Abstract: Background: In current years, numerous algorithms of image encryption were considered and developed through diffusion and confusion, to secure the image against hackers. Materials and Methods: In this study a novel idea is proposed to encrypt and decrypt an image. Here two levels of encryption are used. In the first level the plane encryption is done, due to this there will be an absolute changes in the bits of each plane and in the second level the pixel encryption is done because of this the pixel values are changed. All these things makes harder to decrypt the image by the hackers in the channel. Results: The number of pixel change rate (NPCR) and Unified Average Changing Intensity (UACI) values are calculated to test the randomness of the encrypted images. Horizontal, vertical and diagonal correlations are tested to find the relationship between two adjacent pixels. To compare the pixel distribution of original and encrypted image histograms are plotted. Conclusion: The statistical analysis of data obtained from the results based on the proposed technique can offer a greater quality of information transmission security.
INTRODUCTION
Transmission of highly confidential data in a secured way has become a matter of great concern in the current days. The modern technology has made it effortless for a third person to hack the information. This drawback has given birth to the idea of communication via images. There have been a lot of studies by researchers on image encryption and decryption. Plenty of methods have been evolved. Since the method of plane encryption and pixel encryption proves to furnish extremely accurate results, its relevance to image encryption and decryption is highly appreciated1-5.
The image can be decomposed by several number of bit planes using decomposition techniques. Some of the decomposition techniques are gray code bit plane decomposition (GCBD), binary bit plane decomposition (BBD)1 and fibonacci p-codes bit plane decomposition2-4. The images are divided into several bit planes in which any one bit plane can act as secret key and for encryption, XOR operation is performed between the secret key planes with all other planes5.
Image encryption through confusion and diffusion is performed using Matrix Reordering (MR) and simple XOR operation6-10. Pixel level scrambling can be done by using simple XOR operation7,8. The image is divided into several number of vectors and the number varies based on the key value K and XOR operation is done between two neighbouring vectors to achieve the diffusion and each vector bits are circularly rotated right to achieve confusion8,9. Bit level permutation is done by using chaotic map in order to encrypt the image9-11. Images are encrypted by using value transformation and random permutation10,12. Optical XOR operation is performed between bit plane and key data by polarization encoding technique13.
Image positions are changed and non-linear substitution and diffusion are applied for each row of the image14,15. This substitution and diffusion are repeated for 3 times to create an encrypted image chaos theory and vigenère cipher is used12,15-17. Chaotic sequences created by chaos are arranged as Vigenère then pixel positions are scrambled based on chaotic sequences11,12,14,17-20. Simple classifications on Image encryption are pixel level7, plane level8, matrix compressing21-23 or chaotic sequences24-27.
In pixel level and plane level encryption, the image data is encrypted using simple XOR and XNOR operations with high NPCR, UACI value21-27. Generally, the gray scale images are represented using 8 bits. In this study by using simple pixel and plane level encryption, a significant improvement can be achieved.
MATERIALS AND METHODS
Proposed scheme: Plane level encryption and pixel level encryption have been implemented in this proposed scheme. The XOR and XNOR operations have been used to develop the encryption strategies25,26. This would ensure higher level of image security, since two levels of encryption techniques have been pursued. The experiment has been performed with the standard gray scale images, "Lena", "Cameraman", "Pepper" and "Baboon" of the same size.
Plane level encryption: In this encryption scheme, MSB and LSB planes are encrypted separately as shown in Fig. 1a and 1b, respectively. The plane level encryption algorithm is given. Here confusion and diffusion can be achieved with help of alternate XOR and XNOR operation.
Algorithm 1: |
Pixel level encryption: In the pixel level encryption scheme, the data bits of each pixel in the image is encrypted using data bits of other pixels in the image.
Fig. 1(a-b): | Encrypting the (a) MSB plane and (b) LSB plane |
Fig. 2(a-b): | (a) First half pixels encryption and (b) Second half pixels encryption |
The image is divided into two halfs where the first and second half are encrypted separately as shown in Fig. 2a and b, respectively. Here the XOR operation is performed on even bits of pixels and XNOR operation is performed on odd bits of pixels. The pixel level encryption is given.
Algorithm 2: |
RESULTS AND DISCUSSION
The proposed method was simulated using jdk 1.6. The 256×256 lena image is given in Fig. 3a and its histogram is given in Fig. 3b. After first level encryption the encrypted lena image is given in Fig. 4a and its corresponding histogram is given in Fig. 4b.
Fig. 3(a-b): | (a) Lena image and (b) Histogram of Lena image |
After second level, the encrypted lena image and its corresponding histogram are shown in Fig. 5a and 5b, respectively.
Fig. 4(a-b): | (a) First level encrypted Lena and (b) Histogram of first level encrypted Lena |
Fig. 5(a-b): | (a) Second level encrypted Lena and (b) Histogram of second level encrypted Lena |
Table 1: | Comparison of correlation with the previous studies |
Table 2: | Comparison of NPCR with the previous studies |
In the proposed method the correlation between pixels for encrypted lena image is better than19,23 as given in Table 1.
The NPCR is for lena image good when comparing to the other algorithms16,17,27 and it is shown in Table 2.
The 256×256 cameraman image is given in Fig. 6a and its histogram in Fig. 6d. After first level encryption the encrypted cameraman image and its corresponding histogram is given in Fig. 6b and e, respectively. The second level encrypted cameraman image and its corresponding histogram is given in the Fig. 6c and f, respectively.
The 256×256 pepper image is given in Fig. 7a and its histogram in Fig. 7d. After first level encryption the encrypted pepper image and its corresponding histogram is given in Fig. 7b and e, respectively. The second level encrypted pepper image and its corresponding histogram is given in the Fig. 7c and f, respectively.
The 256×256 baboon image is given in Fig. 8a and its histogram in Fig. 8d. After first level encryption the encrypted baboon image and its corresponding histogram is given in Fig. 8b and e, respectively. The second level encrypted baboon image and its corresponding histogram is given in the Fig. 8c and f, respectively.
Table 3 provides the NPCR, UACI and correlation values of the proposed scheme for different images when comparing to the other algorithms20-22.
Fig. 6(a-f): | (a) Cameraman image, (b) Encrypted image after level 1, (c) Encrypted image after level 2, (d) Histogram of cameraman, (e) Histogram after level 1 and (f) Histogram after level 2 |
Fig. 7(a-f): | (a) Pepper image, (b) Encrypted image after level 1, (c) Encrypted image after level 2, (d) Histogram of pepper, (e) Histogram after level 1 and (f) Histogram after level 2 |
Fig. 8(a-f): | (a) Baboon image, (b) Encrypted image after level 1, (c) Encrypted image after level 2, (d) Histogram of baboon, (e) Histogram after level 1 and (f) Histogram after level 2 |
Table 3: | NPCR, UACI and correlation values of proposed scheme |
CONCLUSION
An image-encryption scheme based on plane level and pixel level using XOR method is presented. The decryption is reverse method of encryption. The advantages and the necessity of the diffusion and confusion in this proposed scheme is demonstrated the simulation results shown is effective, robust and secure to encrypt and decrypt the images high security and computation complexity has been achieved due to the two level encryption process. In future, by using chaotic system can randomize the plane and pixel level encryption. This method is also applicable for colour images.
SIGNIFICANCE STATEMENTS
• | A novel dual level image encryption algorithm is proposed |
• | Encryption at plane and pixel level is done as the first and second layer, respectively |
• | High security and computation complexity has been achieved |