You are on page 1of 25

Model of Noise and Color Restoration :

Application for Restoration and Steganalysis

Hasan ABDULRAHMAM 
Supervisors
Marc CHAUMONT Philippe MONTESINOS
Last Meeting in Friday : 12 / 09/ 2014
First meeting in : 2nd year Date : 10 / 10 / 2014

11/11/22 1
Work done in first year

noise R .F
R residual 34617
Add all
noise
features in
G G .F
residual 34671 one vector
104013
B noise
B.F features
residual
34671

11/11/22 2
Plan for First & Second years

1st year 2ed year 3rd year

 Study the different types of noise.  Build big database 15.000 RGB
color image based on CFA tech.
 Study Steganography and
Steganalysis Techniques.  Test different types of color
image filters.
 Experiment embedding secret
message using four image  Apply more techniques for color
Steganography methods. image Steganalysis.

 Modify ( SRM ) to get 104.013  Apply color filter depending on


features from each RGB color image restoration process to
image, to make good classification. detect Steganography method.

 Analysis the experimental results.

11/11/22 3
Database Sets
 Build new Database sets
 The new database is depending on CFA ( Color Filter Array )
 collected ( Raw color images 24 bits ) from the following :-

1- subset of BOSSBase [1], ( 1000 images )


2- subset of Dresden image database[2] ( 4000 images )

[1]BOSS. http://exile.felk.cvut.cz/boss/BOSS_Final
[2]Dresden database http://forensics.inf.tu-dresden.de/ddimgdb/

11/11/22 4
Work Done
IMG2BMP convert
Cropping
dcraw code
ext
sTIFF2IMG 512X512 512X512 512X512
IMG 1 BMP 1 Image 1

512X512 512X512 512X512


IMG 2 BMP 2 Image 2 25.000
Raw
TIFF IMG
Color
images image image .ppm
512X512 512X512 512X512 color
3900X2600
IMG 3 BMP 3 Image 3 images

512X512 512X512 512X512


IMG 4 BMP 4 Image 4

512X512 512X512 512X512


IMG 5 BMP 5 Image 5

11/11/22 5
Work Done
 First step : Using “dcraw” [3] ( code program in C++ ) to convert all
raw color images into color image type Tiff .

$ max= 5000
$ for (( i=1; i<=$max ; i++ )); do dcraw - T /Boss/$i.cr2; done

[3] Dave Coffin, Main page of dcraw , http://www.cybercom.net/~dcoffin/dcraw

11/11/22 6
Work Done
 Second step : Convert all Tiff images in to IMG file.

$ max= 5000
$ for (( i=1; i<=$max ; i++ )); do sTIFF2IMG /Boss/$i.tiff $i; done

11/11/22 7
Work Done
 Third step : Making the cropping starts from red channel by
choosing the position randomly.
( from each even position xi , yi ).

$ for i in 1 2 3 4 5; do echo $i;


ext –x 512 – y 512 –ixr - iyr in_file_name out_file.$i; done

11/11/22 8
Work Done
 Fourth step : Convert each IMG file in to BMP.

$ IMG2BMP img_file_name bmp_file_name

11/11/22 9
Work Done
 Fifth step : Convert each BMP file in to PPM file.

$ convert bmp_file_name - compress none ppm_file_name

11/11/22 10
Work Done
Cropping sample 1

Crop 1 Crop 4
512 X 512 512 X 512

Crop 3
Original Image 512 X 512
6250X4000

Crop 2 Crop 5
512 X 512 512 X 512
11/11/22 11
Work Done
 Random cropping starts from Red channel.
hasan@$ for i in 1 2 3 4 5 ; do echo $i ; ext -x 512 -y 512 -ixr -iyr sortie2.img 505_new.$i; done
crop 1
random ix : 2116
random iy : 1928
crop 2
random ix : 902
random iy : 1182
crop 3
random ix : 3080
random iy : 436
crop 4
random ix : 1866
random iy : 1778
crop 5
random ix : 650
random iy : 1032

11/11/22 12
Work Done
Cropping sample 2

Crop 1 Crop 4
512 X 512 512 X 512

Crop 3
Original Image 512 X 512
6250X4000

Crop 5
Crop 2
512 X 512
11/11/22 512 X 512 13
Work Done
 Random cropping starts from Red Channel.
hasan@$ for i in 1 2 3 4 5 ; do echo $i ; ext -x 512 -y 512 -ixr -iyr sortie1.img 202_new.$i; done
Crop 1
random ix : 2790
random iy : 932
Crop 2
random ix : 1576
random iy : 186
Crop 3
random ix : 360
random iy : 1528
Crop 4
random ix : 2540
random iy : 782
Crop 5
random ix : 1324
random iy : 36

11/11/22 14
Work Done
Before we fiished embedding secret message in ( 280.000 ) color images
Now we finished embedding secret message in ( 330.000 ) color images
110.000 color images for each Steganography method,
10.000 color images for each palyload ratio.

HUGO WOW S-UNIWARD


Payload
Ratio Red Green Blue Red Green Blue Red Green Blue

0.6 0.15 0.15 0.3 0.15 0.15 0.3 0.15 0.15 0.3
0.6 0.15 0.3 0.15 0.15 0.3 0.15 0.15 0.3 0.15
0.6 0.3 0.15 0.15 0.3 0.15 0.15 0.3 0.15 0.15
0.6 0.1 0.4 0.1 0.1 0.4 0.1 0.1 0.4 0.1
0.6 0 0.3 0.3 0 0.3 0.3 0 0.3 0.3
0.6 0.3 0.3 0 0.3 0.3 0 0.3 0.3 0

11/11/22 15
Work Done
 Finished embedding secret message in ( 330.000 ) color images

HUGO WOW S-UNIWARD


Payload
Ratio Red Green Blue Red Green Blue Red Green Blue

0.3 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1
0.3 0.1 0 0.2 0.1 0 0.2 0.1 0 0.2
0.3 0 0.2 0.1 0 0.2 0.1 0 0.2 0.1
0.3 0.05 0.05 0.2 0.05 0.05 0.2 0.05 0.05 0.2
0.3 0.15 0 0.15 0.15 0 0.15 0.15 0 0.15
0.3 0 0.15 0.15 0 0.15 0.15 0 0.15 0.15

0.3 0.15 0.15 0 0.15 0.15 0 0.15 0.15 0

11/11/22 16
Experimental results
 Previous results with the results of complementary

Payload HUGO WOW S-UNIWARD LSB


Ratio
0.6 Green Channel 95.54% 95.22% 96.12% 100%
0.4 Green Channel 92.25% 91.62% 92.72% 100%
0.2 Green Channel 85.47% 84.67% 85.87% 96.56%
0.3 Red Channel 86.67% 84.30% 85.98% 100%
0.3 Blue Channel 87.19% 85.25% 87.39% 100%
0.15 Red + 0.15 Blue 80.64% 77.78% 79.93% 94.26%
0.2 Green + 0.1 Blue 85.51%

0.1 Red + 0.1 Green + 0.1 Blue 77.82


0.15 Red + 0.3 Green + 0.15 Blue 88.95% 88.18% 89.44% 100%

11/11/22 17
Experimental results
 Steganalysis Detection rate
Payload S-UNIWARD
Ratio
0.6 Green Channel 96.12%
0.4 Green Channel 92.72%
0.2 Green Channel 85.87%
0.3 Red Channel 85.98%
0.3 Blue Channel 87.39%
0.15 Red + 0.15 Blue 79.93%
0.2 Green + 0.1 Blue 85.51%
0.1 Red + 0.1 Green + 0.1 Blue 77.82
0.15 Red + 0.3 Green + 0.15 Blue 89.44%

11/11/22 18
Experimental results
 ROC curves S-UNIWARD payload ( 0.2 Green + 0.1 Blue ) channels
Detection rate = 85.51%

11/11/22 19
Experimental results
 ROC curves S-UNIWARD payload ( 0.1 Red+ 0.1 Green + 0.1 Blue ) channels
Detection rate = 77.82%

11/11/22 20
Experimental results
 Comparison between color Steganalysis Detection rate with
Gray level ( one channel ) steganalysis detection rate.
Payload S-UNIWARD S-INIWARD
Ratio Color image (One Channel)

0.6 Green Channel 96.12% 95.88%


0.2 Green Channel 85.87% 84.39%
0.3 Red Channel 85.98% 85.60%
0.3 Blue Channel 87.39% 86.60%

11/11/22
21
Experimental results
 Comparison between S-UNIWARD payload 0.3 Roc curves

11/11/22 22
Experimental results
 Comparison between S-UNIWARD payload 0.6 Roc curves

11/11/22 23
Experimental results
 Working with BossBase_1.0 ( 10.000 ) database
graylevel images type PGM file format
 Finished embedding secret message in ( 30.000 ) Gray images*
 10.000 for each payload using S-UNIWARD Steganography
Payload S-UNIWARD
Ratio
0.6 10000
0.3 10000
0.2 10000

* BossBase_1.0 ( 10.000 ) gray images type PGM file format.


 Notes : I finished embedding secret message 0.2 ratio in 10.000 graylevel (pgm )
images ( downloaded from Boss website).
 I finished extracting features today.

11/11/22 24
11/11/22 25

You might also like