Image analysis – Image compression or coding – Predictive coding
Reexamination Certificate
1998-12-08
2002-03-05
Couso, Jose L. (Department: 2724)
Image analysis
Image compression or coding
Predictive coding
C382S236000, C382S239000, C348S699000
Reexamination Certificate
active
06353683
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to a method and an apparatus of image processing, and a data storage medium and, more particularly to a method and an apparatus which are capable of coding or decoding an image signal without damaging a regenerated video even if a system which processes an image signal has low processing ability, and a data storage medium which contains a program for implementing coding and decoding the image signal by the method and the apparatus by software.
BACKGROUND OF THE INVENTION
In a prior art coding method for a digital moving image, since temporal correlation, i.e., correlation of image information between frames, is high in the moving image, a difference value between an image signal of a target frame to be coded and a prediction signal obtained from an image signal of a previous frame which has been coded (before the target frame is coded) is coded, in order to eliminate redundant image information between frames and thereby reduce temporal redundancy.
This coding process is called a “differential coding process”,and is performed for each of unit areas into which one frame is divided. The frame corresponds to one frame according to MPEG (Moving Picture Experts Group) 1 and 2, and corresponds to an object area, i.e., an image display space corresponding to an object image according to MPEG4.
To be specific, the differential coding process detects a prediction area in the previous frame which comprises an image signal having the smallest error with respect to an image signal (input image signal) of a target unit area to be coded, by motion prediction using the image information of the previous frame, and performs coding to a difference signal corresponding to a difference value between the image signal of the target unit area and the prediction signal (an image signal in the prediction area) for the target unit area along with prediction relevant information associated with the motion prediction. The prediction relevant information is, for example, information indicating a position of the prediction area, namely, a motion vector.
Meanwhile, a differential decoding process regenerates an image signal based on a coded difference signal generated by the differential coding process. The differential decoding process uses motion compensation to predict a regenerated image signal (decoded image signal) for a target unit area to be decoded and thereby generate a prediction signal (decoded prediction image signal) according to the prediction relevant information, decodes the coded difference signal of the target unit area to generate a decoded difference signal, and adds the prediction signal to the decoded difference signal to generate a regenerated image signal corresponding to the target unit area.
As described above, in the conventional process for coding or decoding the image signal, techniques for compressing image data such as motion prediction or motion compensation have been employed. A specific example of implementation of these is data compression according to MPEG1 or MPEG2 as an international standard.
Attempts are being made to extend standardization of techniques according to MPEG4 being standardized, by introducing new motion prediction or motion compensation into them, in addition to the motion prediction and motion compensation according to MPEG1 or MPEG2.
The motion prediction detects the information indicating the position of the prediction area, i.e., the motion vector, in the coding process. The motion compensation generates the prediction signal according to the motion vector in the coding process or the decoding process.
One of such introduction is overlapping motion compensation. Hereinafter, a description will be given of normal motion compensation and overlapping motion compensation by an image processing technique according to MPEG1 and MPEG2. For the sake of simplicity, a luminance signal is shown as a signal to be subjected to motion compensation. The overlapping motion compensation is commonly performed in the decoding process, including a local decoding process in the coding process, and therefore the decoding process will be discussed.
In a prior art decoding process according to MPEG1 and MPEG2, the motion compensation is performed for each unit area corresponding to a two-dimensional image space (macroblock) comprising 256 pixels (16×16 pixels).
FIGS.
7
(
a
)-
7
(
d
) show motion compensation in the decoding process.
Turning now to FIG.
7
(
a
), for a target macroblock MB(n) in a target frame F(n) being decoded at a current time t(n), a motion vector MV(n) is used to find a prediction area PR(n) in a previous frame F(n-
1
) which has been decoded at a past time t(n-
1
) The target macroblock MB(n) and the corresponding macroblock MB(n-
1
) are present in the same position in the target frame F(n) and the previous frame F(n-
1
), respectively. The prediction area PR(n) is apart from the corresponding macroblock MB(n-
1
) by a displacement indicated by the motion vector MV(n), in the previous frame F(n-
1
)
Then, an image signal included in the prediction area PR(n) is found as a prediction signal (decoded prediction image signal) for a decoded image signal of the target macroblock MB(n).
FIGS.
7
(
b
)),
7
(
c
), and (
d
), show the macroblock MB(n-
1
), the prediction area PR(n), and the target macroblock MB(n) in FIG.
7
(
a
), respectively.
The overlapping motion compensation is performed for each unit area, i.e., a two-dimensional image space (block) comprising 64 pixels (8×8 pixels). 4 of these blocks correspond to one macroblock. In other words, one macroblock is composed of (2×2) blocks, as shown in FIG.
8
.
In the overlapping motion compensation, the motion vector is present for each block or for each macroblock. Assume that motion vectors for the 4 blocks in the macroblock are the same when the motion vector is for the macroblock.
A description will be given of overlapping motion compensation with reference to FIG.
9
.
Herein, a description will be given of generation of a prediction signal (a decoded prediction image signal) of the target block EC(n) in the target frame F(n).
Initially, by the use of the motion vector MVC of the target block BC(n), motion vectors MVU, MVD, MVR, and MVL of neighboring blocks BU(n), BD(n), BR(n), and BL(n) which are positioned above, below, right, and left with respect to the block BC(n), 5 prediction areas of the target block BC(n) are found by using the corresponding block BC(n-
1
) in the previous frame F(n-
1
) as a criterion. The motion vectors MVU, MVD, MVR, and MVL indicate areas PU(n), PD(n), PR(n), and PL(n) in the previous frame F(n-
1
) as the prediction areas of the blocks BU(n), BD(n), BR(n), and BL(n) in the previous frame F(n-
1
), and therefore the 4 prediction areas of the target block BC(n) found by the use of these motion vectors are, as shown in FIG.
10
(
a
), areas PU(n)′, PD(n)′, PR(n)′, and PL(n)′. In addition, by the use of the motion vector MVC of the target block BC(n), the prediction area PC(n) is found for the target block BC(n).
Subsequently, from pixels in the 5 prediction areas PC(n), PU(n)′, PD(n)′, PR(n)′ and PL(n)′, prediction images PGC, PGU, PGD, PGR, and PGL in FIG.
10
(
b
) are found. It should be remembered that unnecessary pixels in respective prediction areas are not used. To be specific, the prediction images of the prediction area PU(n)′, the prediction area PD(n)′, the prediction area PR(n)′, and the prediction area PL(n)′ are formed from pixels in the upper portion, pixels in the lower portion, pixels in the right portion, and pixels in the left portion, respectively.
Then, values of pixels in the prediction images, PGC, PGU, PGD, PGR, and PGL, are weighted by values shown by weighting matrixes WVC, WVU, WVD, WVR, and WVL(see FIG.
10
(
c
)). Weighting values shown in respective weighting matrixes are added for each pixel, to obtain “8”. Pixel values Pg of a composite image PG is found according to the following expression:
Burr & Brown
Couso Jose L.
Do Anh Hong
LandOfFree
Method and apparatus of image processing, and data storage... 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 apparatus of image processing, and data storage..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus of image processing, and data storage... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2878458