Image analysis – Image compression or coding – Adaptive coding
Reexamination Certificate
1999-04-02
2002-01-29
Couso, Yon J. (Department: 2721)
Image analysis
Image compression or coding
Adaptive coding
C382S232000
Reexamination Certificate
active
06343153
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to a high efficiency coding method of a moving video image signal and, more particularly, to a coding compression method which is appropriate for a storage media such as an optical disk.
BACKGROUND OF THE INVENTION
Generally, when the video image data is stored as it is, it requires a memory of a tremendously large capacity Accordingly, a technology of storing video image data in a media with efficiently compressing the video image data becomes that of a great importance.
On the other hand, from an optical disk, usually, video image data of constant rate is reproduced as seen in CD or LD. However, there are generally a video image of a scene including a large quantity of information and a video image of a scene including not so large quantity of information As a coding method effectively utilizing this nature of video image, there is proposed a variable bit rate coding method (as disclosed in Japanese laid-open publication No. Hei. 7-284097).
At first, a video image coding circuit which is generally used in MPEG will be described with reference to FIG.
13
. The coding circuit
2
in
FIG. 13
compresses data using the motion compensation DCT method. The motion compensation DCT method is a method of transmitting one frame which is periodically selected from the input video image with compressing the same by using only the data in that frame, and transmitting the remaining frame with compressing the difference between that frame and the previous frame. For the intra-frame compression and inter-frame compression, a discrete cosine transform as a kind of orthogonal base transformation is employed. In addition, when calculating the difference between frames, detecting the motion vector of the video image among the previous frames, and taking the difference after the motions being corrected, whereby the compression rate is enhanced to a great extent.
Hereinafter, the operation of the coding circuit
2
in
FIG. 13
will be described in more detail. In
FIG. 13
, the real lines show flows of data, and dotted lines show flows of control.
The video image data input from the input terminal
30
is stored in the frame buffer
25
, and thereafter, reordered in the coding order by the frame reordering circuit
26
. The subtracter
10
is used for calculating the difference from the previous frame. The coding control circuit
22
controls the on and off of the refresh switches
23
and
24
in accordance with the type of the pictures to be processed. In other words, the coding control circuit
22
switches of f the refresh switches
23
and
24
(performs in-frame compression) when the pictures to be processed are I pictures (intra-pictures). As a result, the subtracter
10
will not operate.
The input video image data is DCT converted by the DCT conversion circuit
11
. The DCT is usually performed two-dimensionally. When the DCT is to be performed for each 8×8 blocks, it results in 8×8 pieces of coefficients. The data to which DCT is performed is originally continuous quantity, but because the operation is performed using digital circuits,
64
pieces of coefficients are obtained as digital values of a predetermined bit width. This data is subjected to appropriate bit allocation for each frequency component by a quantization circuit
12
. In general, low band component is subjected to a large number of bit allocation because low band component is important for constituting a video image, and high band component is subjected to a small number of bit allocation because high band component is not so important for constituting a video image. A variable length coding (VLC) circuit
13
performs variable length coding of the output of the quantization circuit
12
. The variable length coding is a method of allocating a shorter code length to data, which has statistically higher appearance probability, and this method enables to remove the statistically redundant component included in the data. In this method, Huffmann code is often used.
An inverse quantization circuit
15
reverses the quantization of the output of the quantization circuit
12
. In contrast to the quantization, the inverse quantization circuit
15
restores the amplitude of each frequency component. Each coefficient restored to the original amplitude by the inverse quantization circuit
15
is restored to the original video image data by the inverse DCT circuit
16
. When the restored video image data is intra-frame video image data, an adder
17
will not operate. Thereafter, the restored video image data is delayed by a predetermined number of frames by a frame memory
18
. The delayed video image data is input to a motion vector detecting circuit
20
. The motion vector detecting circuit
20
calculates a motion amount from the input video image data. A motion compensation circuit
19
moves the place of the video data image in accordance with the motion amount. Thus, the video image data subjected to a motion compensation is used to calculate the difference from the next video image data by the subtracter
10
.
The video image of some frames followed by I picture are used for compressing the difference from the video image data of the former frame. A coding control circuit
22
switches the refresh switch
23
or
24
on in case where the picture to be processed is P picture or B picture (intra-frame compression). The refresh switch
23
is switched on in calculating the difference between frames and is used for operating the subtracter
10
.
The refresh switch
24
is repeatedly switched on or off alternatively with the refresh switch
23
at a predetermined period. When it is switched on, it is used to operate adder
17
to add the different data between the frames and the former frame data, and is used to restore the frame. A variable length coding circuit
13
also performs the variable length coding to the inter-frame compression data.
Here, a group from I picture before the next I picture is called
1
GOP (Group of Picture), and it is usually constituted from the video image signal having about 15 frames (about 0.5 second).
The output of a variable length coding circuit (VLC)
13
is output through the buffer circuit
14
to an output terminal
31
. The quantization scale decision circuit
21
indicates, viewing the state of the buffer circuit
14
, a quantization scale to the quantization circuit
12
. To be specific, when it is assumed that a video image signal is output from the buffer circuit
14
at a prescribed constant rate, when the data left in the buffer circuit
14
is small, it is necessary to generate a large amount of data, and therefore, the quantization scale Q is controlled to be smaller than before, thereby to increase the number of the generated bit amount. On the contrary, when the data left in the buffer circuit
14
is too large, it is necessary to make it difficult to generate data, and therefore the quantization scale Q is controlled to become larger than before, thereby to decrease the number of generated bit.
To be specific, the target data amount which is a data amount to be generated in the next section is calculated from the buffer residual quantity, and then the quantization scale is calculated from the target data amount. When the target data amount is large, the quantization scale Q is required to be small, while when the target data amount is small, the quantization scale Q large. In short, the target data amount and the quantization scale are in a relation of inverse proportion.
The target data amount for each section of the video image signal is calculated in advance to be stored in the quantization scale decision circuit
21
, and the target data amount is compulsorily and successively changed for each short section while performing coding, whereby a variable bit rate control is realized,
In
FIG. 14
, the description of the variable bit rate control of two-path system, which is usually employed, is given. The first path inputs the prescribed video image signal such as 2 hours video image signal from the terminal
Hasebe Takumi
Kashiwagi Yoshiichiro
Kawasaki Kojiro
Nakamura Kazuhiko
Couso Yon J.
Wenderoth , Lind & Ponack, L.L.P.
LandOfFree
Coding compression method and coding compression apparatus does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Coding compression method and coding compression apparatus, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Coding compression method and coding compression apparatus will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2830393