Steganography application of effective Genetic Shifting Algorithm against RS Analysis Advisor: MScEE, Vladislav Kaplan Vadim Purinson LV Tailoring Software
Aug 08, 2015
Steganography application of effective Genetic Shifting Algorithm against RS
Analysis
Advisor: MScEE, Vladislav Kaplan
Vadim Purinson LV Tailoring Software
Who owns the information – owns the world
Rotschild Nathan Mayer•What is Steganography?
•Why Steganography?•History of Steganography.•Digital steganography.•Steganalysis.•RS Fridrich analysis.•Genetic shifting algorithm.•Software implementation. •Tests•Conclusions
04/15/2023LV Tailoring Software
2
What is Steganography?
The art and science of hiding information by embedding it in some other media.
•Steganography versus cryptography
•In general, steganography approaches hide a message in a cover e.g. text, image, audio file, etc.
04/15/2023LV Tailoring Software
3
Why Steganography?
•Transfer secret information or embed secret messages into media.
•Data, intellectual property and privacy protection - Digital Water marking, medical data.
04/15/2023LV Tailoring Software
4
History of Steganography.•400 B.C. – writings of Herodotus•1499 – “Steganographia”, Trithemius –
steganography and magic.•1665 – Steganographica, Gaspari Schotti.•1870 – The Pigeon Post into Paris.
Most popular example in history this is use invisible inks.
04/15/2023LV Tailoring Software
5
Terminology and Definitions
•Steganographic system or stegasystem – this is set of tools and methods are used to generate a secret channel of information transmission.
04/15/2023LV Tailoring Software
7
Digital image definitionA digital image is binary representation of a two dimensional image and contains a fixed number of rows and columns of pixels.
• Pixel• Byte• Bit
04/15/2023LV Tailoring Software
8
Equation 1
Message embedding mathematical definition
8 bit Grayscale equivalents to 1 byte per pixel. For example, for the image size of 7 Kbyte maximum message size can be embedded, by using 1 LSB is 7168 bit
04/15/2023LV Tailoring Software
9
Equation 2
Digital steganography.Advantages of digital images
steganography is:
• There are a variety of methods used in which information can be hidden in the images.
• Relatively large volume of digital images representation, that allows the embedding of large amount of information.
• Known size of the cover media, that absence of restrictions, requirements imposed by real-time.
• Presence of relatively large textural regions in most digital images that have noise structure and well suited for information integration.
• Weak sensitivity of the human eye to minor changes the color of the image, brightness, contrast and the noise presence.
• Image steganography has come quite far with the development of fast, powerful graphical computers.
04/15/2023LV Tailoring Software
10
IMAGE STEGANOGRAPHY TECHNIQUES
•Least Significant Bit insertion – LSB
•Masking and filtering
04/15/2023LV Tailoring Software
11
Classification of Steganography Methods.
•Substitution methods in spatial domain = LSB;
•Transform domain = frequency domain;•Spread spectrum techniques;•Statistical methods;•Distortion techniques; •Cover generation methods;
04/15/2023LV Tailoring Software
15
The properties of the human eye used in the steganography
.
•selectivity to brightness fluctuations;
•frequency sensitivity;
•masking effect;
04/15/2023LV Tailoring Software
16
Selectivity to brightness fluctuations
04/15/2023LV Tailoring Software
17
Human eye sensitivity to contrast.
Experimental data by Aubert (1865), Koenig and Brodhun (1889) and Blanchard (1918). It indicates that the Weber-Fechner law - according to which the smallest perceptible change in intensity vs. intensity level I is constant.
Steganalysis
•Steganalysis
•Steganalyst
•Attack on steganography system
04/15/2023LV Tailoring Software
20
Stegattacks classes• Attack with the knowledge of the modified media
only. • Attack with knowledge of unmodified container.
04/15/2023LV Tailoring Software
21
The results of stegattack
• Detect secret message presence.• Recover secret message from stegoimage.• Destroy the message in case no possibility to
recover message.
Main methods of stegattack
•Visual analysis – detect visual image degradation by “naked” eye.
•Statistical Histogram and STD analysis.
•Detection methods are based on data hiding analyzing the characteristics of the probability distribution of the container.
04/15/2023LV Tailoring Software
23
Visual Attacks
04/15/2023LV Tailoring Software
24
Grayscale image visual attack example
Grayscale image filter visual attack example
Statistical Analysis Attack
•Stego Only Attack;
•Known Cover Attack;
•Known Message Attack;
•Blind Steganalysis;
•Semi-blind;
04/15/2023LV Tailoring Software
26
Fridrich RS Analysis
•“regular”•“singular”•“unchanged”
04/15/2023LV Tailoring Software
27
←𝑅−𝑚𝑆−𝑚→𝑅𝑚↔𝑆𝑚𝑅𝑚>𝑆𝑚
𝑅−𝑚>𝑆−𝑚
Genetic Shifting Algorithm (Shen Wang )
04/15/2023LV Tailoring Software
28
Before or After ?
𝐶=∑𝑖
𝑁 (𝑖+1 )− 𝑖𝑁−1
– this is number of pixels, and and are indicate current and next pixel values.
Front Panel04/15/2023LV Tailoring Software
30
Example of Front panel view
• Modern• System• Classic• Express• Control
Design & Simulation
• .Net & ActiveX
• Signal processing
• Add ons• User
Controls• Select and
control• DSC
Module• RF
Communications
• Sound & Vibration
• VisionControls palette view
More functions
•Tools palette • Operating tool• Positioning tool• Labeling tool• Wiring tool
04/15/2023LV Tailoring Software
32
• Wiring
• SubVis
Front panel view04/15/2023LV Tailoring Software
34
1. Source cover image
2. Result stegoimage
3. Cover Histogram and STD statistic window
4. Stegoimage Histogram and STD statistic window
5. RS analysis results on Stegoimage
6. LSB level to be used (up to LSB-4)
7. Start shifting (GSM)
8. Standard deviation evaluation
9. Snaked array length.
10. Open output result text message
11. Decode message from stegoimage
12. Encode message into cover message
13. Stop button
14. Start / stop menu
Steps definition
• Perform basic message coding (Cover Image) up to LSB-4 for gray images.
• Perform basic message recovery (Stego Image) up to LSB-4 for gray images.
• Compare visual image degradation.• Compare visual degradation through common tools (Histogram,
STD). • Perform study of coded message saturation (message of different
length) vs. recovery and image degradation per different LSB coding at gray images.
• Build RS analysis (Fridrich algorithm) routine.• Confirm validity of RS analysis on gray images.• Implement secure genetic steganography method for RS baseline
shifting for LSB-1. (GSM for RS shifting).• Perform basic message recovery with GSM for RS shifting for
LSB-1.• Perform RS analysis comparison for different message length
with GSM for RS shifting and without, use different “snake” division array image representation.
04/15/2023LV Tailoring Software
37
Compare visual image degradation
04/15/2023LV Tailoring Software
38
LSB-1The recovered text file size is 6.21 KB (6,361 bytes), 1177 words text, equivalent to 2.5 pages in WORD format.
Compare visual image degradation04/15/2023LV Tailoring Software
40
LSB-2
The recovered text file size is 12.4 KB (12,737 bytes), 2228 words text, equivalent to 5 pages in WORD format.
Compare visual image degradation
04/15/2023LV Tailoring Software
41
LSB-3
The recovered text file size is 18.6 KB (19,106 bytes), 3317 words text, equivalent to 7.5 pages in WORD format.
Compare visual image degradation04/15/2023LV Tailoring Software
42
LSB-4
The recovered text file size is 24.8 KB (25,477 bytes), 4468 words text, equivalent to 10 pages in WORD format – this is a maximum text file size can be imbedded into image by using 4LSB plane.
Compare visual degradation through common tools
04/15/2023LV Tailoring Software
44
Different depth of the LSB
Blue line is displays Cover image Histogram and red line represents manipulated image distribution.
Compare visual degradation through common tools
04/15/2023LV Tailoring Software
45
Histogram degradation trough of message enlargement for 4LSB level
differences versus message volume.
04/15/2023LV Tailoring Software
49
Image1
Message % of Rm % of Sm diff
Message
length(bytes) % of image
0 54.9 45.1 9.8 0 0
1 53.8 46.2 7.6 699 1.05
2 52.1 47.9 4.2 2,225 3.35
3 51.5 48.5 3 4,055 6.11
4 51.4 48.6 2.8 4,222 6.36
5 50.8 49.2 1.6 6,201 9.34
6 49.8 50.2 -0.4 8,169 12.3
Image2
0 54.2 45.8 8.4 0 0
1 53.9 46.1 7.8 699 1.78
2 52.5 47.5 5 2,225 5.67
3 51.8 48.2 3.6 4,055 10.34
4 51.7 48.3 3.4 4,222 10.76
5 50 50 0 6,201 15.8
6 49.8 50.2 -0.4 6,361 16.33
Image3
0 52.5 47.5 6 0 0
1 52 48 5 699 1.05
2 51.8 48.2 4.4 2,225 3.35
3 51.9 48.1 4 4,055 6.11
4 51.7 48.3 4 4,222 6.36
5 50.9 49.1 3.4 6,201 9.34
6 50.8 49.2 0.6 8,169 12.31
Image 2 RS analysis results.
04/15/2023LV Tailoring Software
51
2LSB
Message
% of
Rm % of Sm diff
Message
length(bytes)% of
image
0 54.2 45.8 8.4 0 0
1 54 46 8 699 1.78
2 53.4 46.6 6.8 2,225 5.67
3 52.8 47.2 5.6 4,055 10.34
4 52.7 47.3 5.4 4,222 10.76
5 52.1 47.9 4.2 6,201 15.8
6 50.9 49.1 1.8 12,737 24.31
3LSB
0 54.2 45.8 8.4 0 0
1 54.2 45.8 8.4 699 1.78
2 54.2 45.8 8.4 2,225 5.67
3 53.9 46.1 7.8 4,055 10.34
4 53.9 46.1 7.8 4,222 10.76
5 53.4 46.6 6.8 6,201 15.8
6 52 48 4 19,106 36.47
4LSB
0 54.2 45.8 8.4 0 0
1 54.1 45.9 8.2 699 1.78
2 54 46 8 2,225 5.67
3 53.9 46.1 7.8 4,055 10.34
4 53.9 46.1 7.8 4,222 10.76
5 53.4 46.6 6.8 6,201 15.8
6 50.3 49.7 0.6 25,477 48.56
13 and 29 division snaked array LSB-1 RS analysis results
04/15/2023LV Tailoring Software
54
Image 1
Message % of Rm
% of
Sm diff
Message
length(bytes)% of imager
0 63 37 26 0 0
1 60.9 39.1 21.8699 1.33
2 56.9 43.1 13.82,225 4.24
3 54.5 45.5 9 4,055 7.74
4 54.5 45.5 9 4,222 8.06
5 52.4 47.6 4.8 6,201 11.83
6 51.1 48.9 2.2 6,361 12.14
Image 2
0 60 40 20 0 0
1 58.9 41.1 17.8699 1.05
2 57.1 42.9 14.22,225 3.35
3 55.3 44.7 10.64,055 6.11
4 55.5 44.5 11 4,222 6.36
5 51.7 48.3 3.4 6,201 9.34
6 51.1 48.9 2.2 8,169 12.3
Image3
0 58.1 41.9 16.20 0
1 56.6 43.4 13.2699 1.05
2 55 45 10 2,225 3.35
3 54.2 45.8 8.4 4,055 6.11
4 54.3 45.7 8.6 4,222 6.36
5 53.3 46.7 6.6 6,201 9.34
6 50.9 49.1 1.8 8,169 12.31
Image 1
Message % of Rm % of Sm diff
Message
length(bytes)
% of
image
0 58.7 41.3 17.4 0 0
1 57 43 14 699 1.05
2 54.5 45.5 9 2,225 3.35
3 53 47 6 4,055 6.11
4 52.8 47.2 5.6 4,222 6.36
5 51.2 48.8 2.4 6,201 9.34
6 50.5 49.5 1 8,169 12.3
Image 2
0 56.9 43.1 13.8 0 0
1 56 44 12 699 1.33
2 53.7 46.3 7.4 2,225 4.24
3 52.1 47.9 4.2 4,055 7.74
4 52.1 47.9 4.2 4,222 8.06
5 49.9 50.1 -0.2 6,201 11.83
6 49.7 50.3 -0.6 6,361 12.14
Image3
0 54.9 45.1 9.8 0 0
1 53.8 46.2 7.6 699 1.05
2 52.7 47.3 5.4 2,225 3.35
3 52.6 47.4 5.2 4,055 6.11
4 52.5 47.5 5 4,222 6.36
5 52.1 47.9 4.2 6,201 9.34
6 50.6 49.4 1.2 8,169 12.31
13 division snaked array 29 division snaked array
Secure genetic steganography method
04/15/2023LV Tailoring Software
55
No shifted 13 division snake array
29 division snake array 51 division snake array