Television – Image signal processing circuitry specific to television – Motion dependent key signal generation or scene change...
Reexamination Certificate
2000-03-06
2003-05-13
Miller, John (Department: 2614)
Television
Image signal processing circuitry specific to television
Motion dependent key signal generation or scene change...
C348S449000, C348S452000, C348S558000
Reexamination Certificate
active
06563550
ABSTRACT:
BACKGROUND
The present invention relates to the field of video processing, and more particularly to methods and apparatuses for detecting the presence of progressive frames in a sequence of video fields.
A telecine is a well-known apparatus that converts a motion picture film into a video format for display on a device such as a television. Both motion picture film and video create the illusion of moving pictures by sequentially displaying a series of still image frames that represent the image at corresponding sequential instants of time. The conversion process must take into account differences in display format as well as differences in image frame rate.
Considering display format first, each portion of a motion picture film frame is displayed simultaneously to the user. By contrast, video images are created by sequentially “painting” dots, called “pixels”, onto a suitable screen, such as a cathode ray tube (CRT). The pixels are supplied in an order that draws horizontal lines on the screen, one line at a time. This is performed at a fast enough rate such that the viewer does not experience the individual pixels, but rather sees the combination of displayed pixels as a single image. The lines of horizontal pixels may be drawn in several different ways. If a progressive scan order is used, the lines are supplied in sequence from, for example, top to bottom. Alternatively, an interlaced scan order can be used, wherein the image frame (which comprises the totality of scan lines to be displayed for the given frame) is divided into even and odd fields. The even field comprises all of the even numbered scan lines, and the odd field comprises all of the odd numbered scan lines. In an interlaced video display system, an entire even field is supplied to the screen, followed by the odd field. This pattern is then repeated for each frame to be displayed.
Considering now differences in display rates, standard motion picture film is shot at a rate of 24 frames per second (fps). By contrast, current existing television systems, such as those operating in accordance with National Television Standards Committee (NTSC), Phase Alternation Line (PAL) and High Definition (HD) television standards, have video frame rates that include 24, 30 and 25 fps.
In converting from a 24 fps film image to a 30 fps video image, the frame rate must increase by 25% so that when the film frames are played back as video they transpire in the same 1 second that they would have on film. This can be accomplished by outputting 2.5 video frames for every 2 film frames. Since a telecine typically needs to generate an interlaced video output comprising alternating odd and even fields, this rate difference equates to outputting 5 video fields for every 4 film fields. One way to accomplish this is by extracting 2 fields from one film frame, and 3 fields from the next. In the 3-field sequence (henceforth referred to as the field “triplet”), the first and third fields are derived from the same film frame, and are therefore identical. The specific conversion from 24 to 30 fps is called 2:3 pulldown (also referred to as 3:2 pulldown). This process is illustrated in FIG.
1
(
a
). The top strip shows a film sequence
101
of a ball moving from left to right across the frame. Each of these frames may be considered to be in a “progressive” format since, if separated into odd and even fields, both fields will have been captured at the same instant in time. In contrast, the interlaced NTSC video format has odd and even fields that are captured {fraction (1/60)} of a second apart.
The second strip in FIG.
1
(
a
) shows the output
103
of the 3:2 pulldown telecine process. In the figure, the label “Ao” denotes the first odd video field, the label “Ae” denotes the first even video field, the label “Bo” denotes the second odd video field, and so on. Each successive pair of odd and even fields constitutes one video frame, capable of being displayed on an interlaced video display device. Note that as a result of the 3:2 pulldown process which selectively duplicates certain fields, the field “Co” is not in the same video frame as that constituted by the fields “Bo” and “Be” even though the field “Co” originated from the same film frame as the fields “Bo” and “Be”. Likewise, although “Co” and “Ce” are in the same video frame, they originated in different film frames.
It is useful to have the capability of detecting whether telecine processing or other processing (e.g., computer-generated video, which would also be in the form of video fields that can be combined to form progressive video frames) has been employed in the generation of video material, and if so, to be able to identify those fields in the sequence that have been “pulled down”. How this information is utilized depends on the type of application that is to take place. For example, when a telecine processed video image is to be compressed (i.e., so that the image can be represented in fewer digital bits) the repeated frames are simply discarded and the compression routine supplies the appropriate field replication markings. In another example, when a telecine processed video image is to undergo interlace-to-progressive format conversion, no processing to generate a synthetic field (either via interpolation or motion compensation techniques) takes place, and the action is merely to bundle back together the appropriate fields into their original progressive frame state. Thus a progressive frame may sometimes be reconstructed by pairing a source field with the field before, or with the field after, or sometimes with either. This is illustrated in FIG.
1
(
b
). The first strip in FIG.
1
(
b
) pairs each field with the immediately preceding field, and is thus labeled “Field−1”. The next strip in FIG.
1
(
b
) pairs each field with the immediately succeeding field, and is thus labeled “Field+1”. Note that while some pairings yield the original progressive film frames (e.g., “AeAo” and “BeBo”), other pairings yield incorrect results (e.g., “BoAe” and “CeCo”). By correctly selecting the pairings which yield the original progressive film frames, a 60 fps progressive output can be achieved as shown in the last strip in FIG.
1
(
b
).
Note in the last strip in FIG.
1
(
b
) that there are two frames that have two correct pairings: a first frame that could either be “BeBo” or “BeCo”, and a second frame that could either be “EoDe” or “EoEe”. This is a characteristic of a field triplet. The center field can be paired with either the immediately preceding field or with the immediately succeeding field, since both are identical. In the video compression application, the second of the identical fields (“Co” and “Ee”) would be labeled as being replicated and would not be subjected to the lengthy compression algorithm.
In a perfect (unedited) 3:2 sequence, replicated fields occur once every fifth field. Prior techniques for pulldown detection utilize this sequence repetition for identifying the field triplet location, and hence the 3:2 pattern. If the 3:2 sequence is not perfect, as is the case with edited material, the pattern is broken. The break in the sequence pattern cannot be detected until the location of the next field triplet arrives and the expected field replication is not found. Thus these conventional techniques must buffer the fields between the triplets or suffer the consequences of incorrect pairing. Decisions have to then be made on how to treat these buffered fields without the knowledge of where they fall in the 3:2 sequence, all of which leads to processing latency.
An example of edited 3:2 material is shown in FIGS.
2
(
a
)-(
e
). FIG.
2
(
a
) illustrates a telecined 3:2 pulldown sequence
201
with edited frames (Bo, Be) and (Do, De) shown in cross-hatch. FIG.
2
(
b
) shows the same sequence
201
′ with the edited frames removed. As in FIG.
1
(
b
), the FIGS.
2
(
c
) and
2
(
d
) show the results of pairings with the immediately preceding field (Field−1) and the immediately succeeding field (Field+1). Again, the strip depicted in FIG.
Kahn Barry A.
Sanders Albert R.
Miller John
Teranex, Inc.
Yenke Brian
LandOfFree
Detection of progressive frames in a video field sequence does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Detection of progressive frames in a video field sequence, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Detection of progressive frames in a video field sequence will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3064617