Académique Documents
Professionnel Documents
Culture Documents
g Steganography
g g p y
By Paul Hammes
What is Steganography?
z Hiding in plain sight
z C
Comes ffrom greek
k
z ‘steganos’ – covered
z ‘graphie’ – writing
z Good Mediums
z Images
I
z Audio
z Text
Uses for Steganography
z Information hiding, convey confidential
information where cryptography is not viable
z Third party cannot know that you have confidential
i f
information
ti
z Authenticate Documents
z If embedded data destroyed then medium has been
tampered with
z Protect Intellectual Property
Aspects of Steg Methods
z Invisibility
z How detectable by observer
z Storage Capacity
z How much data can we hide in the medium
z Robustness against attack
z How secure is it against analysis
z R b t
Robustness against
i t manipulation
i l ti
z Can message be distorded by manipulation
z Resizing,
Resizing cropping
cropping, etc
What is a computer image?
z Computer image is an array of pixels
z Pixels represent different light intensities
z Are encoded as bit patterns
z Bit
Bit--depth
z 8bit grayscale – 256 levels of gray
z 24bit RGB – 16,777,216 million colors
Image Compression
z Lossy vs Lossless Compression
z Lossy (JPEG)
z Discards excess image data, removes details that are too
small for human eye to differentiate
z Smaller file size
z Lossless
L l (BMP
(BMP,GIF)
GIF)
z No loss of data, uncompressed image bit
bit--by
by--bit identical
to original
z Larger file size
Image vs Transform Domain
z Image Domain
z Change bits of image pixels
z Simpler
z Lossless image compression
z Transform Domain
z More Robust
z Lossy and Lossless image compression
Types of Steg Explored
z LSB Hiding
z Patchwork
P t h k
z Decode
z Extract
LSB
S from
f steganography image and make
MSB of new image
LSB Hiding Conclusion
z Visibility of message being hidden depends
on cover image
z Less solid areas are better
z 1bit and 2bit virtually undetectable
message
Cover image
LSB Hiding w/ Patchwork
Encoded Image Encoded Data
(2bits per patch)
LSB Hiding w/ Patchwork
z Reconstructed Message
Patchwork Conclusion
z Visibility depends on cover image and number
of bits encoded
z Can g get better reconstruction
z More Robust against alteration as data is
redundant
z Can not store as much data
Hiding in Frequency Domain
z Using FFT to hide message in the
frequency domain of image