Cryptography Reference
In-Depth Information
Frame
Frame
(a) Object movement
(b) Object deformation
Fig. 7.2.
The relationship between WM Imperceptibility and Motion Picture Prop-
erties.
Block-Matching Techniques
Block-matching techniques [14] are based on between-frame matching of
MPEG macro-blocks (1616-pixel blocks) and determine one motion vector
for each macro-block. The following procedure is done for all macro-blocks in
a frame:
Step 1: The luminance set of the f -th frame consisting of WH pixels is
y
(f )
=y
(f )
i,j
1≤i≤W,1≤j≤H. The sum of absolute difference
d(k, l) between the macro-block of current frame y
(f )
and that of
reference frame y
(f−1)
. This is displaced by (k, l) from the block of
y
(f )
, and is calculated using the following equation (see Fig. 7.3):
15
y
(f )
b
x
+i,b
y
+j
−y
(f−1)
b
x
+k+i,b
y
+l+j
d(k, l)=
,
(7.1)
i,j=0
where (b
x
,b
y
) is the pixel location representing the starting point of
the macro-block.
Step 2: The d(k, l)s are calculated within search range k, l. A range of−15≤
k, l≤15 is typically searched [14], and 3131 d(k, l)s are generated.
Step 3: The motion vector v is identified as vector (k
0
,l
0
) for the minimum
of 3131 d(k, l)s. That is, based on the equation