Active Image Forgery Detection Using Cellular Automata - Cellular Automata in Image Processing and Geometry - page 133

Original Images Output Images

Fig. 7.6 Scenario 1: input and output images

7.4.2 Time Consumption

We select the first eight pixels of the original image to embed the proposed cipher,
as we mentioned in Sections 7.2 and 7.3. Figure 7.8 presents the results of different
pixels selection with different time consumption in which 50 images for each sce-
nario separately. Figure 7.9 shows the same results with using 100 images. In this
experiment we only used 50 and 100 images from the dataset and this number of
images is only a random selective set of digital images we had. These figures show
that minimum time consumption is obtained by embedding the cipher key into the
first eight pixels of the original image.

7.4.3 True and False Alert

In Table 7.6, you can see the percentage of true and false detection of digital im-
age forgery which are performed by the proposed method. We compare these two
indices with different local rule for proposed cellular automata in Sections 7.2
and 7.3, plus different position to embed data into the original image. We use exactly
same images for both scenarios. In this experiment, three various logical operations
(XOR, AND, and OR) and two different arithmetic operations (Addition and Mul-
tiply) are implemented as our cellular automata's local rule.
As shown in this table, and by considering previous results (Time consumption),
scenario 2 using XOR local rule with data embedding in only first eight pixels of
the original images are better in average.