Pulse or digital communications – Bandwidth reduction or expansion – Television or motion video signal
Reexamination Certificate
1999-07-21
2003-03-11
Le, Vu (Department: 2613)
Pulse or digital communications
Bandwidth reduction or expansion
Television or motion video signal
C375S240050, C375S240270
Reexamination Certificate
active
06532262
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to technology of coding a signal such as a video or audio signal and technology of recording coded data on a storage medium.
In coding a video signal, transform coding is usually employed. According to this technique, an image is divided into a plurality of blocks, each composed of a multiplicity of pixels neighboring each other, and an orthogonal transform, like discrete cosine transform, is performed on each of these blocks. Also, a transform coefficient is quantized based on a predetermined quantization unit (quantization parameter), and then compression-coded by a variable-length encoding technique such as Huffman coding.
Also, in coding a moving picture like a TV signal, inter-frame coding is performed by utilizing a correlation between frames. According to the inter-frame coding technique, an image at a frame to be coded is predicted using, as a reference frame, another frame preceding or succeeding the frame in question on the time axis, and a prediction error thereof is coded and transmitted or recorded. The inter-frame coding is performed on the basis of a block composed of a plurality of pixels. And the motion quantity of each block is also transmitted or recorded along with the prediction error.
FIG. 22
illustrates a coding technique according to the MPEG-2 (Moving Picture Experts Group) standard, which is an international standard of moving picture coding. According to the MPEG-2 standard, intra-frame coding, inter-frame forward predictive coding and inter-frame bidirectional coding are performed. A frame, on which intra-frame coding is performed, is called an “I-frame”. A frame, on which inter-frame forward predictive coding is performed, is called a “P-frame”. And a frame, on which inter-frame bidirectional coding is performed, is called a “B-frame”. In a GOP (group of pictures) consisting of a plurality of frames, at least one of the frames is intra-frame-coded.
According to such a coding technique, variable-length encoding is employed, and therefore, the number of bits generated cannot be recognized accurately until the coding processing is over. Thus, feedback control is performed such that a cumulative error, which is an accumulation of errors between numbers of bits generated and target numbers of bits, becomes zero and that an average coding rate during a predetermined interval can be approximated at a target coding rate.
Also, a variable-rate coding technique for coding a video at a coding rate variable with the complexity of coding an individual scene thereof is lately applied to a DVD (digital video disc) and so on. According to this technique, a coding rate is increased for a scene with a high complexity of coding, thereby reducing coding noise. On the other hand, as to a scene with a low complexity of coding (such as a scene with little motion), the coding rate is decreased, thereby balancing the overall image quality of the video and reducing an average bit rate.
Variable-rate coding is implementable most easily by fixing a quantization parameter for quantization processing. If the quantization parameter is fixed, then the resolution of quantization is also fixed, but the coding rate for a scene with a high complexity of coding gets higher than a scene with a low complexity of coding. Accordingly, the coding rate becomes variable.
FIG. 23
is a block diagram illustrating a configuration for a conventional coder for controlling an average rate by a feedback control technique. As shown in
FIG. 23
, an input video signal is quantized by a quantizer
701
using a quantization parameter Q, which has been determined by a rate-controlling-parameter determiner
704
. The output of the quantizer
701
is variable-length encoded by a variable-length encoder
702
and then output as coded data. A bit number calculator
703
calculates a number of bits B of the coded data generated per unit time t.
In the rate-controlling-parameter determiner
704
, an error calculator
711
obtains an error d (=B−BT) by subtracting a target number of bits BT, which is derived from a target coding rate, from an actual number of bits generated B. A cumulative error calculator
712
accumulates the errors d, thereby obtaining a cumulative error D. A quantization parameter determiner
713
controls the quantization parameter Q such that the cumulative error D becomes zero.
If the cumulative error D is a positive large value, then the quantization parameter Q is increased to reduce the number of bits generated B. Conversely, if the cumulative error D is negative, then the quantization parameter Q is decreased to increase the number of bits generated B. The quantization parameter Q may be given by the following equation:
Q=Q′
/(1−
D/T
)
where Q′ is a quantization parameter used in previous quantization processing and T is a time constant used for the feedback control. During the interval T, the cumulative error D becomes zero and the average coding rate during the interval T can be controlled at the target coding rate.
In this case, by setting the time constant T at a sufficiently large value, variable-rate coding can be performed during the interval T. This is because if the time constant T is large enough, the quantization parameter Q hardly varies and is converged at a substantially constant value. As a result, a coding rate increases for a scene with a relatively high complexity of coding, but decreases for a scene with a relatively low complexity of coding. That is to say, coding processing can be performed at a rate variable with the complexity of coding an individual video scene.
The conventional coding processing with average rate control, however, has the following drawbacks.
The sensitivity of the human eyes to coding noise resulting from coding processing greatly differs depending on the complexity of coding an individual video scene. Specifically, coding noise generated in a scene with a low complexity of coding (e.g., a substantially uniform picture with almost no motion such as “blue sky”) is more perceivable to the human eyes compared to a scene with a high complexity of coding (e.g., a complicated picture with a lot of motion such as “crowd”). In other words, the human vision is more sensitive to coding noise generated in a scene with a low complexity of coding than that of a scene with a high complexity of coding.
Thus, in coding a video with a substantially constant quantization parameter, even if a coding noise generated does not seriously affect the image quality of a scene with a high complexity of coding, non-negligible deterioration may be caused by the same noise in a scene with a low complexity of coding.
Also, in controlling an average coding rate according to a conventional technique, if a cumulative error is large when a scene with a high complexity of coding is changed into a scene with a low complexity of coding, the quantization parameter is controlled at a larger value. In a scene with a low complexity of coding, however, even if the quantization parameter is increased, the number of bits generated does not decrease so much. Thus, although the average coding rate can be controlled at a value closer to the target coding rate, deterioration in image quality might surpass the advantageous effects attained by the rate control.
Furthermore, the feedback control might cause oscillation in quantization parameter or number of bits generated, and the image quality might considerably change in such a case.
In general, optimum number of bits and quantization parameter may be determined for an individual scene by estimating the complexity of coding a signal in advance. In such a case, however, since an additional time is needed to estimate the complexity of coding, it is difficult to perform coding processing in real time.
SUMMARY OF THE INVENTION
An object of the present invention is to reproduce a signal of quality at a low coding rate even when a signal continuously supplied should be coded in real time.
Specifically, a method according to the present invention is
Fukuda Hideki
Kondo Satoshi
Nakamura Kazuhiko
Shibata Hideaki
Bugg George A.
Cole Thomas W.
Le Vu
Matsushita Electric - Industrial Co., Ltd.
Nixon & Peabody LLP
LandOfFree
Coding method and apparatus and recorder 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 method and apparatus and recorder, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Coding method and apparatus and recorder will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3026787