Television – Image signal processing circuitry specific to television – With details of static storage device
Reexamination Certificate
2000-11-14
2003-11-04
Hsia, Sherrie (Department: 2614)
Television
Image signal processing circuitry specific to television
With details of static storage device
C348S716000, C375S240260
Reexamination Certificate
active
06642967
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to video data formatting and storage.
2. Description of the Prior Art
A video data storage system has been proposed in which an input video signal having a so-called “long GOP” format (i.e. a group of pictures format related to the MPEG format, involving P and/or B frames) is stored as an I-frame only signal. The stored signal is then returned to the long-GOP form for output.
As well as storing the I frame data (which allows frame-accurate editing of the video signal) some extra data or metadata associated with the video signal could be stored as well. The metadata could provide information about how the signal was coded in long GOP form and so could be used to assist the output encoder to give better results when the long GOP output signal is generated.
Two example possibilities for the metadata are: (a) the coding decisions used in generating the long GOP data, for example, vectors, Q (quantization parameter) values etc., or (b) the actual long GOP bitstream itself. In case (a), the coding decisions would guide the output encoder to make similar or identical decisions to generate the output long GOP bitstream, and in case (b), when no editing has taken place the original long GOP data could be output directly so that an I-frame to long GOP recoding process is needed only at or near to edit points.
Of course, in either case (a) or case (b), the amount of metadata concerned can be highly variable.
In case (a), the amount of data representing the coding decisions can vary from frame to frame. Apart from any other reason, an I frame does not usually have any associated vectors, a P frame has one set of vectors and a B frame may have two sets.
In case (b), the number of bits per frame or per GOP of a long GOP bitstream is not generally regulated to a fixed amount, but instead is controlled in the MPEG system by the fullness of a so-called “virtual buffer”, an average desired data rate over time and by the difficulty of encoding the picture content itself. So, the quantity of data can vary dramatically from frame to frame. Also, it is very difficult to synchronise to the long-GOP data on a frame-by-frame basis.
In contrast, the I frame data recorded by the video data storage system is generally set up to aim towards a fixed data amount per frame. This is particularly important in a tape-based system where the capacity of the storage medium is strictly limited. If any metadata is recorded, the amount of metadata allocated to each I-frame must be subtracted from the available data capacity per I-frame to give a remaining target data amount for the Iframe.
The problem therefore is to store the variable-size metadata within a fixed data allocation per I-frame, and to do so in a manner which allows synchronism between the metadata and the frames of the stored I-frame signal.
Two possibilities for handling this problem are therefore:
(i) to record an average amount of either the coding decision data (a) or the long GOP data (b) with each frame of I-frame data. This gives a predictable data quantity remaining for each I frame. This scenario is shown schematically in
FIG. 1
of the accompanying drawings, in which each fixed-size data allocation
10
contains a predetermined quantity of I-frame data
20
and a predetermined quantity of metadata
30
. However, this means that the data recorded with each I-frame recorded by the video data storage system may bear no relation to the particular image represented by that I-frame.
(ii) to record data actually associated with each I frame alongside that I frame. This is only really possible for the decision data (a), as there is not an easily derivable relationship between long-GOP data (b) and individual frames of the video signal. This scenario is illustrated schematically in
FIG. 2
of the accompanying drawings, in which each fixed size data allocation
10
contains a variable quantity of I-frame data
22
and a variable, complementary, quantity of metadata
32
. This proposal means that there is a good correlation between metadata and the I-frame data if the signal is cut or otherwise edited. However, it requires the target bit rate for each I-frame to be altered from I-frame to I-frame. Also, because the amount of data varies dramatically, some I-frames may be left with insufficient data capacity for adequate coding.
In summary, there is a need for a formatting and/or storage arrangement which allows variable-length metadata which may, for example, comprise or be derived from a long GOP encoding of a video signal.
SUMMARY OF THE INVENTION
This invention provides video data formatting apparatus for formatting video data representing successive pictures for a data handling channel having a predetermined data capacity per picture, the apparatus comprising:
means for receiving an input video signal representing successive pictures, the input video signal having associated with it at least data defining at least some of the coding decisions made during an encoding of pictures represented by the input video signal into a compressed form having group-of-pictures (GOP) format including at least one inter-picture encoded picture;
means for converting the input video signal into an intermediate compressed video signal, the intermediate compressed video signal having a GOP format in which each GOP contains fewer pictures than a GOP associated with the input video signal;
means for deriving a metadata signal from the input video signal, the metadata signal indicating at least the data defining at least some of the coding decisions;
means for generating a data quantity allocation to control the transcoding into the intermediate video signal, whereby each picture of the intermediate video signal is transcoded so as not to exceed a respective data quantity allocation, in which the generating means calculates the data quantity allocation for each picture to be substantially equal to:
the predetermined data capacity per picture
less
the quantity of metadata for the input video signal GOP containing that picture divided by the number of pictures (n) in that input video signal GOP.
Further respective aspects and features of the invention are defined in the appended claims.
The invention is particularly applicable to a system in which the intermediate compressed video signal has a GOP format comprising only intra-picture encoded pictures.
Preferably the apparatus comprises means for generating data packets (e.g. for recording on a storage medium such as a tape medium) each comprising: an encoded picture of the intermediate compressed video signal; and a substantially I
portion of the metadata signal associated with the input video signal GOP from which that picture was derived.
In embodiments of the invention, the solution provided to the problem described above is to determine the quantity of metadata—for example, of type (a) or (b)—associated with a particular GOP of the long GOP signal, and then to record that metadata in substantially equal segments with each of the I-frames corresponding to that GOP.
This allows the metadata and I-frame data to be associated with one another on a GOP by GOP basis. The start of each GOP of metadata can be established by standard synchronising codes within a GOP.
This solution recognises that long-GOP metadata of either type is close to useless if an edit has been made during the GOP, so there is no point having a correlation frame by frame. It also allows the target bit rate to be set once for all of the frames corresponding to a GOP.
The intermediate video signal preferably comprises only I-frames, for convenience of editing. This would generally imply a GOP length of 1, though that need not always be the case.
Preferably the metadata is formatted to the intermediate video signal in such a manner that it is resynchronised at GOP boundaries—in the preferred embodiments these would be boundaries of the long-GOPs, but in a system where the two GOP lengths shared a common multiple it would take place at instances of that common multiple of picture
Frommer William S.
Frommer & Lawrence & Haug LLP
Hsia Sherrie
Savit Glenn F.
Sony United Kingdom Limited
LandOfFree
Video data formatting and storage employing data allocation... 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 data formatting and storage employing data allocation..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Video data formatting and storage employing data allocation... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3174352