to privacy or requiring significant compliance on the part of the individual being recognized. A number of techniques have been applied to face recognition and they can be divided into two categories [7]:

• Geometric feature matching
• Template matching

The performance of face recognition techniques have increased severely. However, the robustness of face recognition needs enhancement. The results of most techniques are effected by circumferential changes, such as illumination deviations, expression disparity and pose variations [9].

Local Binary Pattern (LBP) is defined as a grayscale invariant texture measure; it is a useful tool to model texture images. LBP has shown excellent performance in many comparative studies, in terms of both speed and discrimination performance [10]. In this method, a small window of an image is considered and intensity difference between the center pixel and its N neighbors [11]. The original LBP manipulator labels the pixels of an image by thresholding the 3 × 3 neighborhood of each pixel with the value of the central pixel and concatenating the results to form a number. As illustrating in figure 1, obtaining an LBP micro pattern when the threshold is set to zero [10].
Figure 1: LBP Basic Operator
LBP local pattern is calculated by comparing between the centers symmetric pixels with neighboring pixels without taking into consideration other relative intensity difference between each pixel with other adjacent pixels other than the center one. For these reason, an extension of an LBP is chosen for this paper, Local Neighborhood Intensity Pattern (LNIP) that considers two neighboring pixels for binary pattern calculation. LNIP considers the radius of unit distance since closest neighboring pixels carries more discriminating information for texture descriptors. Thus by using a 3×3 window to calculate the binary pattern by exploring the mutual information with respect to the adjacent neighbors. Figure 2 defines the meaning of adjacent neighbors in a 3×3 window.
Figure 2: Explains the adjacent neighbor relation for each of the 8 neighbors (Zi i=1,2,…8) of ZC , Zi has 4 adjacent neighbors if i is even otherwise, Zi has only 2 adjacent neighbors for odd values of i.
For signed LNIPS, the sign of relative difference between one of the 8 neighbors of the center pixel (ZC) and its corresponding adjacent neighbors Zi. An N bit pattern with respect to Zi where N is the number of element in Si. A binary pattern A1,i corresponding to Zi using equation (1). Similarly, another binary pattern A2,i of same size considering the same neighboring as in Si using equation (2).
A 1,i ( j ) = Sign( S i ( j ),  Z i ) where j = 1 to N .. Eq ( 1 ) MathType@MTEF@5@5@+= feaagGart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGbbWdamaaBaaaleaapeGaaGymaiaacYcacaWGPbaapaqabaGc daqadaqaa8qacaWGQbaapaGaayjkaiaawMcaa8qacaqGGaGaaeypai aabccacaWGtbGaamyAaiaadEgacaWGUbWdamaabmaabaWdbiaadofa paWaaSbaaSqaa8qacaWGPbaapaqabaGcdaqadaqaa8qacaWGQbaapa GaayjkaiaawMcaa8qacaGGSaGaaeiiaiaadQfapaWaaSbaaSqaa8qa caWGPbaapaqabaaakiaawIcacaGLPaaacaqGGaWdbiaadEhacaWGOb GaamyzaiaadkhacaWGLbGaaeiiaiaadQgacaqGGaGaeyypa0Jaaeii aiaaigdacaqGGaGaamiDaiaad+gacaqGGaGaamOtaiaabccacqGHMa cVcqGHMacVcqGHMacVcqGHMacVcaGGUaGaaiOlaiaabccacaWGfbGa amyCaiaadwhacaWGHbGaamiDaiaadMgacaWGVbGaamOBaiaabccapa WaaeWaaeaapeGaaGymaaWdaiaawIcacaGLPaaaaaa@6D52@ A 2,i ( j ) = Sign( S i ( j ),  Z C ) where j = 1 to N .. Eq ( 2 ) MathType@MTEF@5@5@+= feaagGart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGbbWdamaaBaaaleaapeGaaGOmaiaacYcacaWGPbaapaqabaGc daqadaqaa8qacaWGQbaapaGaayjkaiaawMcaaiaabccacaqG9aWdbi aabccacaWGtbGaamyAaiaadEgacaWGUbWdamaabmaabaWdbiaadofa paWaaSbaaSqaa8qacaWGPbaapaqabaGcdaqadaqaa8qacaWGQbaapa GaayjkaiaawMcaa8qacaGGSaGaaeiiaiaadQfapaWaaSbaaSqaa8qa caWGdbaapaqabaaakiaawIcacaGLPaaacaqGGaWdbiaadEhacaWGOb GaamyzaiaadkhacaWGLbGaaeiiaiaadQgacaqGGaGaeyypa0Jaaeii aiaaigdacaqGGaGaamiDaiaad+gacaqGGaGaamOtaiaabccacqGHMa cVcqGHMacVcqGHMacVcqGHMacVcaGGUaGaaiOlaiaabccacaWGfbGa amyCaiaadwhacaWGHbGaamiDaiaadMgacaWGVbGaamOBaiaabccapa WaaeWaaeaapeGaaGOmaaWdaiaawIcacaGLPaaaaaa@6D2E@
The single bit with respect to Zi is evaluated simply by comparing the change in the structure of these two binary patterns A1,i and A2,i. The structural change in the bit pattern is calculated by taking bitwise XOR operation between these two patterns. In order to calculate the information of total deviation of these neighbors from a particular pixel (Zi) LNIPM is used.
M i = 1 M k=1 M |  S i (k) Z i | ..Eq ( 3 ) MathType@MTEF@5@5@+= feaagGart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGnbWdamaaBaaaleaapeGaamyAaaWdaeqaaOWdbiabg2da9maa laaapaqaa8qacaaIXaaapaqaa8qacaWGnbaaamaawahabeWcpaqaa8 qacaWGRbGaeyypa0JaaGymaaWdaeaapeGaamytaaqdpaqaa8qacqGH ris5aaGccaGG8bGaaiiOamaavababeWcbaGaamyAaaqab0qaaiaado faaaGccaGGOaGaam4AaiaacMcacaGGtaIaamOwa8aadaWgaaWcbaWd biaadMgaa8aabeaakiaacYhacaqGGaWdbiabgAci8kabgAci8kabgA ci8kaac6cacaGGUaGaamyraiaadghacaWG1bGaamyyaiaadshacaWG PbGaam4Baiaad6gacaqGGaWdamaabmaabaWdbiaaiodaa8aacaGLOa Gaayzkaaaaaa@5D0F@ T C = 1 8 i=1 8 | Z i   Z c |  Eq ( 4 ) MathType@MTEF@5@5@+= feaagGart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGubWdamaaBaaaleaapeGaam4qaaWdaeqaaOWdbiabg2da9maa laaapaqaa8qacaaIXaaapaqaa8qacaaI4aaaamaawahabeWcpaqaa8 qacaWGPbGaeyypa0JaaGymaaWdaeaapeGaaGioaaqdpaqaa8qacqGH ris5aaGccaGG8bGaamOwa8aadaWgaaWcbaWdbiaadMgaa8aabeaak8 qacaGGtaIaaeiiaiaadQfapaWaaSbaaSqaa8qacaWGJbaapaqabaGc caGG8bGaaeiia8qacqGHMacVcqGHMacVcqGHMacVcqGHMacVcaqGGa GaamyraiaadghacaWG1bGaamyyaiaadshacaWGPbGaam4Baiaad6ga caqGGaWdamaabmaabaWdbiaaisdaa8aacaGLOaGaayzkaaaaaa@5AF7@
Mi is the mean deviation about the ith neighbor of Zc from its corresponding Si as indicated in equation (3) calculated for all the 8 neighboring pixels of Zc in a 3 × 3 window. Tc is the threshold value calculated by taking the mean deviation of the neighbors Zi about the center Zc as shown in equation (4). The final magnitude pattern value corresponding to Zc is calculated using equation (5).
LNI P M ( Z c )= i=1 8 2 i1  ×Sign( Z i , T c ) .Eq ( 5 ) MathType@MTEF@5@5@+= feaagGart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGmbGaamOtaiaadMeacaWGqbWdamaaBaaaleaapeGaamytaaWd aeqaaOWaaeWaaeaapeGaamOwa8aadaWgaaWcbaWdbiaadogaa8aabe aaaOGaayjkaiaawMcaaiabg2da98qadaGfWbqabSWdaeaapeGaamyA aiabg2da9iaaigdaa8aabaWdbiaaiIdaa0WdaeaapeGaeyyeIuoaaO GaaGOma8aadaahaaWcbeqaa8qacaWGPbGaeyOeI0IaaGymaaaakiaa cckacqGHxdaTcaWGtbGaamyAaiaadEgacaWGUbGaaiikaiaacQfapa WaaSbaaSqaa8qacaWGPbaapaqabaGcpeGaaiilaiaabsfadaWgaaWc baGaam4yaaqabaGccaGGPaGaaeiiaiabgAci8kabgAci8kabgAci8k abgAci8kaac6cacaWGfbGaamyCaiaadwhacaWGHbGaamiDaiaadMga caWGVbGaamOBaiaabccapaWaaeWaaeaapeGaaGynaaWdaiaawIcaca GLPaaaaaa@6868@
This enhancement step was necessary by making the patterns more resistant to illumination changes and the mutual relationship among the adjacent neighbors is well explored in this pattern. Besides that, not only the sign of the intensity variation between the central pixel and one of its neighbors Zi was taken into account, but also the sign of variation values between Zi and its adjacent neighbors along with the central pixels. Most of the local patterns including premier LBP concentrate mainly on the sign information ignoring the magnitude. The sign information is more important for any binary pattern it cannot be removed be as it plays a contributory role to supply complementary information texture descriptor. Therefore, both sign pattern LNIPS and magnitude pattern LNIPM poses complementary information that gives superior performance as compared with the original LBP [11].
Steganography and Cryptography
With the strong development of computing, large amounts of media are continuously being downloaded and streamed across the Internet. The variety of these media leads to difficulties in analyzing normal and abnormal content. Also, as most processes in the Internet are driven by humans, estimating behavior and analyzing abnormality is a complicated process that may require high computing power and sophisticated algorithms [12]. The main purpose of Steganography, which means ‘writing in hiding’ is to hide data in a cover media so that others will not be able to notice it. Mainly it is of three categories, (i) steganography in image, (ii) steganography in audio, and (iii) steganography in video [2]. In image steganography the confidential message is concealed inside an image in such a way that the change in quality of the image cannot be noticed. Image steganography techniques can be classified into two major classes: spatial domain techniques and frequency domain techniques. In spatial domain techniques the confidential message is concealed inside the image by applying some manipulation over the different pixels of the image. In frequency domain techniques the image is converted to another form by applying a transformation like discrete wavelet transform and then the message is concealed by applying any of the ordinary embedding techniques. The image in which a confidential message concealed is called the stego-image. There are different categories of spatial domain, (i) LSB steganography, (ii) RGB based steganography, (iii) pixel value differencing steganography, (iv) mapping based steganography, (v) palette based steganography, (vi) collage based steganography, (vii) spread spectrum steganography, (viii) code based steganography, and (ix) others [13].

A steganography system is a quintuple ℘ =(C, M, K, DK, EK), where 𝐶 is the set of all coverings used in communication, 𝑀 is the set of all confidential messages that need to be transported using the coverings, 𝐾 the set of confidential keys, 𝐸𝐾: 𝐶 × 𝑀 × 𝐾 → 𝐶, and 𝐷𝐾: 𝐶 × 𝐾 → 𝑀 two functions, the embedding and the extraction functions respectively such that: 𝐷𝐾(𝐸𝐾(𝒄, 𝒎, 𝒌), 𝒌) = 𝒎 [12].

Spread Spectrum Image Steganography (SSIS) is an example of robust steganography algorithm since it can hold some amount of distortion before message wastage. The fundamental concept of SSIS is the embedding of the concealed information within noise, which is then added to the digital image. This noise is typical of the noise inherent to the image acquisition process and, if kept at low levels, is not perceptible to the human eye nor is susceptive to detection by computer analysis [12].

The encoding algorithm takes as input the steganography key, the input message, the cover image and the output is the stegoimage [13]. The major processes of the stegosystem encoder are sketched in figure 3.

As indicated in figure 3, the voter’ choice is first encrypted with RSA cryptosystem. The steps of RSA ciphering is summarizes as fellow:

i. Choose P and Q randomly, both should be a prime number.
ii. Calculate N = P × Q, ɸ(N) = (P-1)×(Q-1).
iii. Choose an integer E, such that GCD(ɸ(N),E) = 1, 1< E< ɸ(N).
iv. Calculate D, such that (D × E)% ɸ(N)=1
v. Public Key (E,N) and Private Key (D,N), public key is used for encryption while private key is used for decryption.
Figure 3: Stegosystem using SSIS
Proposed System Design
Figure 4 summarizes the use case diagram for the two main personnel who are deal with this system. The main tasks for both of them is shown in this figure, the administrator is responsible for managing the whole voting process even registering people who are allowed to vote. On the other hand, voters have only to enter their voter’s ID number as written in their electoral card besides capturing an instant image for authorization process. Figure 5 and 6 describes the sequence diagrams for both system administrator and voters

The authorization process applied in this system is based on face recognition. Face recognition steps summarize in figure 7.
Figure 4: System Use Case Diagram
Figure 5: Administrator Sequence Diagram
Figure 6: Voter Sequence Diagram
Figure 7: Flow Diagram of Face Detection Methodology
Image retrieval from server database is implemented using Local Neighborhood Intensity Pattern feature (LNIP). Five distance equations are used to calculate image matching for both the input voter’s instant image and the stored images, based on those equations images with highest matching values are put at the beginning in order to decide whether this person could vote or not.

For integrity insurance of the whole process especially voting remotely, this secure e-voting system provides besides secure login, additional security issues represented by steganography and RSA ciphering. Figure (8) presents the ciphering method implemented in this system for safely casting voter’s choice. By encrypting his/her decision first with RSA algorithm then the input image of 512×512 pixels is used as a cover image with voter’s ID number as key. Spread Spectrum Image Steganography (SSIS) approach is applied to get the stego-image, at this step the stego-image is posted to the server.
Proposed System Implementation
For implementing this system PHP version 5.6 is used with SQL server version 5.2 for executing server side scripting side by side with Matlab version 9.1 to implement face image processing algorithm then encrypt the voter’s choice using RSA algorithm and SSIS steganography approach.
Figure 8: Flow Diagram of ciphering voters’ choice Methodology
First the administrator login to set the system ready as indicated in figure 9. Then if the electoral period is started, voters would be able to access the application and make a choice as indicated in figure 10. The eligible voters require an ID number besides an instant image to verify their identity. An instant image is capturing with limited size and clear Facial features as shown in figure 11.
Figure 9: Administrator Login Stepogy
Figure 10: Voter Login
Figure 11: Voter’s instant image
In case if the entered ID number is correct and voter’s image retrieved from server database, a voter ID card information would be presented as indicated in figure 12.

Each eligible voter is allowed to vote only once, otherwise a message indicating that the person with this ID number had already been voted and is unable to vote again as shown in figure 13.

This system guaranteed the right if a voter chooses to repeal his ID number without making a choice as indicated in figure 14. A confirmation message is appeared in case a voter decided not to vote as shown in figure 15.

Otherwise, a voter can make a choice among the appeared electoral lists and their members then casting his/her choice as indicated in figure 16.
Figure 12: Voter ID card Information
Figure 13: Election right is done once only
Figure 14: voting or repealing your voter ID number
Figure 15: Card Revocation Confirmation Message
Figure 16: Election lists with members
Conclusion
This paper discusses the designing and implementing of a secure e-voting web application. Security is provided in two approaches, first a voter’s authorization process is applied based on (LNIP) method for face recognition besides his/her unique voter ID number, otherwise he/she is deprived of the election process. Secondly, and to ensure that voter’s choice is transferred securely without altering, Cryptography and steganography methodologies are used to introduce two levels of security. RSA public key algorithm is applied to encrypt a voter’s choice to provide confidently at the first level then, by using his/her instant image as a cover image side by side with the Voter ID number as a key to from a stego-image that will be sent to the server. If any attack by a hacker or eavesdropping is happened, it would be difficult to him to retrieve the original text especially within the allowed electoral period. This system was experimented on thirty persons; twenty-three were recognized and able to vote. This system could be distributed to cover the whole country with more servers and administrators.
ReferencesTop
  1. Hazzaa F, Kadry S. New System of E-Voting using Fingerprint. International Journal of Emerging Technology and Advanced Engineering. 2012;2(10):355-363.
  2. Neha G. Study on Security of Online Voting System using Biometrics and Steganography. International Journal of Computer Science and Communication. 2014;5(1):29-32.
  3. Alaguvel R, Gananavel G, Jagadhambal K. Biometrics using Electronic Voting System with Embedded Security. International Journal of Advanced Research in Computer Engineering and Technology. 2013;2(3):1065-1072.
  4. Mukesh DR, Bharat SB. Face Recognition using Local Patterns. International Journal on Recent and Innovation Trends in Computing and Communication. 2015;3(10):5884-5889.
  5. Annadate MN, Sunil Gandhi S, Nivita Ravi K, Pushkar Satish N. Online Voting System using Biometric Verification. International Journal of Advanced Research in Computer and Communication Engineering. 2017;6(4):276-281.
  6. Patil Rahul H, Tarte Babita B, Wadekar Sapana S, Zurunge Bahakti S, Phursule Rajesh. A Secure E-Voting System using Face Recognition and Dactylogram. International Engineering Research Journal. 2016;2(2):758-762.
  7. Geetha M, Mohamed S, Vetharaj YJ. Face Recognition based on Local Deivative Tera Pattern. Journal on Image and Video Processing. 2017;7(3):1393-1400.
  8. Mamta J, Saroj Kumar L, Sunil Kumar V. Adaptive Circular queue image Steganography with RSA cryptosystem. Perspectives in Science. 2016;8:417-420.
  9. Sharadamani D, Naga Raju C. Face Recognition using Gradient Derivative Local Binary Pattern. International Journal of Applied Engineering Research. 2017;12(7):1316-1323.
  10. Zhang B, Gao Y, Zhao S, Liu J. Local Derivative Pattern versus Local Binary Pattern: Face Recognition with High-Order Local Pattern Descriptor. IEEE Transactions on Image Processing. 2010;19(2):533-544.
  11. Prithaj B, Ayan Kumar B, Avirup B, Partha Pratim R, Subrahmny M. Local Neighborhood Intensity Pattern- A New texture Feature descriptor for image Retrieval. Expert Systems with Applications. 2018;113.
  12. Dragos D, Ioan-Mihail S, Emil S. Steganography Techniques. Excellence Research Grants Program, University of Politechnica of Bucharest. 2016.
  13. Nurhayati, Ahmad SS. Steganography for inserting message on digital image using least significant bit and AES cryptographic algorithm. IEEE International Conference on Cyber and IT Service Management. 2016.
  14. Padmasri B, Amutha Surabi M. Spread Spectrum Image Steganography with advanced encryption key Implementation. International Journal of Advanced Research in Computer Science and Software Engineering. 2013;3(3):713-720.
 
Listing : ICMJE   

Creative Commons License Open Access by Symbiosis is licensed under a Creative Commons Attribution 3.0 Unported License