Pulse or digital communications – Bandwidth reduction or expansion
Reexamination Certificate
1999-11-29
2004-02-24
An, Shawn S. (Department: 2713)
Pulse or digital communications
Bandwidth reduction or expansion
C375S240230, C375S240240, C375S240180, C375S240030, C382S246000, C382S250000, C382S251000, C382S248000
Reexamination Certificate
active
06697425
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention generally relates to dynamic image encoding apparatuses and methods for encoding a dynamic image that achieve a high encoding ratio and prevent image quality from being deteriorated by decoding and reproducing.
In order to transmit or store dynamic images, an encoding technology is desired in which a high compression rate is achieved without causing deterioration of images from reproduction. In addition, excellent recovery from a transmission error is desired. Thus, the present invention provides a dynamic image apparatus and a method for encoding a dynamic image that can realize the above technologies.
2. Description of the Related Art
Such a dynamic image encoding technology is described, for example, in ITU-T (International Telecommunication Union—Telecommunication Standardization Sector) H.263, the disclosure of which is hereby incorporated for reference. This is called a dynamic image hybrid encoding method. Also, Annex K, Annex Q and Annex P are known as ITU-T H.263.
FIG. 1
is a diagram for briefly explaining ITU-T H.263 Annex Q. In
FIG. 1
, when a picture changes slowly, that is, when activity (motion activity) is small, an input image
41
composed of 352×288 pixels CIF (Common Intermediate Format) is divided and encoded at 16×16 pixels per macroblock (MB) so as to be encoded to an image
42
. In this case, the input image
41
is divided into macroblocks of 22 columns×18 rows and then these macroblocks are encoded with motion compensations.
When the picture changes quickly, that is, when activity is large, a large volume of information is incurred. Thus, the input image
41
is divided and encoded at 32×32 pixels per MB so as to be an image
43
. In this case, the input image
41
is divided into macroblocks of 11 columns×9 rows. Then, these macroblocks, which volume is reduced to half volume, are encoded with motion compensations.
In this case, the motion compensation is performed for each macroblock composed of 32×32 pixels. Down-sampling corresponding to 16×16 pixels is performed for an estimation error of a macroblock composed of 32×32 pixels. Then, a DCT (discrete cosine transform), a quantization and a variable length coding are performed on the reduced estimated error. In this method, it is possible to maintain a high resolution for static background images. Also, it is possible to perform the down-sampling only for motion fields (dynamic image areas) and then to encode reduced motion fields. Thereafter, image information including block division information is transmitted. When the image information of the input image
41
is received at a receiver side, the image information is decoded so as to be a decoded image
44
composed of 352×288 pixels CIF as shown in FIG.
1
.
FIG. 2
is a diagram for briefly explaining ITU-T H.263 Annex P. In
FIG. 2
, when a picture changes slowly, an input image
51
composed of 352×288 pixels CIF is divided and encoded at 16×16 pixels per macroblock to form an image
52
in the same manner as Annex Q. When the picture changes quickly, the input image
51
is reduced so as to be an image
53
composed of 176×144 pixels QCIF. The image
53
is further divided and encoded at 16×16 pixels per MB. In this case, the encoded image has a different size from that of the reduced image. Thus, an estimated image is obtained by regenerating the original size. Also, when image information of the input image
51
is received at the receiver side, the image information is decoded based on CIF or QCIF information so as to be a decoded image
54
composed of 352×288 pixels CIF as shown in FIG.
2
.
As mentioned above, when the activity is small, the input image is encoded at a high resolution. On the contrary, when the activity is large, an overload of information is prevented by a division of information into smaller blocks or by reducing the volume of information of the input image. Thus, in order to avoid such an overload of information, the input image is divided by large blocks or the input image is reduced. Thereafter, the input image is encoded at a low resolution.
A block structure and a bit stream structure will be now explained. When the activity is large in a frame (t), the input image is reduced or divided by a large block. An image corresponding to QCIF after being reduced or divided is shown in FIG.
3
A. In this case, the input image is divided into 9 rows×11 columns of macroblocks. Thereafter, encoded macroblocks on each macroblock line are grouped to compose a GOB (Group Of Blocks). In this example, the divided image has 9 rows. Thus, the divided image is scanned along each row so as to produce 9 block groups of data, GOB
0
through GOB
8
.
In this case, the bit stream structure is composed of PH, GOB
0
, GOBH, GOB
1
, GOBH, GOB
2
, . . . , GOBH and GOB
8
. It should be noted that the PH means a picture header and the GOBH means a block group header. That is, a picture header PH is added at the beginning of GOB
0
. Also, block group headers GOBH are added at the beginning of the other block groups of data GOB
1
through GOB
8
.
When the activity is small in a frame (t+1), the input image corresponding to the CIF is divided by a small block. An image corresponding to QCIF after being divided is shown in FIG.
3
B. In this case, the input image is divided into 18 rows×22 columns of macroblocks. Thereafter, data of encoded macroblocks at a macroblock line is grouped so as to produce
18
block groups of data, GOB
0
through GOB
17
. In this case, the bit stream structure is composed of PH, GOB
0
, GOBH, GOB
1
, GOBH, GOB
2
, . . . , GOBH and GOB
17
.
The picture header and the block group header will be now explained.
FIG. 4A
is a diagram showing a picture header PH. The picture header PH includes a synchronous code (Picture Start Code), an image size, a coding mode and so on.
FIG. 4B
is a diagram showing a block group header GOBH. The GOBH includes a synchronous code (GOB/Slice Start Code) and location information in an image.
FIG. 4C
is a diagram showing a bit stream from a sender and
FIG. 4D
is a diagram showing a bit stream from a receiver. And
FIG. 4E
is a diagram showing a block group header GOBH which is in conformity to H.263. In
FIG. 4E
, the block group header GOBH includes the following codes: A GBSC is a unique synchronous code shown as “0000 0000 0000 0000 1”. A GN is a block group (GOB) number composed of 5 bits. A GFID is an ID number of the block group GOB. A GQUANT shows a default value for a quantization of this block group GOB. An MBD is a block group of data.
When the receiver normally receives the bit stream from the sender as shown in
FIG. 4C
, the receiver can recognize all information in the bit stream, including the image size of the picture header PH, the coding mode and so on. Thus, it is possible to successfully decode all information of the bit stream and reproduce one screen composed of a plurality of macroblock lines at the receiver side. Accordingly, an image successfully reproduced is displayed. However, when the receiver can not receive the picture header PH because of an occurrence of a transmission error, it may be impossible for the receiver to get information of the image size, a coding mode and so on. As a result, it is impossible to define the division number of an image that shows how many macroblocks the image is divided into. That is, even if the receiver obtains the image, the receiver can not recognize the number of macroblock lines composing one screen. In this case, it fails to decode and reproduce the image received at the receiver side.
Recently, in wireless mobile communication, it is possible to encod a dynamic image by such as a hybrid encoding method and transmit the encoded dynamic image. In such a mobile communication system that has improved to achive practical use, such as IMT-200, a maximum 2-Mbps transmission is realized. With regard to this transmission performance, it is easily realize
Anan Taizo
Morimatsu Eishi
Nakagawa Akira
Niem Wolfgang
An Shawn S.
Fujitsu Limited
Katten Muchin Zavis & Rosenman
LandOfFree
Dynamic image encoding apparatus and method for encoding a... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Dynamic image encoding apparatus and method for encoding a..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Dynamic image encoding apparatus and method for encoding a... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3349029