Pulse or digital communications – Bandwidth reduction or expansion
Reexamination Certificate
1998-04-16
2001-02-13
Kelley, Chris S. (Department: 2713)
Pulse or digital communications
Bandwidth reduction or expansion
C375S240120
Reexamination Certificate
active
06188725
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to a method and apparatus for high-efficiency encoding to efficiently convert an interlaced type of video signal into a stream of compressed code, for the purpose of transmission or storage. In particular, the invention relates to encoding processing which uses bidirectional motion prediction encoding, applied to an interlaced type of video signal.
In the following, the term “picture” will be used as a general term for referring to the contents of afield of an interlaced video signal, or a frame of aprogressive scanning (i.e., non-interlaced) type of video signal.
DESCRIPTION OF THE PRIOR ART
A method of high-efficiency video encoding for an interlaced type of video signal is known whereby one in every m successive frames (where m is an integer of 2 or more) is encoded either independently by internal encoding or by unidirectional predictive encoding, while the remaining frames (referred to as the B frames)are encoded by bidirectional predictive encoding using preceding and succeeding ones of the aforementioned specific frames (i.e., I or P frames).
Such predictive encoding of a video signal is now well known in the art, being described for example in Japanese patent laid-open number HEI 2-192378, of the assignee of the present invention, etc. The technique is also used with the MPEG-1 system (ISO/IEC-11172), and the MPEG-2 system (ISO/IEC-13818).
It can be understood from the above that with such a method three different picture types are established, in accordance with the form of encoding that is applied, i.e. the I frames, the P frames and the B frames, and that only the I frames and P frames are used as reference frames for encoding.
Assuming that the frame period is 1/30 second, the first and second fields of each frame of the interlaced video signal are time-displaced by 1/60 second, and are also mutually displaced by one scanning line position, in the vertical direction of the picture. As a result, appropriate prediction for the purpose of picture encoding cannot be achieved by using a simple inter-frame prediction method. In such a case, a method is used such as with the NPEG-2 standard etc., whereby processing is performed in units of fields, with a plurality of fields being used to constitute a reference picture, or whereby processing is basically performed in units of frames, but with prediction being switched to perform local prediction in units of fields when necessary.
In particular, in the case of the MPEG-2 standard, each of the aforementioned picture types (i.e., I, P, B) must be established in units of frames. That is to say, even if prediction processing is performed in units of fields, the I-pictures and B-pictures must each be set as respective consecutive pairs of fields.
Whichever method is used, when motion is detected between the contents of successive fields, prediction is performed in units of fields. In that case, due to the scanning line configuration of the interlaced signal, large amounts of aliasing components will be generated and encoded, due to the one-line vertical displacement between successive interlaced fields. As a result, even if the picture motion only consists of a parallel shift of the picture, comparatively large amounts of prediction error values will be generated and encoded, so that the predictive encoding process will generate excessively large amounts of code, i.e. the objective of achieving a high efficiency of data compression through encoding will not be met.
FIG. 5
shows an example of the configuration of a prior art type of video encoding apparatus which uses bidirectional prediction for encoding the B fields. It will be assumed that prediction is performed in units of fields, but that the I, P and B picture types are established in units of interlaced frames as described above.
The interlaced video signal which is input to the video input terminal
7
is supplied to the input signal selection switch
56
which is controlled to operate in synchronism with successive fields of the input video signal such that the I and P frames are supplied to a subtractor
51
while the B frames are supplied to a frame delay element
61
. It should be noted that the term “video signal” as used herein signifies a digital video signal.
One out of every m successive frames of the input video signal is selected as an I or a P frame, (where m will in general have a value of 2 or 3). The proportion of I frames to P frames is a matter of design choice. The subtractor
51
subtracts an inter-picture prediction signal (i.e., consisting of successive predicted values for respective pixels of a frame) that is produced by an inter-picture prediction section
57
from the I or P frame signal which is supplied thereto, and supplies the resultant difference values, i.e. prediction error values, to a DCT section
52
. The DCT section
52
performs DCT (Discrete Cosine Transform) conversion processing on successive sets of prediction error values which correspond to respective blocks of 8×8 (or 16×16) pixels of a picture, and the transform coefficients thereby obtained are supplied to a quantizer
53
. The quantizer
53
performs quantization of the coefficients, using a predetermined quantization step size, and the resultant fixed-length encoded coefficients are supplied to a variable-length encoder
54
and to a dequantizer
55
.
The variable-length encoder
54
performs array conversion of the 2-dimensional 8×8 sets of coefficients into a 1-dimensional sequence, using zig-zag sequence processing, and encodes the result by Huffman encoding, i.e. using the numbers of runs of coefficient values of zero or of coefficient values other than zero. The resultant code sequences into which the I and P frames have been respectively converted are multiplexed with the code sequences which are obtained for the B frames, by the multiplexer
13
, and the resultant code stream is supplied to the code output terminal
14
.
The dequantizer
55
and the inverse DCT section
60
perform the inverse processing to that of the quantizer and the DCT section
52
, to thereby reproduce the inter-picture prediction error values, and the values thus obtained are added to the prediction signal by the adder
59
, to obtain values expressing successive reconstructed pictures, which are supplied to the picture memory
58
. The reconstructed pictures which are thus stored in the picture memory
58
are thereafter read out and supplied to the inter-picture prediction section
57
at appropriate timings.
The inter-picture prediction section
57
generates different prediction signals in accordance with respective types of picture (i.e., I, P or B), supplies the prediction signals derived for the I and P frames to the subtractor
51
, and supplies the prediction signals derived for the B frames to one input of the subtractor
17
.
Since no prediction is performed for an I frame, the prediction signal values for an I frame are always zero. In the case of a P frame, the prediction signal is obtained based on a preceding I or P frame. In the case of a B frame, the prediction signal is obtained based on preceding and succeeding I or P frames.
With this method, since prediction is performed in units of fields, both even fields and odd fields of a reconstructed frame may be used as reference pictures. Of these, the field which results in the smallest amount of prediction error is used as a reference for deriving the prediction signal.
When a B frame signal is selected by the switch
56
, the frame delay section
61
applies a delay of (m−1) frames, and the delayed B frame signal is then supplied to the subtractor
17
. Since the picture type is established in units of frames, the delay must be established in units of frame periods. The resultant delayed picture signal (i.e. successive pixel values) is input to the subtractor
17
in synchronism with predicted values supplied from the inter-picture prediction section
57
, to obtain respective prediction error values for the B frame, which are encoded by the DCT sect
Anderson Kill & Olick P.C.
Kelley Chris S.
Victor Company of Japan Ltd.
Wong Allen
LandOfFree
Interlaced video signal encoding and decoding method, by... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Interlaced video signal encoding and decoding method, by..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Interlaced video signal encoding and decoding method, by... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2600119