Study with the several resources on Docsity
Earn points by helping other students or get them with a premium plan
Prepare for your exams
Study with the several resources on Docsity
Earn points to download
Earn points by helping other students or get them with a premium plan
This lecture was delivered by Dr. Paresh Sapan at Biju Patnaik University of Technology, Rourkela. This lecture is part of lecture series on Multimedia Applications course. It includes: Basic, Video, Compression, Techniques, Motion, Compensation, Vectors, Consecutive, Frames, Temporal, Redundancy
Typology: Slides
1 / 30
This page cannot be seen from the preview
Don't miss anything!
docsity.com
10.1 Introduction to Video Compression
images.A video consists of a time-ordered sequence of frames, i.e.,
Compressioncoding based on previous frames.An obvious solution to video compression would be predictive
proceeds
by
subtracting
images:
subtract
in
time order and code the residual error.
parts of the image to subtract from the previous frame.It can be done even better by searching for just the right
docsity.co
Motion Compensation
Each image is divided into
macroblocks
of size
N
×
N
(^).
By default,
= 16 for luminance images.
For chrominance images,
= 8 if 4:2:0 chroma subsampling is adopted.
Motion compensation is performed at the macroblock level.
The current image frame is referred to as
Target Frame
match
is
sought
between the
macroblock
in
the
Target
Frame
(referred to asand the most similar macroblock in previous and/or future frame(s)
Reference frame(s)
roblock is called aThe displacement of the reference macroblock to the target mac-
motion vector
Figure 10.1 shows the case of
forward prediction
in which the Refer-
ence frame is taken to be a previous frame.
docsity.co
Target frame
Matched macroblock
Reference frame
Macroblock
Search window
N
N
2 p (^) + 1
( x, y
)
( x, y
)
( x 0 , y
0 (^) )
2 p (^) + 1
MV
( x 0 , y
0 (^) )
Fig. 10.1:
Macroblocks and Motion Vector in Video Compression.
horizontal and vertical displacements in the range [MV search is usually limited to a small immediate neighborhood — both
p, p
This makes a search window of size (
p (^) + 1)
p
docsity.co
Sequential Search
Sequential search
: sequentially search the whole (
p (^) + 1)
(^) ×
(
p
Full Search).
a
macroblock
centered at
each
of
the
positions
within
frame, pixel by pixel, and their respectivethe window is compared to the macroblock in the Target
M AD
is then
derived using Eq. (10.1).
The vector (
i, j
) that offers the least
M AD
is designated
as the
MV
( u, v
) for the macroblock in the Target frame.
vector for a single macroblock is (2absolute value, addition), the cost for obtaining a motionpixel comparison requires three operations (subtraction,sequential search method is very costly — assuming each
p +1)
· (
p
+1)
· N
·^ 3
⇒
O
( p 2 N
(^2)
).
docsity.co
PROCEDURE 10.
Motion-vector:sequential-search
begin
min M AD
/* Initialization */
for
p
to
p
for
j
=
p
to
p
cur M AD
i, j
if
cur M AD < min M AD
min M AD
cur M AD
u
=
i ;
/* Get the coordinates for
v
=
j ;
end
docsity.co
2
1
1 1
1
1 1 1
1
1
2
3 3
3
3 3 3
3
3
2
2
2 2
2
2
MV
( x 0 − (^) p, y
0 − (^) p )
( x 0 − (^) p, y
0 (^) +
(^) p )
( x 0 (^) +
(^) p, y
0 (^) +
(^) p )
( x 0 (^) +
(^) p, y
0 − (^) p )
( x 0 , y
0 (^) ) d p/ 2 e
Fig. 10.2:
2D Logarithmic Search for Motion Vectors.
docsity.co
PROCEDURE 10.
Motion-vector:2D-logarithmic-search
begin
offset =
d (^2) p e ;
they are centered at (Specify nine macroblocks within the search window in the Reference frame,
x 0 , y
0 (^) ) and separated by offset horizontally and/or
while lastvertically;
Find one of the nine specified macroblocks that yields minimum
offset =if offset = 1 then last = TRUE;
d offset
e ;
Form a search region with the new offset and new center found;
end
docsity.co
Downsample
by a factor of 2 Downsample
by a factor of 2
Motion
Estimation Motion
Estimation Motion
Estimation
Motion Vectors
Level 2 Level 1 Level 0
Fig. 10.3:
A Three-level Hierarchical Search for Motion Vectors.
docsity.co
Hierarchical Search
(Cont’d)
Given the estimated motion vector (
u k , v^
(^) k ) at Level
k , a 3
(^) ×
(^3)
neighborhood
centered
at
(
· u
k ,^ (^2)
·
v k )
at
Level
k
−
1
is
searched for the refined motion vector.
The refinement is such that at Level
k
−
(^) 1 the motion vector
( u k −
1 , v
k −
1 ) satisfies:
(
u k − 1 ≤ u k − 1 ≤ 2 u k
,
2 v (^) k
−
1
≤
v (^) k −
1
≤
2 v (^) k
Let (
x 0 k , y
(^0) k (^) ) denote the center of the macroblock at Level
k
in the Target frame.
The procedure for hierarchical motion
vector search for the macroblock centered at (
x 00 , y
(^) ) in the
Target frame can be outlined as follows:
docsity.co
Table 10.
Vector Search based on examplesComparison of Computational Cost of Motion
Search Method
OP S per second
for 720
480 at 30 fps
p
= 15
p
= 7
Sequential search
9
9
2D Logarithmic search
9
9
3-level Hierarchical search
9
9
docsity.co
10.4 H.
H.
:
An relatively early digital video compression stan-
later video compression standards.dard, its principle of MC-based compression is retained in all
encing and other audiovisual services over ISDN.The standard was designed for videophone, video confer-
The video codec supports bit-rates of
p
×
64 kbps, where
p
ranges from 1 to 30. (Hence also known as
p
∗
64).
directional video conferencing.150 msec so that the video can be used for real-time bi-Require that the delay of the video encoder be less than
docsity.co
Table 10.
Video Formats Supported by H.
Video
Luminance
Chrominance
Bit-rate (Mbps)
H.
format
image
image
(if 30 fps and
support
resolution
resolution
uncompressed )
QCIF
176
×
144
88
×
72
required
CIF
352
×
288
176
×
144
optional
docsity.co
I P P P P P P
I
I
Fig. 10.4:
H.261 Frame Sequence.
docsity.co
