Pulse or digital communications – Bandwidth reduction or expansion – Television or motion video signal
Reexamination Certificate
1999-12-29
2001-12-25
Britton, Howard (Department: 2613)
Pulse or digital communications
Bandwidth reduction or expansion
Television or motion video signal
C375S240030, C375S240040, C375S240120, C375S240240, C375S240250
Reexamination Certificate
active
06333949
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to video coding apparatus and video decoding apparatus, and more particularly, to a video coding apparatus that performs predictive coding of digital video signals and a video decoding apparatus that reproduces the original motion images from the predictive-coded video signal produced by the video coding apparatus.
2. Description of the Related Art
The ITU-T standard H.261 and the ISO standards MPEG-1 and MPEG-2, for example, are well-acknowledged international standards for motion picture coding techniques. Those standards use hybrid coding algorithms, where the coding process will proceed as follows: (1) a source picture is divided into blocks of pixels, (2) orthogonal transformation (e.g., discrete cosine transform) and motion compensation are applied independently on each block, and (3) quantized video data is compressed by entropy coding.
When a motion of considerable magnitude or a full scene transition happened in the middle of a sequence of video frames, the above-described hybrid video coding techniques may suffer from an overwhelming amount of coded frame data that exceeds a certain standard level allowed for each frame. In this case, the coder will forcibly reduce the amount of coded data in an attempt to regulate it at the standard level. This will cause extreme degradation in image quality and coarse frame subsampling (or a drop in frame update rates), thus resulting in unacceptably poor pictures when reconstructed at the receiving ends.
A video coding system aiming at avoidance of the above problem is proposed in Japanese Patent Application No. 8-75605 (1996), for instance, by the same applicant of the present invention. In this proposed system, the video coding apparatus reduces the resolution of input frame signals to regulate the amount of coded frame data when a full scene transition or a massive motion has happened in the middle of a sequence of video frames.
FIG. 14
is a block diagram of this video coding apparatus proposed in the Japanese Patent Application No. 8-75605. The apparatus of
FIG. 14
supports two kinds of picture resolutions: Common Intermediate Format (CIF, 352×288 pixels) and quarter-CIF (QCIF, 176×144 pixels). A CIF/QCIF selection controller
125
determines which picture resolution should be used to encode source pictures, considering the amount of coded frame data produced in a predictive coding, quantizer step size, and some other parameters. For example, the CIF/QCIF selection controller
125
normally chooses the high resolution CIF, while it chooses the low resolution QCIF when a large amount of data has been produced as a result of the coding.
A frame memory
122
is used to store reconstructed (or decoded) pictures of the previous frames. Comparing the source picture of the current frame with a decoded picture that is retrieved from the frame memory
122
as the reference picture, a prediction parameter calculation unit
112
computes motion vectors of the current frame. Here, a picture is partitioned into a plurality of blocks and the comparison of frame data is performed on a block-by-block basis. Each source frame picture is subjected to either an intraframe coding or an interframe coding. A prediction parameter calculation unit
112
determines which coding scheme should be applied to the source frame picture. When the interframe coding is activated, a prediction picture generation unit
113
produces a prediction picture of the current frame based on the decoded image of the previous frame and the motion vectors calculated by the prediction parameter calculation unit
112
.
A prediction error signal generation unit
114
produces a prediction error signal by calculating differences between the source picture and the prediction picture on a block-by-block basis. A CIF/QCIF converter
131
changes the resolution of this prediction error signal, which is originally CIF, to what is chosen by the CIF/QCIF selection controller
125
. More specifically, the CIF/QCIF converter
131
outputs the prediction error signal as it is when the CIF resolution is selected by the CIF/QCIF selection controller
125
, and it in turn converts the resolution to QCIF when the QCIF resolution is selected.
A coding controller
124
receives information regarding the amount of the resultant coded data from an entropy coding unit
117
(described later), as well as obtaining information on buffer occupancy from a coded data buffer
118
(described later). Based on such information, the coding controller
124
determines the quantizer step size and distributes it to a quantizer
116
, a dequantizer
119
, the CIF/QCIF selection controller
125
, and the entropy coder
117
.
A DCT processor
115
applies an orthogonal transform, or a digital cosine transform (DCT), to the output of the CIF/QCIF converter
131
, and a quantizer
116
quantizes the obtained DCT coefficients in accordance with the quantizer step size specified by the coding controller
124
.
The entropy coder
117
receives the quantized DCT coefficients from the quantizer
116
, the picture resolution from the CIF/QCIF selection controller
125
, and the motion vectors and coding scheme information from the prediction parameter calculation unit
112
. Entropy coding is a data compression process that assigns shorter code words to frequent events and longer code words to less frequent events. Out of a predefined code word table, the entropy coder
117
retrieves code words relevant to each combination of the above received data, thereby outputting the coded frame data.
The quantized DCT coefficients produced by the quantizer
116
are also supplied to the dequantizer
119
for inverse quantization, or dequantization. The resultant output signals are then subjected to an inverse discrete cosine transform (IDCT) process that is executed by an IDCT processor
120
to reproduce the original prediction error signal. When the reproduced prediction error signal has the QCIF format as a result of the resolution reduction by the CIF/QCIF converter
131
, a QCIF/CIF converter
132
reconverts it to regain the original CIF resolution. A decoded picture generator
121
reconstructs a picture by adding the prediction error signal outputted by the QCIF/CIF converter
132
to the prediction picture produced by the prediction picture generator
113
. This fully decoded picture is then transferred to a frame memory
122
for storage.
As described above, the proposed video coding apparatus monitors the amount of coded frame data and the like, and if any significant increase is expected in the amount of coded frame data, the apparatus will reduce the resolution of the prediction error signal from CIF to QCIF.
The CIF/QCIF converter
131
performs such resolution reduction through a downsampling process as exemplified in FIG.
15
. More specifically, white dots in
FIG. 15
represent CIF pixels and lower-case alphabetic characters placed in them indicate their respective prediction error signal values. Black dots represent QCIF pixels, and upper-case letters beside them signify their respective prediction error signal values. The downsampling process calculates the QCIF prediction error signal values A, B, C, and D by averaging four values of the CIF pixels surrounding each of the QCIF pixels. For example, the pixel value A is obtained as
A=
(
a+b+e+f
)/4. (1)
In contrast to that, the QCIF/CIF converter
132
performs a QCIF-to-CIF resolution conversion through an upsampling process as shown in FIG.
16
. More specifically, black dots represent QCIF pixels, and upper-case letters beside them indicate their respective prediction error signal values, while white dots represent CIF pixels and lower-case letters in them indicate their respective prediction error signal values. To obtain the CIF prediction error signal values a, b, c, and so on, the upsampling process calculates a weighted average value of four QCIF pixels surrounding each CIF pixel. For example, the pixel value f is obtai
Kazui Kimihiko
Morimatsu Eishi
Nakagawa Akira
Shimizu Takahiro
Britton Howard
Fujitsu Limited
Staas & Halsey , LLP
LandOfFree
Video coding apparatus and decoding 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 Video coding apparatus and decoding apparatus, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Video coding apparatus and decoding apparatus will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2583041