Pulse or digital communications – Bandwidth reduction or expansion – Television or motion video signal
Reexamination Certificate
1999-06-08
2001-10-16
Rao, Andy (Department: 2613)
Pulse or digital communications
Bandwidth reduction or expansion
Television or motion video signal
C375S240090
Reexamination Certificate
active
06304602
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates generally to a method and a system for processing a set of data on motion pictures and particularly to a method and a system for processing a set of data on a temporal sequence of two or three dimensional pictures with a time-dependent motion intermittently picked up therein, by using measures for a compensation of the motion in combination with an image segmentation of the pictures.
2. Description of the Related Art
Recent years have observed an increasing interest in a processing of a set of digital data on motion pictures, in particular in a field of art relating to a compression coding of such data.
The motion pictures appear as a temporal sequence of pictures each having an image region associated with a transient state of a motion. The transient state in an arbitrary picture has a mathematically analyzable correlation to that in a previous picture, so that the former is predictable from the latter in combination with an estimated correlation therebetween or by compensating the latter therewith, subject to the adequacy of an employed model of the motion for analyzing the correlation.
A predicted current state is comparable with a sampled current state to determine a difference therebetween as a prediction error. The prediction error is encodable together with associated parameter values of the model to obtain a compressed code, which permits a current state of the motion to be calculated in combination with a past calculated state thereof at a decoding end.
An adequate model provides an effective prediction to achieve a significant reduction in redundancy of the code.
A typical modelling is based on a motion compensation interframe prediction, in which a motion is defined in terms of a displacement of a block of a picked up image between a pair of frames each corresponding to a picture. Typically, the picture is divided into a set of square blocks each having an area about an order of 16×16 pixels. A total number of blocks in any picture is identical for this purpose, and respective blocks in an arbitrary picture have a one-to-one correspondence to those blocks in any other picture. With respect to each block, a past decoded picture is compensated by an estimated motion, to predict a picture to be compared with a current input picture. Thus, motion compensation is made in blocks, together with associated pixel data as they are inherent.
FIG. 1
exemplarily shows an encoder of a conventional motion picture coding-decoding system using the motion compensation interframe prediction. This conventional system is well known by the ITU-T(CCITT) Recommendations H.261, as a video coded for audiovisual services at p×64 kbit/s.
In
FIG. 1
, designated at reference character
100
is an entirety of the conventional system,
100
a
is the encoder,
101
is an input terminal of the encoder
100
a
, and
102
and
109
are frame memories, respectively.
The input terminal
101
inputs a sequence of pixel data D
101
of a current picture P
101
. The frame memory
102
stores therein the data D
101
of the current input picture P
101
. besides a set of data D
100
of a past input picture P
100
it received from the input terminal
101
and stored therein in a last frame. The frame memory
109
has stored therein a set of data D
102
of a local decoded picture P
102
that is a matrix of restored pixel data of the past input picture P
100
.
The data D
100
of the past input picture P
100
is sequentially read from the frame memory
102
, and the data D
102
of the local decoded picture P
102
from the frame memory
109
. They are either selected by a switch
103
a
, as a sequence of data D
103
representing a reference picture P
103
(to be P
100
or P
102
), and input to a motion estimator
103
, which concurrently receives the data D
101
of the current input picture P
101
.
Incidentally, as shown in
FIG. 2
, the conventional system
100
has, as a common field Fc to a variety of associated computations, an imaginary orthogonal coordinate system defined by a combination of an axis of abscissa X corresponding to a bottom side of a picture frame Fp and an axis of ordinate Y corresponding to a left lateral side of the picture frame Fp. The picture frame Fp, as well as any block B (&bgr;/&agr;) therein and a geometrical gravity center G (&bgr;/&agr;) thereof, is congruently mapped in the imaginary field Fc each time a set of associated data is processed in the system
100
, where “&agr;” is a picture identification number and “&bgr;” is a block identification number.
As illustrated in
FIG. 2
, the motion estimator
103
estimates, by calculation for each block B(i/
101
) (i=an arbitary integer) of the current input picture P
101
, a motion as a displacement vector Vd
i
in terms of a combination of a sense and a magnitude of a displacement of a gravity center G(i/
101
) of an i-th block B(i/
101
) in the current input picture P
101
relative to that G(i/
103
) of a corresponding i-th block B(i/
103
) in the reference picture P
103
, thereby obtaining a set of data D
104
of a set Vm of motion vectors of which an i-th one Vm
i
consists of an X-component Vx
i
as a projection of the displacement vector Vd
i
to the axis X and a Y-component Vy
i
as that to the axis Y.
The data D
104
of the motion vector set Vm is sequentially output from the motion estimator
103
to a motion compensator
104
and an encoding multiplexer
110
.
The multiplexer
110
encodes the data D
104
of the vector set Vm into a sequence of corresponding codes.
At the motion compensator
104
the data D
104
of the motion vector set Vm are processed together with the data D
102
of the local decoded picture P
102
input from the frame memory
109
so that a gravity center B(i/
102
) of each block B(i/
102
) in the local decoded picture P
102
has a position thereof displaced or coordinate-component wise compensated along the axes X and Y by equivalent distances to X- and Y-components Vx
i
and Vy
i
of a corresponding motion vector Vm
i
, respectively, to thereby obtain a motion-compensated picture P
104
as an interframe predicted one for the current input picture P
101
.
As a result, each pixel data in each displaced block in the motion-compensated picture P
104
is updated by data that a corresponding pixel in that block in the local decoded picture P
102
had been carrying.
The motion compensator
104
sequentially outputs a set of data D
105
of the motion-compensated picture P
104
to a subtractor
105
and an adder
108
.
The subtractor
105
performs a pixel-mode subtraction of the motion-compensated picture P
104
from the current input picture P
101
, obtaining a set of data D
106
representative of a differential picture P
105
therebetween, i.e. a picture having elementwise distributed thereon a matrix of prediction errors due to a motion compensation in a 16×16-pixel block mode by the compensator
104
. Accordingly, the data D
101
of the current input picture P
101
is converted into a compressed set of data as the data D
106
representing the prediction errors.
The prediction error data D
106
of the differential picture P
105
is sequentially input from the subtractor
105
to a unit
106
adapted for a discrete cosine transformation and quantization (hereafter “DCT-Q”) process, i.e. for a data compression. The data is mapped in an 8×8-pixel block mode from a real measure space through a discrete cosine transform function into a related frequency field, to be expressed in terms of a combination of cosine coefficients of a corresponding cosine series to an associated 8×8-pixel block. The cosine coefficients are then quantized.
As a result, the error data D
106
is further compressed into a set of combinations of data D
107
each representative of a quantized coefficient, so that the set of data D
107
represents the differential picture P
105
.
The compressed data D
107
of the differential picture P
105
is sequentially output from the DCT-Q unit
106
to the encoding multiplexer
110
NEC Corporation
Ostrolenk Faber Gerb & Soffen, LLP
Rao Andy
LandOfFree
Method and system for processing data on motion pictures by... 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 and system for processing data on motion pictures by..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and system for processing data on motion pictures by... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2614123