Artificial intelligence

Scientific journal

ISSN 2710-1673

ONLINE: ISSN 2710-1681

Select your language


Encryption of graphic information by means of transformation matrixes for protection against decoding by neural algorithms

Yunak O.1, Stryxaluk B.1, Yunak O.3
1 Lviv National Polytechnic University
3 Separated structural unit «College of telecommunications and computer technologies» of National University «Lviv Polytechnic»

Full text (PDF)

UDC: 004.8; 004.7
Publication Language: Ukrainian
Stuc. intelekt. 2020; 25(2):15-20

Abstract: The article deals with the algorithm of encrypting graphic information (images) using transformation matrixes. It presents the actions that can be done with the image. The article also gives algorithms for forming matrixes that are created with the use of random processes. Examples of matrixes and encryption results are shown. Calculations of the analysis of combinations and conclusions to them are carried out. The article shows the possibilities and advantages of this image encryption algorithm. The proposed algorithm will allow to transmit encrypted information through open communication channels. The algorithm uses all possible actions that can be done with the image, namely: image resizing, adding redundant elements to the image, changing the location of the pixels, changing the brightness of the image pixels, inverting the image pixels, changing the colours of the images pixels. A randomized approach to matrix formation will make it impossible to use neural systems. The algorithm of the implementation of writing encryption code is rather simple. It is not possible to decrypt the image without a key as the number of possible combinations is extremely large. This algorithm can be used in the military, intelligence, information security. Encryption does not require significant computing or RAM resources. This algorithm can be implemented in WEB-technology and in mobile applications, it has no recursive functions and loop in loop. The algorithm hides the image size, which in turn adds extra protection. Redundant elements will not allow neural networks to compare pixels. Changing colors, changing brightness, inversion by random processes will not allow neural networks to find the decryption function. The article considers an image encryption algorithm that will give a possibility to generate a key. The key will be a set of two-dimensional arrays formed in a randomized manner. A randomized approach to matrix formation will make it impossible to use neural systems. The algorithm uses all possible actions that can be done with the image, namely: (1) Image resizing. (2) Adding redundant elements to the image. (3) Changing the location of the pixels. (4) Changing the brightness of the image pixels. (5) Inverting the image pixels. (6) Changing the colours of the images pixels. The pixel positioning matrix gives us (Nx • Ny)! combinations. The pixel brightness change matrix gives us 8 * (Nx • Ny)! combinations. The pixel inversion matrix gives us 8 * (Nx • Ny)! combinations. The pixel colour change matrix gives us 6 * (Nx • Ny)! Combinations. The total number of combinations N is equal to: It is not possible to decrypt the image without a key as the number of possible combinations is extremely large. The algorithm of the implementation of writing encryption code is rather simple. This algorithm can be used in the military, intelligence, information security. The advantage of this algorithm is the fact that the encrypted image can be transmitted through open channels.

Keywords: encryption, decryption, graphic information, matrix transformation, security, neural algorithms, ran- dom processes, image, RGB, encryption key

References:

  1. V. Jemecʹ Sučasna kryptohrafija: Osnovni ponjattja / V. Jemecʹ, A. Melʹnyk, R. Popovyč. – Lʹviv: BaK, 2003. – 144 s.: il.
  2. Raškevyč, Ju.M., Peleško, D.D., Kovalʹčuk, A.M., Peleško, M.Z. (2008) Modyfikacija alhorytmu RSA dlja dejakych klasiv zobraženʹ. Technični visti 1(27), 2(28). S. 59 – 62.
  3. Y. Rashkevych, A. Kovalchuk, D. Peleshko, M. Kupchak. (2009) Stream Modification of RSA Algorithm For Image Coding with precize contour extraction. Proceedings of the X-th International Conference CADSM. 24-28 February 2009, Lviv- Polyana, Ukraine, Pp. 469-473.
  4. Krasylenko, V.H., Hrabovljak, S.K. (2012) Matryčni afinno-perestanovočni alhorytmy dlja šyfruvannja ta dešyfruvannja zobraženʹ. Systemy obrobky informaciï. # 3(101). – S. 53-61 Vinnycja.
  5. Krasylenko, V.H., Ohorodnyk, K., Flavycʹka, Ju. (2010) Modeljuvannja matryčnyx afinnyx alhorytmiv dlja šyfruvannja kolʹorovyx zobraženʹ, Kompjuterni texnolohiji: nauka i osvita: tezy dopovidej V Vseukr. nauk.-pr. konf. – K., 2010. – S. 120-124.
  6. Krasylenko, V.H., Nikolʹsʹkyj, O.I., Lazarjev, O.O. (2008) Modeljuvannja modyfikovanoho alhorytmu stvorennja 2-D ključa v kryptohrafičnyx zastosuvannjax. Nauka i navčalʹnyj proces: naukovo-metodyčnyj zbirnyk naukovo-praktyčnoji konferenciji. – Vinnycja, 2008. – S. 107-109.
  7. Krasylenko, V.H., Hrabovljak, S.K., (2011) Matryčni afinni šyfry dlja stvorennja cyfrovyx slipyx pidpysiv na tekstohrafični dokumenty. Systemy obrobky informaciji: zb. nauk. pr. – X.: XUPS, 2011. – Vyp. 7 (97). – S. 60-63.
  8. Krasylenko, V.H., Svirenjuk, S.A. (2006) Rozrobka metodu kryptohrafičnoho zaxystu informaciji tekstohrafičnoho typu. Nauka i navčalʹnyj proces: naukovo-metodyčnyj zbirnyk naukovo-praktyčnoji konferenciji. – Vinnycja, 2006. – S. 73-74.
  9. Krasylenko V.H., Hrabovljak, S.K. (2012) Ocinjuvannja stijkosti ta času zlamuvannja u matryčno-perestanovočnyx alhorytmax kryptoperetvorenʹ. Nauka i navčalʹnyj proces: naukovo-metodyčnyj zbirnyk materialiv naukovo- praktyčnoji konferenciji VSEI Universytetu “Ukrajina”. – Vinnycja, 2012. – S. 173-174.
  10. Krasylenko, V.H., Hrabovljak, S.K. (2012) Modeljuvannja matryčnoho afinno- perestanovočnoho alhorytmu dlja kryptoperetvorenʹ zobraženʹ. Nauka i navčalʹnyj proces: naukovo-metodyčnyj zbirnyk materialiv naukovo-praktyčnoji konferenciji VSEI Universytetu “Ukrajina”. – Vinnycja, 2012. – S. 171-172.

View full text (PDF)