Image analysis – Image compression or coding – Interframe coding
Reexamination Certificate
1995-12-12
2002-07-02
Chen, Wenpeng (Department: 2624)
Image analysis
Image compression or coding
Interframe coding
C382S238000, C375S240140
Reexamination Certificate
active
06415055
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to a moving image encoding method and apparatus which is suitable in recording or reproducing a moving image signal to or from a recording medium such as a magneto-optical disc or magnetic tape to display it on a display, or in transmitting a moving image signal from a transmitting side to a receiving side by way of a transmission line such as video conference systems, video telephone systems, or equipments for broadcasting. The invention also relates to a moving image decoding method and apparatus for reproducing a moving image signal, from a signal obtained by motion compensation predictive coding.
2. Description of the Related Art
Heretofore, in a case where a moving image is digitalized, recorded or transmitted, coding (compressing) of data has been performed because data size becomes massive. As a representative coding method, there is motion compensation predictive coding.
FIG. 1
shows the principles of the motion compensation predictive coding. The motion compensation predictive coding is a method which makes use of a correlation of the time axis direction of video signals. That is, the motion compensation is a method where the present motion vector of an object to be coded is estimated from a video signal which has already been decoded and reproduced, the decoded and reproduced video signal is moved in accordance with the motion of a signal, and the movement data (motion vector) and a prediction error obtained at that time are transmitted, so as to compress data size necessary for coding. is compressed by transmitting data of this movement (motion vector) and a prediction error obtained at that time.
Moving Picture Expert Group (MPEG) is known as a representative of this motion compensation predictive coding. The MPEG is a popular name of the moving image encoding collected together in the Working Group (WG)
11
of Sub Committee (SC)
29
of Joint Technical Committee (JTC)
1
of ISO and IEC.
In the MPEG, one picture (frame or field) is divided into small units called a macroblock which is constituted by 16 lines ×16 pixels, and motion compensation predictive coding is performed at units of this macroblock. The motion compensation predictive coding is roughly grouped into two methods: intra coding and non-intra coding. The intra coding is a coding method which uses only information of a self-macroblock, and the non-intra coding is a coding method which uses both the information of a self-macroblock and information obtained from a picture which appears at another time.
In the MPEG, each frame picture is coded as any of three kinds of pictures: an intra coded picture (I-picture), a predictive coded picture (P-picture), and a bidirectionally predictive coded picture (B-picture). That is, as shown in
FIGS. 2A and 2B
for example, video signal of 17 frames of frames F
1
to F
17
is considered as a group of pictures (GOP) which is one unit of processing.
As shown in
FIGS. 2A and 2B
, the video signal of first frame F
1
of the GOP is coded as an I-picture, that of second frame F
2
is coded as a B-picture, and that of third frame F
3
is coded as a P-picture, for example. The frames F
4
to F
17
are alternately processed as a B-picture or a P-picture. In
FIGS. 2A and 2B
, an arrow from one picture to another represents a direction of prediction (the same shall apply hereinafter).
For the video signal of the I-picture, the video signal of one frame, as it is, is coded and transmitted. For the video signal of the P-picture, as shown in
FIG. 2A
, basically there is coded and transmitted a difference between the video signal of the P-picture and either of the video signal of the I-picture or the past P-picture being in the past point of time. Also, for the video signal of the B-picture, as shown in
FIG. 2B
, basically there is coded and transmitted a difference between the video signal of the B-picture and either of a frame being in the past point of time and a frame being in the future point of time, or there is coded and transmitted a difference between the video signal of the B-picture and both of a frame being in the past point of time and a frame being in the future point of time.
The principles of a method of coding a moving image signal are shown in
FIGS. 3A and 3B
. As shown in
FIGS. 3A and 3B
, since the first frame F
1
is processed as an I-picture, all of the macroblocks are intra-coded and transmitted as transmission data F
1
X to a transmission line. For the frame F
3
of the P-picture, with the past frame F
1
being in the past point of time as a reference picture, a prediction error (SP
3
) from the frame F
1
is calculated, and transmitted as transmission data F
3
X, together with a motion vector x
3
(forward predictive coding). In this case, the original data of the frame F
3
, as it is, is transmitted as transmission data F
3
X (SP
1
)(intra coding). These methods can be switched at units of a macroblock.
For the frame F
2
of the B-picture, there is calculated a prediction error between the frame F
2
and either or both of the frame Fl being in the past point of time and the frame F
3
being in the future point of time, and this is transmitted as transmission data F
2
X. For the process of this B-picture, there are four kinds of processes at a macroblock unit: (1) intra mode (intra coding), (2) forward predictive mode (forward predictive coding), (3) backward predictive mode (backward predictive coding), and (4) bidirectionally predictive mode (bidirectionally predictive coding).
The process in the intra mode is the process (SP
1
) of transmitting the data of the original frame F
2
, as it is, as transmission data F
2
X, and is the same process as the case of the I-picture. The process in the forward predictive mode is the process of transmitting a prediction error SP
3
obtained from the reference frame F
1
being in the past point of time and also transmitting the motion vector x
1
(motion vector between the frames F
1
and F
2
). The process of the backward predictive mode is the process of calculating a prediction error (SP
2
) with the reference frame F
3
being in the future point of time and transmitting the error (SP
2
) and the motion vector x
2
(motion vector between the frames F
3
and F
2
).
The process in the bidirectionally predictive mode is the process of obtaining a prediction error SP
4
from an average value of two prediction pictures obtained from both of the past reference frame F
1
and the future reference frame F
3
and also transmitting this error as transmission data F
2
X, together with the motion vectors x
1
and x
2
. For the B-picture, the aforementioned four kinds of methods can be switched at units of a macroblock. Among these methods, the processes of the forward predictive mode, the backward predictive mode, and the bidirectionally predictive coding are a non-intra coding method.
The moving image encoding apparatus should select a method whose coding efficiency is best among the aforementioned four modes, in coding the macroblock of the B-picture. Ideally, it is desirable that the macroblock be coded with four kinds of methods and then a method where the size of transmission data is least be selected. However, this method has the problem that the scale of the hardware becomes large.
As a method for solving this problem, there has been proposed a method in U.S. patent Ser. No. 08/123560, filed on Sep. 17, 1993, now (U.S. Pat. No. 5,461,420, issued on Dec. 24, 1995) in the process (motion estimation: ME) of estimating forward and backward motion vectors of a macroblock, forward and backward motion vector estimation errors (ME errors) are obtained and, based on these values, the non-intra predictive coding of the macroblock is selected.
The motion vector estimation error is obtained by, for example, calculating the sum of absolute value of a difference in each pixel between a prediction macroblock obtained from the motion vector and a macroblock of an object to be coded. This motion vector estimation error is obtained for both the fo
Chen Wenpeng
Frommer William S.
Frommer & Lawrence & Haug LLP
Polito Bruno
Sony Corporation
LandOfFree
Moving image encoding method and apparatus, and moving 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 Moving image encoding method and apparatus, and moving image..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Moving image encoding method and apparatus, and moving image... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2869431