Image analysis – Image compression or coding – Interframe coding
Reexamination Certificate
1999-07-28
2003-07-08
Johnson, Timothy M. (Department: 2625)
Image analysis
Image compression or coding
Interframe coding
C375S240130, C382S238000
Reexamination Certificate
active
06591015
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to video coding method and apparatus and motion vector estimator for use in recording or transmitting a video signal.
In general, a current frame of a moving picture has high correlation with the previous frame thereof unless scenes are changed between them. Accordingly, in coding a moving picture, a motion-compensated inter-frame coding technique is ordinarily employed, i.e., a current frame is predicted and coded using a pair of frames preceding and succeeding the current frame.
Since inter-frame prediction always requires information about other frames, it is impossible to randomly access a desired frame. Thus, an intra-frame coding technique, i.e., coding a current frame using only the information about the frame itself, is also employed periodically. For example, according to the MEG (moving picture experts group) standards, a unit of intra-frame coding is called a “group of pictures (GOP)”.
In coding a moving picture, a target frame is divided into a plurality of blocks, each consisting of a multiplicity of pixels neighboring each other, and each of these blocks divided is coded independently. That is to say, coding is performed on a block-by-block basis. For instance, in performing predictive coding, a block with the highest correlation with the target block is extracted as a predicted block from a reference frame preceding or succeeding the target frame. A predicted error block, which represents a difference between the target block and the predicted block, is quantizes and coded. When coding is performed using a reference frame preceding a target frame, the coding technique is called “forward predictive coding”. Conversely, when coding is performed using a reference frame succeeding the target frame, the coding technique is called “backward predictive coding”. And when coding is performed using a frame representing an average between the pair of frames preceding and succeeding the target frame as a reference frame, the coding technique is called “bidirectional predictive coding”.
Since a predicted block should preferably be most similar to the target block, a block with the highest correlation with the target block is extracted as the predicted block from the blocks located within a search range in the reference frame. Specifically, differential blocks are obtained based on respective differences in luminance signal level between the target block and candidate blocks within the search range in the reference frame. By calculating the sum of absolute or squared pixel values within each of these differential blocks, one of the candidate blocks with a differential block having the smallest sum of absolute or squared pixel values is extracted as the predicted block.
In a moving picture, however, the location of the predicted block within the reference frame is different from that of the target block within the target frame. The direction and quantity representing this position al difference is called a “motion vector”, which is also coded along with the predicted error block and used in decoding. Such a technique is called “motion compensation”.
According to the MPEG standards, pictures are classified based on the coding mode thereof into the three types of: I-pictures, P-pictures and B-pictures. An I-picture is composed of nothing but intra blocks, in which only the information within the frame is coded without performing prediction. A P-picture is composed of forward-predicted blocks and intra blocks. And a B-picture is composed of forward-predicted blocks, backward-predicted blocks, bidirectional-predicted blocks and intra blocks.
Also, when inter-frame prediction is performed, frame prediction can be adaptively switched into field prediction, and vice versa, based on the state of the picture to increase the predictive efficiency. For example, an interlaced picture may sometimes be coded more efficiently by the field prediction technique.
As can be seen, a plurality of coding modes coexist according to the MPEG standards.
FIG. 16
is a block diagram illustrating a configuration of a conventional video coder. This video coder is adapted to select one of a plurality of coding modes and to code a target block based on the coding mode selected. In the following example, selectable coding modes are supposed to include forward prediction mode and intra-frame coding mode.
As shown in
FIG. 16
, the video coder includes: block divider
10
; encoding section
40
; decoding section
50
; frame memory section
60
; predicted block generator
70
; and coding mode determining section
1600
.
The coding mode determining section
1600
includes: a sum-of-squared-differences calculator
1601
; a variance calculator
204
; and a coding mode determiner
205
.
The predicted block generator
70
receives a reference frame, which has been output from the frame memory section
60
, and a target block S
k
, which has been output from the block divider
10
. The generator
70
extracts a block with the highest correlation with the target block S
k
from the reference frame, thereby outputting the extracted block as a forward-predicted block P
k
.
In the coding mode determining section
1600
, the variance calculator
204
calculates a variance V
k
of the target block S
k
and outputs the variance to the coding mode determiner
205
. Receiving the target block S
k
and forward-predicted block P
k
, The sum-of-squared-differences calculator
1601
obtains a predicted error Ep
k
, that is, a sum of absolute or squared differences between the blocks of these two types, and outputs the predicted error Ep
k
to the coding mode determiner
205
.
Comparing the variance V
k
to the predicted error Ep
k
, the coding mode determiner
205
selects one of the coding modes allowed by the picture type specified, and outputs the coding mode selected to the encoding and decoding sections
40
and
50
. For example, if an I-picture has been specified, then the determiner
205
determines the coding mode as intra-frame coding.
FIG. 17
illustrates how the coding mode is determined by a test model (TM) technique according to the MPEG standards. If a P-picture, requiring prediction, has been specified, then the coding mode is determined as shown in FIG.
17
.
In this manner, the conventional video coder obtains the variance of the target block and the predicted error of the predicted block, and selects one of a plurality of coding modes based on these values.
The method for compensating for a motion based on the magnitude of a predicted error in luminance signal level between blocks supposes that brightness conditions are the same between a target frame and a reference frame. In other words, this method supposes that an actually associated object on these frames is presented at substantially the same luminance signal level. However, in several types of pictures, such as a fade-in picture gradually brightens on the whole screen, a fade-out picture gradually darkening and a flash picture on which the brightness of the frame changes instantaneously, the luminance changes substantially uniformly over the entire screen. Accordingly, even if an object on a target frame is actually associated with an object on the reference frame, these objects are presented at mutually different luminance signal levels. Thus, in such a situation, the motion cannot be compensated for properly and an appropriate coding mode cannot be selected, either.
Next, the features of fade-in and fade-out pictures (in this specification, these pictures will be collectively referred to as “fading pictures”) will be described in greater detail.
FIG. 18
illustrates a variation in luminance signal level between a pixel located on a line within a frame (i.e., the 65
th
frame in the illustrated example) of a quasi-still fade-out picture with almost no motion and a corresponding pixel located on the same line within a frame (i.e., the 70
th
frame in the illustrated example) appearing later than the former frame. As shown in
FIG. 18
, although almost no motion happens in the fade-out picture,
Fukuda Hideki
Yasunari Tomoko
Cole Thomas W.
Johnson Timothy M.
Matsushita Electric - Industrial Co., Ltd.
Nixon & Peabody LLP
LandOfFree
Video coding method and apparatus with motion compensation... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Video coding method and apparatus with motion compensation..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Video coding method and apparatus with motion compensation... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3099100