Image analysis – Image compression or coding – Transform coding
Reexamination Certificate
1997-11-17
2001-04-10
Boudreau, Leo H. (Department: 2721)
Image analysis
Image compression or coding
Transform coding
Reexamination Certificate
active
06215909
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates generally to methods and apparatus for data compression and decompression, and in particular, is directed to methods and systems for performing the discrete cosine transform (DCT) and weighting processes of a digital video encoder and the inverse weighting and inverse discrete cosine transform (IDCT) processes of a digital video decoder.
Digital video is the term used to describe video signals represented in digital form. Digital video offers several advantages over traditional analog systems. For example, recordings of digital video signals can be copied indefinitely without significant loss of quality. Also, compressed digital video signals may require less storage space than analog recordings for the same or better picture quality. Finally, digital formats allow audio, video, and other data to be easily combined, edited, stored, and transmitted.
A direct conversion into digital format of analog video signals having fast frame rates, many colors, and high resolution, however, results in digital video signals with a high data rate, creating difficulties for storage and transmission. Many digital video systems, therefore, reduce the amount of digital video data by employing data compression techniques that are optimized for particular applications. Digital compression devices are commonly referred to as encoders; devices that perform decompression are referred to as decoders. Devices that perform both encoding and decoding are referred to as codecs.
The “DV” format is an industry digital video format specification for use primarily in consumer-level video tape recorders VTRs). The specification for DV format has been adopted by most of the major manufacturers of high-quality digital video cassette recorders (DVCRs) and digital video camcorders. See Specifications of Consumer-Use Digital VCRs, HD Digital VCR Conference, December 1994. The DV format is currently used in such commercially available products as digital camcorders.
Video displays have traditionally consisted of a series of still pictures, or “frames”, painted by scan lines and sequentially displayed at a rate of, for example, thirty frames per second to provide the illusion of continuous motion. Each frame consists of a pair of interlaced “fields.” A field contains half the number of lines of a frame. Fields are interleaved with lines from either a previous or subsequent field to create a frame. When storing or transmitting video data, the amount of data may be reduced by taking advantage of redundancy within fields (intrafield) or between neighboring fields (interfield). DV format uses both intrafield and interfield data reduction.
FIG. 1
is a basic flow diagram showing the encoding, or data compression, process of a prior art digital video codec. Codecs employing DV format use a DCT-based data compression method. In the blocking step, the image frame is divided into N by N blocks of pixel information including, for example, brightness and color data for each pixel (Step
100
). A common block size is eight pixels horizontally by eight pixels vertically. The pixel blocks are then “shuffled” so that several blocks from different portions of the image are grouped together (Step
110
). Shuffling enhances the uniformity of image quality.
Different fields are recorded at different time incidents. If a video scene contains a large amount of motion, the two fields within a frame contain significantly different image information, and DV encoders use an intrafield data reduction process to remove redundancy within a field.
In video images without substantial motion, the two fields of a frame contain similar image information, and DV encoders use an interfield data reduction process to remove redundancy between fields. For each block of pixel data, a motion detector looks for the difference between two fields of a frame (Step
115
). The motion information is sent to the next processing step (Step
120
).
In step
120
, pixel information is transformed using a DCT. There are at least two common DCT modes: 8—8 DCT mode and 2-4-8 DCT mode. The 8—8 DCT mode refers to a DCT that takes eight inputs and returns 8 outputs in both vertical and horizontal directions. In the 2-4-8 DCT mode, an 8 by 8 block of data is divided into two 4 by 8 fields, each field consisting of 4 horizontal lines of 8 components. A two-dimensional 4 by 8 transform is performed on each field, each 4×8 transform consisting of a one-dimensional transform taking 4 inputs and returning 4 outputs in the vertical direction, and a one-dimensional transform taking 8 inputs and returning 8 outputs in the horizontal direction. The DV format specification recommends that the 8—8 DCT mode be used when the difference between two fields is small. By contrast, the 24-8 DCT mode should be used when two fields differ greatly.
In the 2-4-8 DCT mode, 8×8 blocks of pixel information are divided into two 4×8 blocks of pixel information. The first block represents the sums the rows; the second block represents the differences of the rows. Each 4×8 block is transformed into a 4×8 matrix of corresponding frequency coefficients using a two-dimensional DCT. In the following equations, P(x,y) represents an input block of pixel information with symbols x and y representing pixel coordinates in the DCT block. Q′(h,v) represents the resulting output block of DCT coefficients for the sum information and Q′(h,v+4) represents the output block of DCT coefficients on the difference information. The DCT in the 2×4×8 mode may be described mathematically as follows:
For h=0, 1, . . . 7 and v=0, 1, . . .
3
,
Q
′
(
h
,
v
)
=
C
(
h
)
C
(
v
)
∑
m
=
0
3
∑
x
=
0
7
(
P
(
x
,
2
m
)
+
P
(
x
,
2
m
+
1
)
)
COS
(
π
h
(
2
x
+
1
)
16
)
COS
(
π
v
(
2
m
+
1
)
8
)
Q
′
(
h
,
v
+
4
)
=
C
(
h
)
C
(
v
)
∑
m
=
0
3
∑
x
=
0
7
(
P
(
x
,
2
m
)
-
P
(
x
,
2
m
+
1
)
)
COS
(
π
h
(
2
h
+
1
)
16
)
COS
(
π
v
(
2
m
+
1
)
8
)
where
C
(
h
)
=
{
1
2
2
,
h
=
0
1
2
,
h
=
1
to
7
and
C
(
v
)
=
{
1
2
2
,
v
=
0
1
2
,
v
=
1
to
7
The DCT coefficients are then weighted by multiplying each block of DCT coefficients by weighting constants (Step
124
). This process may be described mathematically as follows:
Q (h, v)=W (h, v) Q′(h, v)
The following weighting coefficients are standard for the DV format.
W
(
h
,
v
)
=
{
1
4
,
h
=
0
,
v
=
0
(
w
(
h
)
w
(
2
v
)
)
2
,
0
<
v
<
4
(
w
(
h
)
w
(
2
(
v
-
4
)
)
)
2
,
4
≤
v
<
7
re w(O)=1
w (1)=CS4/(4×CS7×CS2)
w (2)=CS4/(2×CS6)
w (3)=1/(2×CS5)
W(4)=7/8
w(5)=CS4/CS3
w (6)=CS4/CS2
w (7) =CS4/CS1 and CSm =COS (mII/16).
The weighted DCT coefficients, Q(h,v), are stored to a buffer (Step
125
).
The weighted DCT coefficients are quantized in the next step (Step
140
). Quantization increases the efficiency of video data transmission, but may result in error propagation. To reduce the magnitude of errors, each DCT block is classified into one of four activity classes described in the DV format specification (Step
130
). The four classes represent four different quantizing schemes. The amount of data in the variable length codeword using each quantizer is estimated (Step
135
) and the quanfizer that best will compress one or more successive weighted DCT coefficients into a same size block as a synchronization block is selected.
Quantization rounds off each DCT coefficient within a certain range
Chang Ching-Fang
Lee C. C.
Yanagihara Naofumi
Boudreau Leo H.
Finnegan Henderson Farabow Garrett & Dunner L.L.P.
Schirduan Paul H.
Sony Electronics Inc.
LandOfFree
Method and system for improved digital video data processing... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and system for improved digital video data processing..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and system for improved digital video data processing... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2464492