Active Image Forgery Detection Using Cellular Automata - Cellular Automata in Image Processing and Geometry - page 126

Ta b l e 7 . 2 An example of cellular automata

Cell Number 0 1 2 3 4

Input Sequence (time t ) 0 1 1 1 0

Cell[X] t+1 = Cell[X-1] t (AND) Cell[X+1] t Cellular Automata Rule

Output Sequence (time t + 1) 0 0 1 0 0

drawing a block diagram of our method (Figure 7.2), we just assume that the input
type would be a .jpeg image.
Cellular automata have been implemented to create the required cipher key bit
sequence. The XOR local rule used to generate the result in this chapter. We generate
a cipher key by using specific cellular automata with an XOR local rule on six cells
(see Table 7.3). In 2013 Sharma et al. [14] proposed a text security approach based
on a XOR rule within 2D cellular automata and get well formed results. In 2011
Prasad Panda et al. [12] proposed a cellular automata encryption and decryption
algorithm for block cipher based on XOR rules. Our experiments also show that
a XOR logical operation often suffices to obtain a very sensitive cipher key. We
have achieved a better rate for 'True alert' indicates true forgery detection, by using
XOR rule rather than some other logical rules and also arithmetic rules (Table 7.6).
Based on the experimental validations, our proposed XOR rule could improve PSNR
(Table 7.7). Furthermore, it can be done extremely fast on contemporary CPUs that
mostly provide a specific instruction to do a XOR operation [7].
We only use three number of statistical information of the LU decomposition
matrices of the original image to generate the cipher key. This information consists
of arithmetic mean, median, and the statistics range (Table 7.3). If anybody wants to
modify a digital image, then the statistical information of these particular matrices
will be changed, so the output of the proposed cellular automata will be damaged.
In this proposed method we consider single iteration cellular automata, imple-
menting zero padding to obtain a valid value for boundary cells (Cells 0 and 5 in
Table 7.3). We add zeros to both front and rear of our cellular automata so that the

t + 1

Ta b l e 7 . 3 Proposed six cells for the cellular automata with XOR local rule: Cell [ X ] =

t t

Cell [ X − 1 ] (XOR) Cell [ X + 1 ]

Cell Number Input Value (time t )
0 Mean of the values in the L Matrix from LU decomposition of the original image

1 Mean of the values in the U Matrix from LU decomposition of the original image

2 Median of the values in the L Matrix from LU decomposition of the original image

3 Median of the values in the U Matrix from LU decomposition of the original image

4 Range of the values in the L Matrix from LU decomposition of the original image

5 Range of the values in the U Matrix from LU decomposition of the original image