Professional Documents
Culture Documents
1c DVW Motion Estimation
1c DVW Motion Estimation
1c DVW Motion Estimation
0-7803-84O6-7/04/$20.00@ I EEE.
2004
2576
Authorized licensed use limited to: Moradabad Institute of Technology. Downloaded on December 17, 2008 at 02:58 from IEEE Xplore. Restrictions apply.
watennarking bits are directly embedded into the 3. Watermarking method using
niacro blocks of Y or Cb (Cr) component of Chain Cord
I-fmine b!, referring to {lie niotion vector n-lien Fonii the point of view about real-time
P - h “ has a inotion vector. processing by software. 8 directions of motion
This watennarking method is performed by the vectors is represented into tlie octal fonn with
following inanner. simpiicity.
1) When the watermark bit is ‘. l’.. the Y pixel The originally referred macro block of I-fmme is
value of macro block of I-frame that is referring divided into 8 regions on the basis of the direction
to a motion vector is cllanged into an even of motion vectors o f the macro blocks of F-frame
, ,
number. Watermark bits can be embedded into the
2) When the watennark bit is “0”. the Y pixel original inacro block of I-frame by referring to
value of imcro block of I-frame that is refemng motion vectors of P-frame as shown in Fig.2.
N frame N+3 frame
to a niotion vector is changed into an odd number. (Ifmme) ~ (Pfm;~ ~
Direaim of Flas
And also we can embed the waternurk bits into
QLJ
0 PI bll
7 Directicri ofa
positions of watennark bits into the I ~ C I T )block The algorithm of embedding watermark bits by
of I-frame by referring to a motion vector. C h i n Code is described in the following steps
I . Y component (1) Divide the referred macro block A of
2. Cb or Cr component I-frame into 8 regions having 16x16 pixels 011
3. D.C component of the quantized value of Y tlie basis of the direction of niotion vector of
coinponenl after DCT transfonnation the macro block 33 of P-frame.
4. D.Ccomponent uf the quantizcd value of Cb or (2) Embed the 3 bits wateninrks into one of the 8
Cr component after DCT transfonnation regions of inacro block of I-frame and also embed
These embedded I-frames are replaced by the a flag bit to disctiiizinate the embedding position
original I-frames in tlie IBBP buffer and then the of watermark bits in 8 regions very easily.
watermarked out put stream are derived after (3) One of 8 directions for embedding a flag bit is
Variable Lengtll Coder (VLC) is applied to the selected to be a position among (0---7) directions
embedded I-fraiiie. clockwise by the direction of motion vectors.
These iiietliods can be realized in real-time (4)Let the walcnnark bits be W (i) = { O l O l l - - - j
processing using software very easily. As the octal Chain Code for embedding is
represented to be Wzin = { 101}, the watermark bit
is embedded into the “5” region of the original
2577
Authorized licensed use limited to: Moradabad Institute of Technology. Downloaded on December 17, 2008 at 02:58 from IEEE Xplore. Restrictions apply.
macro block of I-frame. The traditional motion estimation method with
(5) The embedding position of ma
co block uses
r tlie optimal block-based motion estimation is
the pisel of the inacro block that is on a constant adopted.[4] In this block based motion estiniation
distance From the center of inacro block and method. the searching area of block matching IS
nearly lies on the line of the "5" direction on the esecuted within f 15 macro blocks.
basis of thi: direction of the motion vector. The watermarking method using Cliain Code
(6) After [he value of tlie specific pisel for the algorithm is evaluated by using AVI video
embedding position is initialized into an even (MPEG-1 format) that has 30 frames/sec.
nuniber. an odd number of a flag bit is embedded 32Ox240pixels, 24 bits color and 192 frames in
when tlie watermark bit has an even number. length. The sparse motion estimation method
Then. the extraction algorithm of the enlbedded adopts one macro block apart in order to avoid the
nag bit having an odd value is described in the superposition of motion vectors.
following steps. The experimental results are shown in Table I.
(1) Find ciut the embedding position of the flag
Edddinp
FWllKM 1 AvmgcSNR AvmgtSNR EwmedPIImFalr
b e f m d s l d m g j Mlwtmbdddng/ d v l l a & i b * l l
2578
Authorized licensed use limited to: Moradabad Institute of Technology. Downloaded on December 17, 2008 at 02:58 from IEEE Xplore. Restrictions apply.
I-frame after DCT transfonnation. After all. the techniques for MPEG video by partly changing
error mate of eh3racted watermark bits became 0%. the construction of the ordinarv MPEG encoder
The actual embedded video fraiiie (AVI fonnat) and decoder by means of software.
is shown in Fig.3 wliich picture qualily is better Final goal is to integrate the watermarking
rilan MF’EGI fonnal tecl~niquesin the distributed video production and
distribution environments as the enabling
technologies for electronic commerce and for
digital market places to ensure copyrights for
videos.
imbhm h e o f l k mb4hdwarnnini;sihl ? L h m oFlh mibrddedvinermprlu
2579
Authorized licensed use limited to: Moradabad Institute of Technology. Downloaded on December 17, 2008 at 02:58 from IEEE Xplore. Restrictions apply.