Pulse or digital communications – Bandwidth reduction or expansion – Television or motion video signal
Reexamination Certificate
2000-11-01
2003-07-01
Britton, Howard (Department: 2613)
Pulse or digital communications
Bandwidth reduction or expansion
Television or motion video signal
C348S705000, C375S240260, C386S349000, C386S349000
Reexamination Certificate
active
06587506
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a video editing apparatus, a video editing method, and a medium for storing a video editing program, and relates more particularly to technology for extracting a plurality of contiguous frames (scenes) from a bitstream encoded according to a Motion Picture Expert Group (MPEG) standard, and producing a new bitstream by combining a plurality of the extracted scenes.
2. Description of Related Art
MPEG is a family of international standards for encoding moving pictures (hereafter referred to as simply “video”). It includes MPEG-1, which is used for video CD and PC video data, for example, and MPEG-2, which is used with DVD and digital broadcast satellite. Other applications for the MPEG standards continue to be found.
More specifically, MPEG has been adopted by the International Standards Organization (ISO) as a standard for a video coding method defining bitstream interpretation and decoding techniques. The MPEG-1 standard has been adopted as ISO-11172, and MPEG-2 as ISO-13818.
MPEG-1 defines a compression technique for compressing and storing video to a digital storage medium with a 1.5 Mbps transfer rate.
MPEG-2 extends MPEG-1, and defines a compression technique more specifically considering applications with communications and broadcast media, in addition to storage media.
Under MPEG-1, video data consists of a sequence of picture frames, enabling the pictures to be compressed using correlations within each frame (intra-frame coding) and correlations between frames (inter-frame coding). Combining these coding techniques yields three picture types based on the compression technique(s) used: I-pictures, or intra-coded pictures; P-pictures, or predictive-coded pictures based only on temporally preceding pictures; and B-pictures, or bidirectionally predictive-coded pictures.
I-pictures are coded based solely on the data within that picture frame, and thus have no correlation to any other frame. P-pictures are coded with reference (correlation) to a temporally preceding (past) frame. B-pictures are coded with correlation to temporally preceding (past) and/or following (future) frames.
FIG. 13
shows the correlation between pictures in an MPEG-1 bitstream. Each square in
FIG. 13
represents one picture (frame).
Each frame is labelled with the picture type and ordinal sequence. I indicates an I-picture, P a P-picture, and B a B-picture. Note that this same designation is used throughout the figures and this specification to indicate the picture type.
The frames are further shown in display order from left to right, and the arrows in
FIG. 13
indicate the correlation between frames. For example, from
FIG. 13
we know that frame B
3
is coded with reference to frames I
1
and P
4
.
Because a specific frame can thus be coded with reference to a temporally following (future) frame, the sequence in which frames are presented (the display order, shown on the top row in
FIG. 14
) to the viewer and the sequence in which frames are stored on the data storage medium (the coding order or data cumulation order in buffer, shown on the bottom row in
FIG. 14
) are different in an MPEG-1 bitstream containing B-pictures.
Generally speaking, the compression efficiency of these picture types is:
I-pictures<P-pictures<B-pictures
and the code size is conversely
I-pictures>P-pictures>B-pictures.
The MPEG-2 scheme can be applied to picture data having a frame structure or a field structure. Video scanning methods include, broadly, non-interlaced scanning and interlaced scanning.
In non-interlaced scanning all pixels in one frame are sampled at the same time. In this case the video is a collection of frames, and thus has a frame structure.
With interlaced scanning every other line in one picture frame is sampled at the same time. The first set of lines sampled at a first time is referred to as the first field, and the second set of lines sampled at a second time is referred to as the second field. Each frame in interlaced scan video thus consists of two fields, and the video has a field structure.
The picture structure in MPEG-2 video having a frame structure is the same as in MPEG-1. However, picture correlations in field structure video are more complicated. Picture correlations in field structure video are shown in FIG.
15
.
In
FIG. 15
each square represents one field, and the fields are arranged in display order. As will be known from
FIG. 15
, a P-field can be referenced to the most recently decoded I-field, an I-field and a P-field, or to two P-fields.
However, if the P-field is coded using an I-field as the first field and a P-field as the second field, the P-field can only use the I-field, which is the first field, for prediction. For example, field P
2
is coded only with reference to field I
1
.
A B-field is coded using the two most recently decoded temporally preceding and following I- and P-fields, that is, two temporally preceding and two temporally following fields. For example, field B
3
uses preceding fields I
1
and P
2
, and following fields P
5
and P
6
.
The display order and coding order of field structure video is shown in
FIG. 16
on the top and bottom rows, respectively.
Two particular tasks to be solved with the related art of the present invention are described next.
First Task to be Solved
When an MPEG video stream compressed using both intra and inter coding is edited by extracting a plurality of consecutive frames (scenes) from the bitstream and then combining a selected subset of the extracted scenes to produce a new frame sequence, the pictures referenced for predictive coding might be lost, resulting in pictures that cannot be reproduced.
The reason for this is explained next with reference to FIG.
17
. The arrows in
FIG. 17
indicate the correlations between pictures. When specific scenes, that is, pictures B
3
to B
11
, are extracted from this picture sequence, the links to referenced pictures indicated by the Xs are lost. In this example, the correlations between pictures I
1
and B
3
, between I
1
and P
4
, and between I
13
and B
11
, are lost.
While picture B
3
is coded with reference to picture I
1
, picture I
1
is not in the extracted sequence from B
3
to B
11
, and picture B
3
therefore cannot be reproduced. Pictures P
4
and B
11
also cannot be reproduced for the same reason.
Second Task to be Solved
FIG.
18
(
a
) shows an idealized decoder, referred to as a system target decoder
2
, under the MPEG-1 system, and related peripheral components. Encoded MPEG-1 data is input to buffer
1
at a constant bit rate, and data for one decoded picture is read from buffer
1
at a specific decode timing. Picture data is then output either directly or by way of a reordering buffer
3
. Differences in the display order and the coding order are absorbed by the reordering buffer
3
.
An MPEG-1 encoder codes video while varying the compression rate to adjust the code size (buffer control) by calculating the buffer capacity needed by the decoder during decoding to prevent both data overflow and data underflow states, that is, the data to be temporarily stored to the decoder buffer exceeds buffer capacity, or the buffer is temporarily depleted because the decoder reads data faster than it is stored to the buffer.
FIG.
18
(
b
) shows the change over time in the amount of data stored temporarily to the buffer (buffer fullness). Where the buffer fullness line drops perpendicularly to the x-axis in FIG.
18
(
b
) (i.e., has a slope of −□) indicates when one picture is read from the buffer
1
by system target decoder
2
. The height of the vertical drop in this line is indicative of the code size of one picture. As~ noted above, the code size depends on the picture type where
I-picture>P-picture>B-picture.
Data is input to the buffer
1
at a constant rate (slope is a constant positive value) in the periods between when the decoder reads picture data from the buffer.
The buffer
1
of decoder
2
will neither overflow nor underflow when decoding bu
Arimura Koji
Ikeda Jun
Kawaguchi Yuichi
Noridomi Kenichi
Britton Howard
Wenderoth , Lind & Ponack, L.L.P.
LandOfFree
Video editing apparatus, video editing method, and data... 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 editing apparatus, video editing method, and data..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Video editing apparatus, video editing method, and data... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3083446