Pulse or digital communications – Bandwidth reduction or expansion – Television or motion video signal
Reexamination Certificate
2001-04-19
2004-11-16
Philippe, Gims (Department: 2613)
Pulse or digital communications
Bandwidth reduction or expansion
Television or motion video signal
C375S240000, C375S240010, C375S240030, C375S240120, C375S240170, C375S240250
Reexamination Certificate
active
06819714
ABSTRACT:
BACKGROUND OF THE INVENTION
(1) Field of the Invention
The present invention relates to a video encoding apparatus and particularly to the video encoding apparatus that adjusts the amount of codes by skipping encoding of image data.
(2) Related Art
In video encoding according to the MPEG standards, the amount of codes is controlled by predicting the amount of pictures to be accumulated in a receiver buffer of a decoding apparatus. This is referred to as control of the code amount performed by the Video Buffering Verifier (VBV) model.
FIG. 1A
shows changes of the predictive accumulation amount of the receiver buffer. As shown in the figure, pictures are inputted in the receiver buffer at a predetermined bit rate. At the time indicated by DTS (Decoding Time Stamp), one picture is outputted from the receiver buffer for being decoded. If the decoding apparatus has a display unit adopting NTSC (National Television System Committee) format, DTS is set at every {fraction (1/30)} seconds when one frame is allocated to one picture, and when one field is allocated to one picture, DTS is set at every {fraction (1/60)} seconds. Also, if the decoding apparatus has a display unit adopting PAL (Phase Alternation Line) format, DTS is set at every {fraction (1/25)} seconds when one frame is allocated to one picture, and when one field is allocated to one picture, DTS is set at every {fraction (1/50)} seconds.
Normally, when D1 denotes the amount of bits generated for decoding a picture at DTS
1
, the predictive accumulation amount of the receiver buffer decreases from V
1
to V
1
* (=V
1
−D
1
) at DTS
1
as shown in FIG.
1
A.
The VBV model controls the amount of codes so that the predictive accumulation amount of the receiver buffer does not cause an overflow or an underflow.
In some circumstances, i.e. during scene changes, an underflow of the receiver buffer may occur due to a sequence of pictures which each require a large amount of bits, as shown in FIG.
1
B. When D
3
denotes the amount of bits generated for decoding a picture at DTS
3
, the predictive accumulation amount of the receiver buffer is below zero at DTS
3
since V
3
−D
3
<0. This phenomenon happens because the picture to be decoded at DTS
3
is yet to be inputted in the receiver buffer. To avoid such an underflow of the receiver buffer, the quantization scale is increased so as to decrease the generation bit amount.
On the other hand, an overflow of the receiver buffer may occur due to a sequence of pictures which each require a small amount of bits, as shown in FIG.
1
C. When the predictive accumulation amount of the receiver buffer prior to decoding at DTS
3
is V
3
(=V
2
*+R, where R is the amount of bits inputted into the receiver buffer of the decoding apparatus, in other words, the amount of bits transmitted from the encoding apparatus to the receiver buffer, during each time interval of DTS (the time interval between decoding of two consecutive pictures)), V
3
exceeds the storage capacity of the receiver buffer. To avoid such an overflow of the receiver buffer, the quantization scale is decreased so as to increase the generation bit amount.
However, when the generation bit amount is significantly decreased by increasing the quantization scale to prevent an underflow, image quality will be deteriorated. To prevent such deterioration, the following methods are conventionally employed along with the adjustment of the quantization scale.
The first method uses so called skipped macroblocks. According to the MPEG standards, encoding is performed on a macroblock basis, the macroblock being a block of 16×16 pixels. A skipped macroblock is a macroblock composed of a special code indicating to display an image that is identical to the reference image at the location, and its data amount is extremely small. Accordingly, when an underflow is likely to occur, skipped macroblocks are transmitted instead of performing normal encoding of the original image.
However, the above method is problematic because a macroblock that is not a skipped macroblock shows a part of the original image, and a skipped macroblock shows a part of the reference image, which makes the reconstructed image lack in consistency as a whole.
FIG. 2A
shows examples of images displayed in a normal case.
FIG. 2B
shows examples of images displayed when a skipped macroblock is used. In
FIG. 2B
, as a skipped macroblock is not used for the upper macroblock of the second frame, a part of the original image of the second frame is displayed in the upper macroblock. On the other hand, as a skipped macroblock is used for the lower macroblock of the second frame, a part of an image displayed in the lower macroblock of the first frame is also displayed in the lower macroblock of the second frame. This makes the reconstructed image in the second frame lack in consistency as a whole.
The second method is to encode pseudo image data. Pseudo image data is image data whose pixel values are medians of possible pixel values. As one example, when a pixel value is expressed in eight-bit, its median is 128. According to the MPEG standards, the difference between each pixel value of the image data and its median is encoded, and so the amount of data resulting from encoding such image data whose pixel values are medians becomes minimum. Accordingly, when an underflow is likely to occur, on a macroblock basis, image data of the macroblock whose pixel values are medians may be encoded instead of performing normal encoding on the macroblock.
However, the problem with the above method is that a gray-colored image is displayed where the image data whose pixel values are medians has been encoded.
FIG. 2C
shows examples of images displayed when a pseudo image is used. In the second frame, a part of the original image is displayed in the upper macroblock since normal encoding has been performed thereon. On the other hand, a gray-colored image is displayed in the lower macroblock where pseudo image data has been encoded. This makes the reconstructed image in the second frame lack in consistency as a whole.
By the way, Japanese Patent No. 2871316 discloses a method for skipping encoding of image data of one field or one frame (hereafter referred to as “skipping a picture”)
FIG. 3
shows the construction of a video encoding apparatus disclosed in Japanese Patent No. 2871316. Referring to the overall construction of this video encoding apparatus, input moving image data is subjected to an encoding process performed by an orthogonal transformation circuit
6
and other units, and generated pictures are stored in a buffer memory
20
. When a transmission rate excess judging circuit
24
judges that a transmission rate of each picture exceeds a predetermined threshold, a SKIP code stored in a SKIP code storage memory
22
is outputted, or when the judgment result of the transmission rate excess judging circuit
24
is negative, the picture stored in the buffer memory
20
is outputted.
According to the method described above, when a picture is to be skipped, not only macroblocks occupying a part of the picture but also all the macroblocks except a first and a last macroblock in each slice layer are replaced with skipped macroblocks. This enables a decoding apparatus to display an image that is identical to a previously decoded image, avoiding displaying such an image that lacks in consistency as described above.
However, the above method has the following problems.
The first problem is that the buffer memory
20
shown in
FIG. 3
is necessary for temporarily storing generated pictures for the judgment whether the amount of each generated picture exceeds the transmission rate. In more detail, when the judgment is performed on each macroblock as to whether encoding an image of each macroblock is to be skipped as described above, a buffer with a small capacity can be provided for storing data encoded on a macroblock basis. However, with the present method, each picture as a whole is subjected to the judgment as to whether the picture is to be skipped, wh
Isomura Tsuyoshi
Morishita Masahiro
Yamada Nobuhiko
Matsushita Electric - Industrial Co., Ltd.
Philippe Gims
Wong Allen
LandOfFree
Video encoding apparatus that adjusts code amount by... 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 encoding apparatus that adjusts code amount by..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Video encoding apparatus that adjusts code amount by... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3332589