Pulse or digital communications – Bandwidth reduction or expansion – Television or motion video signal
Reexamination Certificate
1998-07-28
2001-03-20
Kelley, Chris S. (Department: 2713)
Pulse or digital communications
Bandwidth reduction or expansion
Television or motion video signal
C375S240170
Reexamination Certificate
active
06205176
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to a motion-compensated coder, a decoder, a method of motion-compensated coding, and a method of decoding.
2. Description of the Prior Art
A motion compensated coder for motion-compensated-coding video data is known and a decoder for decoding the motion-compensated-coded video data is also known.
The motion compensation is generally effected every unit of 16×16 pixels to 8×8 pixels. The motion of an image, i.e., the motion vector is obtained at the same unit. An accuracy of the motion vector(MV) is generally one or a half pixel. If the motion compensation is used in coding, the motion vector is coded and sent to the decoder side.
FIG. 7
is a block diagram of a prior art motion-compensated coder.
Input video data is supplied to a subtractor
2
and to a motion estimator
71
through an input terminal
1
. The subtractor
2
subtracts a motion-compensated-predicted signal
10
a
from the input video data and supplies the difference as a prediction error to a DCT (Discrete Cosine Transform)
3
.
The DCT
3
effects a discrete cosine transforming process at 8×8 pixels and supplies the obtained coefficients to a quantizer
4
. The quantizer
4
quantizes the coefficients with predetermined step amount and outputs a fixed length coded coefficients to the variable length coder
5
and to an inverse quantizer
9
.
Generally, the quantizing step amount is controlled in accordance with an amount of codes to keep the data rate constant.
The variable length coder
5
converts coefficients of two dimensional 8×8 pixels into one dimensional array through zigzag scanning and codes the coefficients by Huffman coding. This inter-frame prediction error signal is multiplexed with coded motion vectors by a multiplexer
14
.
On the other hand, inverse processes of the DCT
3
and the quantizer
4
are executed by the inverse quantizer
9
, an inverse DCT
13
to reproduce the inter-frame prediction error. The reproduced prediction error is added to motion-compensated-predicted signal
10
a
to provide a reproduced video data which is stored in a video memory
73
.
Reproduced video from the video memory
73
is supplied to the motion estimator
71
and to a motion-compensated predictor
72
. The processing from the inverse quantizer
9
to the video memory
73
is called as local decoding which is essentially the same processing of the corresponding decoder.
The motion-compensated predictor
72
shifts video data stored in the video memory
73
every block in accordance with the motion vectors from the motion estimator
71
to obtain a motion-compensated-predicted signal
10
a
which is supplied to the subtractor
2
and to the adder
12
.
The motion estimator
71
effects block matching between the reproduced video data stored in the video memory
73
and the input video data with the reproduced video data shifted every motion compensation block and determines the motion vector MV showing best matching (lowest error). The obtained motion vector is supplied to a motion vector coder
74
for coding the motion vector and to the motion-compensated predictor
72
.
The motion vector coder
74
obtains a difference between horizontal components of the motion vectors at the one block previous block (generally left) and the present block and a difference between vertical components of the motion vectors at the one block previous block (generally left) and the present block and codes the difference values with Huffman codes to supply the obtained code train (bit stream) of the motion vector to the multiplexer
14
which multiples the code train (bit stream) of the motion vector with the code train (bit stream) of the inter-frame prediction error signal, i.e., the output of the variable length coder
5
to output coded signal
15
.
FIG. 8
is a block diagram of a prior art decoder corresponding to the motion-compensated prediction coder shown in FIG.
7
.
The coded signal
15
is inputted from an input
21
and supplied to a separator
22
which separates the coded signal into the code train of the inter-frame prediction error and the code train of the motion vectors. The inter-frame prediction error is converted into the fixed length code by a variable length decoder
23
which outputs 8×8 pixels of coefficients which are supplied to an inverse quantizer
9
. The inverse quantizer
9
and an inverse DCT
13
outputs reproduced prediction errors. On the other hand, the code train of the motion vectors is supplied to a motion vector decoder
81
which decodes the code train of the motion vectors and supplies the obtained motion vector data to a motion-compensated predictor
82
. The motion-compensated predictor
82
generates inter-frame prediction signal with the video data stored in a video memory
83
motion-compensated in accordance with the motion vector data from the motion vector decoder
81
. The adder
12
adds the inter-frame prediction error signal to the reproduced prediction errors to output reproduced video data
124
which is stored in the video memory
83
. The adder
12
, the inverse quantizer
9
, and the inverse DCT
13
in
FIG. 8
have the same structure as those shown in
FIG. 7
respectively.
In these prior art motion-compensated prediction coder and motion-compensated prediction decoder, the accuracy of the motion vector is fixed. Therefore, in the case of the image showing a low self-correlation (there is a lot of amounts of high frequency component), it is possible to make the prediction error low by the high accuracy motion prediction. However, in the case of the image showing high self-correlation (there is a few amount of high frequency component) the high accuracy motion compensation prediction does not contribute to reduction in the prediction error, so that the motion vector data is not efficiently used.
SUMMARY OF THE INVENTION
The aim of the present invention is to provide a superior motion-compensated coder, a superior decoder, a superior method of motion-compensated coding and a superior method of decoding.
According to the present invention, a motion-compensated predictive coding apparatus is provided, which includes: a motion vector estimation circuit responsive to video data and local decoded video data for obtaining N kinds of motion vectors at N different accuracies every first block of pixels of the video data respectively; a vector combining circuit responsive to the motion vector estimation circuit for combining the N kinds of motion vectors at every second block including M of the first blocks every the N different accuracies, M and N being natural numbers, N being more than one; a motion vector accuracy selection circuit responsive to the vector combining circuit, the video data, and the local decoded video data for obtaining N total amounts of code of the N kinds of motion vectors and motion-compensated-predictive-coded video data at the N different accuracies, selecting one of the N different accuracies showing the lowest one of the N total amounts, generating accuracy data indicative of one of the N different accuracies, and outputting the motion vectors corresponding to the selected one of the N different accuracies and the accuracy data; and a motion-compensated predictive coding circuit responsive to the video data and motion vector accuracy selection circuit for motion-compensated-predictive-coding the video data using the outputted motion vectors at the selected one of the N different accuracies, generating the local decoded video data, and outputting the motion-compensated-predictive-coded video data, the outputted motion vectors, and the accuracy data.
In the motion-compensated predictive coding apparatus, the motion vector accuracy selection circuit may include N first amount detection circuits for coding the motion vectors at every second block and thereby, obtaining the N amounts of code of the N kinds of motion vectors at every N different accuracies; N second amount detection circuits responsive to the video data, the local decoded video data
Gopstein Israel
Kelley Chris S.
Philippe Gims
Victor Company of Japan Ltd.
LandOfFree
Motion-compensated coder with motion vector accuracy... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Motion-compensated coder with motion vector accuracy..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Motion-compensated coder with motion vector accuracy... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2451726