Image analysis – Image compression or coding – Interframe coding
Reexamination Certificate
1999-12-15
2004-05-04
Chen, Wenpeng (Department: 2624)
Image analysis
Image compression or coding
Interframe coding
C382S253000
Reexamination Certificate
active
06731810
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to a method and an apparatus for coding a moving image and to a recording medium for recording a program of coding a moving image, and more particularly to a method and an apparatus for high rate compression coding of moving image data in TV, animation, color-graphic game and so on, and further, to a recording medium for recording a program of coding used for the same.
BACKGROUND OF THE INVENTION
Heretofore, a system for inter-frame coding with movement compensation (movement-compensated inter-frame coding) has been known which is capable of information-compressed coding of moving-image signals, such as in TV, at high efficiency.
FIG. 1
illustrates a prior art, in which the construction of a conventional system for movement-compensated inter-frame coding is shown.
In
FIG. 1
, video-buffer
51
stores sequentially current image data VD inputted. Meanwhile, frame memory
57
stores preceding image data FD, of one frame earlier, reproduced through decoding. The preceding image data FD is read out for inter-frame expectation coding of current image data VD with the movement of image compensated. In more detail, movement-vector calculation unit
58
calculates the moving of the image (pixel block) between the frames by block-matching search calculation to output the optimum movement vector MV.
In an example of block-matching search calculation, assuming that pixel data of pixel block BD are B
k
(k=1 to 16), pixel data of i-th pixel block RD
i
of 8×8 pixels in search region R in frame memory
57
are Y
i,k
(k=1 to 16), differential absolute value sum Si between the both pixel blocks Si is calculated by block-matching calculation:
S
i
=&Sgr;∥B
k
−Y
i,k
∥(
k
=1 to 16)
and then the optimum pixel block Y
i,k
that makes differential absolute value sum S
i
minimal is found out, thereby the optimum movement vector MV is obtained.
Meanwhile, variable delay buffer
59
serves to extract pixel block Y
i,k
corresponding to (optimum) movement vector MV from preceding image data FD to make expected block data PD for movement compensation. Further, subtracter
52
subtracts from respective pixel data VD to be coded the corresponding expected data for movement compensation PD to produce residual difference data PE. Quantumizer
53
quantumizes residual difference data PE to produce coding data CE to be transmitted.
In this state, dequantumizer
54
dequantumizes coding data CE, producing residual difference data PE′. Adder
55
adds expected data PD for movement compensation stated above to residual difference data PE′, to reproduce current pixel data VD′. Frame buffer
56
accumulates in sequence current pixel data VD′ thus reproduced. After the data for one frame are accumulated, the reproduced data for the one frame are transferred to frame memory
57
as image data FD for the preceding frame.
But the method above in which residual difference data PE for each pixel are quantumized cannot reduce the redundant information associated with the original image itself, thus, high rate of data compression cannot be expected. In this respect, MPEG (Moving Picture Experts Group) system which is popular among the recent systems of moving image compression has achieved a relatively high ratio of data compression by performing inter-frame expectation with movement compensation in a block of 16×16 image elements (pixels), performing two-dimensional DCT (Discrete Cosine Transform) in a unit of 8×8 pixels related to the expected residual difference thus obtained, quantumizing the sequency thus obtained and performing Haffmann coding.
The predominant portion of the expected residual difference by inter-frame expectation for movement compensation of this kind, however, tends to be concentrated to the peripheral portion of the original image block. This tendency is observed significantly particularly when the original image consists of animation image or color graphic game image containing flat portions and peripheral portions with steep gradient. Therefore, if expected residual difference is developed immediately by a system-fixed orthogonal base system (DCT) such as in conventional MPEG system above, many development coefficients (sequency) containing lower and higher frequency components are required, and a high ratio of data compression cannot be obtained. Further, if high-frequency components are quantumized with low precision in order to elevate data compression ratio, not only image information of the peripheral portions is lost, but also image quality is deteriorated by mosquito noises generated in the peripheral portions.
SUMMARY OF THE INVENTION
Accordingly, it is an object of the invention to provide a method for coding of a moving image capable of obtaining high image quality and high ratio of data compression (coding efficiency).
It is another object of the invention to provide an apparatus for coding of a moving image capable of obtaining high image quality and high coding efficiency.
It is still another object of the invention to provide a recording medium for recording a program of coding a moving image in high quality and with high efficiency.
According to the first feature of the invention, the object of the invention stated above is accomplished by a method of coding a moving image comprising the steps of:
comparing a pixel block of current image to be coded and a pixel block of preceding image within a predetermined region of preceding frame, one after another;
seeking a specific preceding-image pixel block minimizing matching error;
if said matching error relative to said specific preceding-image pixel block exceeds an acceptable value, obtaining one or more of orthogonal base systems for approximating AC component vector in said current-image pixel block by inter-frame adaptive orthogonal transformation having, as nest thereof, preceding image data within a predetermined region including said specific preceding image pixel block; and
thereby coding said image data.
According to the first feature of the invention, a high image quality and a high ratio of data compression are obtained by the constitution in which current pixel block B having a matching error relative to the image data of the preceding frame exceeding an acceptable value is approximated by one or more orthogonal base systems by inter-frame adaptive orthogonal transformation utilizing the image data of the preceding frame of a moving image which enable generally to obtain a high correlation between the frames. Further, current pixel block B (alternating current component vector) can be coded with improved efficiency by a small number of orthogonal bases, owing to the constitution utilizing the preceding image data within a predetermined region N including the specific preceding pixel block F minimizing the matching error relative to the current pixel block B as the nest of inter-frame adaptive orthogonal transformation (corresponding to the code book of vector quantumization). Further, remarkable improvement of the coding efficiency (reduction in the amount of codes) can be expected with the original image quality maintained honestly, particularly when the original image consists of animation image or color graphic game image containing a lot of flat portions and peripheral portions with steep gradient.
In the invention, “preceding frame” may be a frame immediately before a current frame in the order of displaying a moving image, or a frame which is displayed after the current frame in the order of displaying the moving image, and image data of which is prepared in advance of the display of the current frame for reference of the coding in the current frame.
According to the second feature of the invention, the first basis for approximating AC component vector <B> of the current pixel block is produced based on AC component vector <F> of the specific preceding pixel block, according to the first feature of the invention.
As specific preceding pixel block F minimizes the match
Itagaki Fumihiko
Kawashima Miyuki
Miura Takashi
Barnes & Thornburg
Chen Wenpeng
Conte Robert F. I.
Hudson Soft Co. Ltd.
LandOfFree
Method and apparatus for coding moving image and medium for... 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 for coding moving image and medium for..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for coding moving image and medium for... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3246340