Active Image Forgery Detection Using Cellular Automata - Cellular Automata in Image Processing and Geometry - page 135

Ta b l e 7 . 6 True and false detection of the proposed approaches. Three different logical oper-
ations and two arithmetic operations are used to evaluate a local rule of the proposed cellular
automata. We use a hundred images for both scenarios.

Approach True Alert % False Alert %

Scenario 1: XOR operation + embedding into first eight pixels 94.61 5.39

Scenario 1: XOR operation + embedding into first and middle eight pixels 94.61 5.39

Scenario 1: XOR operation + embedding into first, middle and end eight pixels 94.79 5.21

Scenario 1: AND operation + embedding into first eight pixels 88.67 11.33

Scenario 1: AND operation + embedding into first and middle eight pixels 88.68 11.32

Scenario 1: AND operation + embedding into first, middle and end eight pixels 88.71 11.29

Scenario 1: OR operation + embedding into first eight pixels 88.29 11.71

Scenario 1: OR operation + embedding into first and middle eight pixels 88.29 11.71

Scenario 1: OR operation + embedding into first, middle and end eight pixels 88.32 11.68

Scenario 1: Addition operation + embedding into first eight pixels 81.44 18.56

Scenario 1: Addition operation + embedding into first and middle eight pixels 81.47 18.53

Scenario 1: Addition operation + embedding into first, middle and end eight pixels 81.50 18.50

Scenario 1: Multiply operation + embedding into first eight pixels 91.14 8.86

Scenario 1: Multiply operation + embedding into first and middle eight pixels 91.15 8.85

Scenario 1: Multiply operation + embedding into first, middle and end eight pixels 91.19 8.81

Scenario 2: XOR operation + embedding into first eight pixels 96.23 3.77

Scenario 2: XOR operation + embedding into first and middle eight pixels 96.25 3.75

Scenario 2: XOR operation + embedding into first, middle and end eight pixels 96.29 3.71

Scenario 2: AND operation + embedding into first eight pixels 86.09 13.91

Scenario 2: AND operation + embedding into first and middle eight pixels 86.09 13.91

Scenario 2: AND operation + embedding into first, middle and end eight pixels 86.17 13.83

Scenario 2: OR operation + embedding into first eight pixels 86.01 13.99

Scenario 2: OR operation + embedding into first and middle eight pixels 86.03 13.97

Scenario 2: OR operation + embedding into first, middle and end eight pixels 86.08 13.92

Scenario 2: Addition operation + embedding into first eight pixels 84.73 15.27

Scenario 2: Addition operation + embedding into first and middle eight pixels 84.73 15.27

Scenario 2: Addition operation + embedding into first, middle and end eight pixels 84.77 15.23

Scenario 2: Multiply operation + embedding into first eight pixels 92.32 7.68

Scenario 2: Multiply operation + embedding into first and middle eight pixels 92.32 7.68

Scenario 2: Multiply operation + embedding into first, middle and end eight pixels 92.37 7.63

purposes, we can examine how pixel's value changes results appear in related values
of the cipher key. This table shows that even small changes in pixel values of the
original image can make a different value for the cipher key.
In this scenario, we propose a digital image forgery detection algorithm based on
the cellular automata and LU decomposition. Experiments for this scenario will be
described in detail in Section 7.4.