A Novel Partial Thermal Image Encryption Scheme Based on Multiple Chaotic Systems

In this paper, proposed an novel partial encryption scheme for protecting thermal images during transmission. The adopted technique having pixel level and bit level permutation process. To reduce time and computational complexity , an partial encryption technique is adopted. Arnold cat map technique involved in the pixel level permutation process to decrease correlation of original image. The proposed scheme selects only higher four binary planes for the bit level shufﬂing process. Where this process consists of row and column wise circular shift operation with the help of pseudo random numbers generated from chaotic map system. Mixing of binary planes of the thermal image will leads to prevent the differential attacks. Experimental results shows that proposed scheme exhibits good performance in terms entropy, NPCR and the speed of system.


Introduction
Now a days fast growing of information and communication technologies, there are many ways of transmitting information from source to destination. While transmitting, securing of information is very important. Thermal image is one which requires security. Ciphering is one way to achieve security of thermal images. Yueping etc. al [1] proposed a hyper chaos based image encryption algorithm. which consists of both pixel and bit level permutation process. where 5-D multi-wing hyper chaotic system enabled to generate key stream and pseudo random numbers.Chaos based expand and shrink techniques [2] applied for image encryption to increase the security. Multiple times of chaos based permutation technique applied to reduce the correlation among the pixels and substitution increases the entropy of encrypted image.Sukalyan Som etc.al [3] presents an chaos based partial image encryption scheme. Plain image decomposed to bit planes. in order to separate significant and insignificant bit planes threshold based autocorrelation techniques adapted. Anish Goel and Kaustubh Chaudhari [4] explained abut median based pixel selection for partial image encryption. Encrypted masks are used to select the pixels of interest and AES technique applied to each region of interest blocks to encrypt.Infrared target based selective encryption of thermal image proposed [5]. Which uses the chaotic maps for the encryption Contour model used to detect the infrared region. Next block cross model used for encrypting pixels of extracted infrared region of interest. Panduranga etc. al [6] introduced an partial image encryption scheme based on block wise shuffling and chaotic systems. plain image divided into non overlapping blocks of different size and pixels of each blocks gets permuted with the help of random sequence generated from the chaotic systems.Laiphrakpam Dolendro Singh and Khumanthem Manglem Singh [7] explained about medical image encryption based on improved EIGamal algorithm. Separate encoding of plain message adopted and data expansion case is removed and that will enhance the execution speed of the system. Novel method for securing thermal images has been proposed [8]. proposed method depends on chebyshev chaotic map and s8 symmetric group of permutation based substitution boxes. Parameters to map are considered as the secrete key of the encryption system.

Chaotic Map
Selecting proper map has been one of the important stages in a digital chaotic. As far as the complexity is concerned, chaotic maps exhibit important characteristics such as, chaotic interval, cycle length, chaotic properties, periodic windows, etc. and it has highly sensitive to its initial conditions. There are instances where the weaknesses of the chaotic map were ignored and the systems suffered a breakdown. Hence there is necessity to design cryptosystems independent of the chaotic maps. That is, there is no necessity to possess complete knowledge of chosen chaotic map to make the good and efficient cryptosystem independent. In order to simplify the mathematics concepts, either logistic map or tent map can be used. Below equation represents the logistic map:

Proposed Partial Thermal Image Encryption
The following block diagram of proposed partial thermal image encryption scheme contain two processes namely a. Pixel Level Permutation b. Bit Level Permutation Row int = f loor(mod(X * (10 14 )), m) + 1 Col int = f loor(mod(Y * (10 14 )), 4 * n) + 1 for the permutation operation circular shifting of row and column of the image is considered.

Parameters for the evaluation of an partial thermal image encryption scheme 4.1. Information entropy analysis
Entropy defined as amount of information, which also indicates important characteristics of image disorder and randomness. Entropy H(X) of a source x, we have: Where X represents the image to testx i denotes the i t h possible value in X, and Pr(x i ) is the probability of X =x i . Theoretical value for H(X)=8 then it indicates that randomness of image is more and that will secure the input image [10].

Mean Square Error (MSE)
MSE can be calculated between input and encrypted image by taking mean of squared difference between them. MSE value more amount of nose introduced more and decreases signal strength. Let I1 and E1 are source image and cipher image respectively, then MSE given by Eq. 5 [11].
where h, w is row and column of picture and I1(p,q) is source image and E1(i,j) is cipher image.

Peak Signal to Noise Ratio (PSNR)
Peak signal-to noise ratio inversely pro-portional to Mean Square Error (MSE). PSNR re ects the ciphering quality. MSE is more PSNR is less and vice versa. PSNR value indicates how signal strength is more. Mathematically as in [11].

UACI and NPCR
To check the proposed ciphering technique is sensitive to source image and keys, they are two tests: Number of pixels change rate (NPCR) and Unified average changing intensity (UACI) [9]. The equation to calculate UACI is Eq. 7.
Where, m represents number of rows, n indicates number of column, I(p,q) and E(p,q) are the original and cipher image respectively. NPCR can be calculated by Eq. 8.
Where, m represents number of rows, n indicates number of column and where D(i,j) defined as follows where I(i,j) and E(i,j) are the original and cipher image respectively. Figure 2 shows the Plain images of MRI, Man, Hand and Legs respectively. Each image size of 256*256 and undergo permutation process with the help of Arnold cat map and obtained the permuted images of MRI, Man, Hand and Legs respectively as shown in figure  3. Figure 4 shows the encrypted images of MRI, Man, Hand and Legs respectively.
From the table 1, E1 and E2 are the entropy of original image and encrypted image respectively. It is observe that entropy of encrypted image is increased and approached to ideal value. Amount of encryption increases as the means square error value increases. Other Parameters are also increased and approached to ideal values. Because of encrypting only upper four bit planes of thermal image, encryption speed of the system increased.

Conclusion
Partial encryption technique for protecting thermal image have been proposed to reduce time and the computation complexity of traditional encryption algorithms. Arnold cat map technique involved in the pixel level permutation process. The proposed scheme selects only higher four binary planes for the bit level shuffling process.
Where this process consists of row and column wise circular shift operation with the help of pseudo random numbers generated from chaotic map system. Proposed method can be adopted for practical image transmission due to its fast encryption speed and robustness against cryptanalytic attacks.