DIGITAL WATER MARKING

Document Sample
DIGITAL WATER MARKING Powered By Docstoc
					    GEOMETRICALLY INVARIANT
WATERMARKING USING FEATURE POINTS



             Presented By
             Shital Desai
        Madhukar V Kakaraparthi

               Instructor
           Dr. Donald Adjeroh
        Cs591 Multimedia Systems
      What is WaterMarking?
• Is the practice of imperceptibly altering a image to
  embed a message about that image
Two Important properties of WM:
 Imperceptibility– The modifications caused
 by watermark embedding should be below the
 perceptible threshold.
 Robustness– The ability of the watermark to
 resist distortion introduced by standard or
 malicious data processing.
     A watermark may be

Fragile:     distorted or broken under slight changes


Semi-fragile:        designed to break under all changes
that exceed a user- specified threshold


Robust:      withstand moderate to severe signal
processing attacks (compression, rescaling, filtering)
  Applications Of WaterMarking
1. Copyright Protection
    Most prominent application
    Embed information about the owner to prevent others from claiming
     copyright
    Require very high level of robustness

2. Database Retrieval
    For the identification of the content.
    Robustness not decisive

3.Content Authentication
    Embed a watermark to detect modifications to the cover
    The watermark in this case has low robustness, “fragile”
             Synchronization In
               WaterMarking
 The detection of the watermark requires a
  synchronization step to locate the embedded
  mark in the content.


We can illustrate this idea in:-
 1. Audio.
 2. Digital Images.
     Geometrical Distortions
Global Transformation
  Rotations, Translations…

Local Transformation
  StirMark Attack
 Classical Self Synchronization
           Techniques

 Using Periodical Sequences
 Using Templates insertions
 Using Invariant transforms
 Using original image.
            Comparison of the Sync
                  Methods
              Local            Global           Notes
              Transformation   Transformation
              robustness       Robustness
Periodic             No               Yes       -
insertion

Template             No               Yes       Can be
insertion                                       removed

Invariant            No               Yes       -
transform

Non-blind            Yes              Yes       Computational
                                                cost
 Content Based WaterMarking

 This represents the new class of the water
  marking techniques which link the water mark
  with image semantics and not the image
  coordinates.

 The problem of geometrical synchronization is
  solved because the image content represents an
  invariant     reference    to      geometrical
  transformations.
Principle of a content based
  water Marking Scheme
          Previous Work
 Duric

 Sun

 Alghoniemy and tewfik

 Dittman
                 Embedder Side



In the embedder side:
    Detect robust feature points in an image.
    Generate a triangular tessellation of the image based on
    the set of feature points.
    Map a triangular spread sequence (watermark) into each
    triangle of the tessellation via affine transform.
     Add the mapped sequence on each triangle.
                   Detector Side



In the detector side
    Reconstruct the tessellation.
    Map each triangle to the shape of the original triangular
    watermark.
    Compute the correlation of each mapped triangle with the
    original watermark.
    Accumulate the correlations to detect the watermark over
    the whole image.
         Feature Point Detectors
 Harris Detector
 Susan Detector
 Achard Rouquet Detector
Detector Bench Mark

Score=     Nb pre  Nbcre  Nbdes 
                 Nb pre  Nbdes
Basing on the score Harris Detector is choosen.
Embedding Process
•   A random sequence Tw is generated of the shape of a right angle isosceles
    traingle(64 X 64).
•   Improved Harris detector is used for the detection of feature points
           
     P  pi  R 2 , i  1,...., n         
•   A Delaunay tessellation of the set P is the unique triangulaton of the convex cover of
    P such as the interior of the containing circle C  pi , p j , pk of each traingle of P 3 does
    not contain another vertex of P
•   An affine transformation is performed on triangle Tw to map the shape of Tk .The
    affine function A is defined by six real parameters a,b,c,d,e,f.


                        xm  a b  xw   e   xw 
         Axw , yw      
                        y  c d  y    f   L y   T
                                         
                        m        w     w 
• Tmis multiplied by a visual mask            is
                                             which
  based on image luminance.

            Tp i, j     i, j  Tm (i, j )
• The marked triangle        Ts  T p  Tk

• Ts is substituted for Tk
Detector Side
•
•
    A random sequence      T is generated of the shape of a right angle isosceles traingle(64 X 64).
                           wis used for the detection of feature points
    Improved Harris detector
•   A Delaunay tessellation      T   i0  i  N
                                         T ,                        is performed using image

•   feature points. The mark is inserted in each triangle T  T
                                                             k
•   The traingle k is warped into right angled isosceles triangle (64 X 64).
               T                                                    TL
•   TL    is given to a wiener filter and using                           ˆ
                                                  Twstatistics to obtain Tw

                                 Vi Tw i, j 
         Tw i, j  
          ˆ                                                TL i, j   M l TL i, j 
                         Vi Tw i, j   Vl TL i, j 
• Weiner prediction is a denoising operation as it allows separation of
  the image components from the marked components
• The mark is detected inside the traingle only if:
                     corr (Tw ; Tw )   Pfa 
                                 ˆ
• A global decision is obtained using the global sum of                                ˆ
                                                                           corr (Tw ; Tw )

                                             corr T                  
                                     1                             ˆ
• The mean of correlation is Z                         i   w   ; Tw
                                     N
                                                    Pfa 
                                         i 1,...., N

• The mark can be also detected if:      Z
                                                        N

• The final decision is obtained from the different detection results
                   Conclusion
So to summarize the paper:
 The detection of the mark doesnot require the original
  image.
 Using content-based techniques we obtain automatic
  resynchronization of the mark after both local and global
  geometrical transformations.
 The orientation of the signature is carried by the content
  of the image and consequently cannot be erased,
 Content Based techniques donot depend on the
  template insertion which is easily removed.
                Limitations
• The robustness of the scheme is
  dependent on
     The capacity of the feature point detector to
     preserve the feature points after geometrical
     transformations.
     The content of the image. Like highly textured
     images are more error prone.

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:4
posted:9/23/2011
language:English
pages:24