Pulse or digital communications – Bandwidth reduction or expansion – Television or motion video signal
Reexamination Certificate
2001-11-13
2002-11-19
Britton, Howard (Department: 2613)
Pulse or digital communications
Bandwidth reduction or expansion
Television or motion video signal
C348S699000
Reexamination Certificate
active
06483877
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to a method of coding and decoding an image by applying global motion compensation to the whole image based on linear interpolation and/or extrapolation or bilinear interpolation and/or extrapolation.
BACKGROUND OF THE INVENTION
In the highly efficient coding of a dynamic image, it has been known, in recognition of the similarity of the frames that are close to each other in regard to time, to use motion compensation in compressing the data. The most widely used motion compensation system at present image coding technology is block matching, employed in Standards H.261, MPEG1 and MPEG2 which are international standards for a dynamic image coding system. According to this system, the image to be coded is divided into a number of blocks, and a motion vector is found for each of the blocks.
FIG. 1
illustrates the constitution of a coder
100
of the H.261 Standard which employs a hybrid coding system (adaptive interframe/intraframe coding method) which is a combination of block matching and DCT (discrete cosine transform). A subtractor
102
calculates the difference between an input image (original image of present frame)
101
and an output image
113
(that will be described later) of an interframe/intraframe switching unit
119
, and outputs an error image
103
. The error image is transformed into a DCT coefficient through a DCT processor
104
and is quantized through a quantizer
105
to obtain a quantized DCT coefficient
106
. The quantized DCT coefficient is output as transfer data onto a communication line and is, at the same time, used in the coder to synthesize an interframe predicted image. A procedure for synthesizing the predicted image will be described below. The quantized DCT coefficient
106
passes through a dequantizer
108
and an inverse DCT processor
109
to form a reconstructed error image
110
(the same image as the error image reproduced on the receiving side).
An output image
113
(that will be described later) of the interframe/intraframe switching unit
119
is added thereto through an adder
111
, thereby to obtain a reconstructed image
112
of the present frame (the same image as the reconstructed image of the present frame reproduced on the receiving side). The image is temporarily stored in a frame memory
114
and is delayed in time by one frame. At the present moment, therefore, the frame memory
114
is outputting a reconstructed image
115
of the preceding frame. The reconstructed image of the preceding frame and the input image
101
of the present frame are input to a block matching unit
116
where block matching is executed.
In the block matching, an image is divided into a plurality of blocks, and a portion most resembling the original image of the present frame is taken out for each of the blocks from the reconstructed image of the preceding frame, thereby synthesizing a predicted image
117
of the present frame. At this moment, it is necessary to execute a processing (local motion estimation) for detecting how much the blocks have moved from the preceding frame to the present frame. The motion vectors of the blocks detected by the motion estimation are transmitted to the receiving side as motion data
120
. From the motion data and the reconstructed image of the preceding frame, the receiving side can synthesize an estimated image which is the same as the one that is obtained independently on the transmitting side.
Referring again to
FIG. 1
, the estimated image
117
is input together with a “0” signal
118
to the interframe/intraframe switching unit
119
. Upon selecting either of the two inputs, the switching unit switches the coding either the interframe coding or the intraframe coding. When the predicted image
117
is selected (
FIG. 2
illustrates this case), the interframe coding is executed. When the “0” signal is selected, on the other hand, the input image is directly DCT-coded and is output to the communication line. Therefore, the intraframe coding is executed.
In order to properly obtain the reconstructed image on the receiving side, it becomes necessary to know whether the interframe coding is executed or the intraframe coding is executed on the transmitting side. For this purpose, a distinction flag
121
is output to the communication line. The final H.261 coded bit stream
123
is obtained by multiplexing the quantized DCT coefficient, motion vector, and interframe/intraframe distinction flag into multiplexed data in a multiplexer
122
.
FIG. 2
illustrates the constitution of a decoder
200
for receiving a coded bit stream output from the coder of FIG.
1
. The H.261 bit stream
217
that is received is separated through a separator
216
into a quantized DCT coefficient
201
, a motion vector
202
, and an intraframe/interframe distinction flag
203
. The quantized DCT coefficient
201
is decoded into an error image
206
through a dequantizer
204
and an inverse DCT processor
205
. To the error image is added an output image
215
of an interframe/intraframe switching unit
214
through an adder
207
to form a reconstructed image
208
.
The interframe/intraframe switching unit switches the output according to the interframe/intraframe coding distinction flag
203
. A predicted image
212
that is used for executing the interframe coding is synthesized by a predicted image synthesizer
211
. Here, the decoded image
210
of the preceding frame stored in the frame memory
209
is subjected to a processing of moving the position of each of the blocks according to the motion vector
202
that is received. In the case of intraframe coding, on the other hand, the interframe/intraframe switching unit outputs the “0” signal
213
.
Block matching is a motion compensation system that is now most widely utilized. When the whole image is expanding, contracting, or turning, however, the motion vectors of all of the blocks must be transmitted, causing a problem of low coding efficiency. To solve this problem, global motion compensation (e.g., M. Hotter, “Differential Estimation of the Global Motion Parameters Zoom and Pan”, Signal Processing, Vol. 16, No. 3, pp. 249-265, Mar., 1989) has been proposed to express the motion vector field of the whole image while not using many parameters. According to this motion compensation system, the motion vector (ug(x, y), vg(x, y)) of a pixel (x, y) in an image is expressed in the form of:
u
g
(
x, y
)=
a
0
x+a
1
y+a
2
v
g
(
x, y
)=
a
3
x+a
4
y=a
5
. . . Equation 1
or
u
g
(
x, y
)=
b
0
xy+b
1
x=b
2
y+b
3
v
g
(
x, y
)=
b
4
xy+b
5
x+b
6
y+b
7
. . . Equation 2
and the motion compensation is executed using the motion vectors. In these equations, a0 to a5 and b0 to b7 are motion parameters. In executing the motion compensation, the same predicted image must be generated both on the transmitting side and on the receiving side. For this purpose, the transmitting side may directly transmit values of a0 to a5 or b0 to b7 to the receiving side or may instead transmit motion vectors of several representative points.
As shown in
FIG. 3A
, assume that the coordinates of the pixels at the left upper, right upper, left lower and right lower corners of an image
301
are expressed by (0, 0) , (r, 0), (0, s) and (r, s) (where r and s are positive integers). Here, letting the horizontal and vertical components of the motion vectors of the representative points (0, 0), (r, 0) and (0, s) be (ua, va), (ub, vb) and (uc, vc), respectively, Equation 1 is rewritten as:
u
g
(
x
,
y
)
=
u
b
-
u
a
r
x
+
u
c
-
u
a
s
y
+
u
a
v
g
(
x
,
y
)
=
v
b
-
v
a
r
x
+
v
c
-
v
a
s
y
+
v
a
Equation
3
This means that the same function can be fulfilled even when ua, va, ub, vb, uc and vc are transmitted instead of transmitting a0 to a5. This state is shown in
FIGS. 3A and 3B
. The motion vectors
306
,
307
and
308
(the motion vectors are defined to start from points of the original image
LandOfFree
Method of coding and decoding image does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method of coding and decoding image, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method of coding and decoding image will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2932574