Pulse or digital communications – Bandwidth reduction or expansion – Television or motion video signal
Reexamination Certificate
1997-02-21
2002-11-12
Rao, Andy (Department: 2613)
Pulse or digital communications
Bandwidth reduction or expansion
Television or motion video signal
C375S240130
Reexamination Certificate
active
06480542
ABSTRACT:
BACKGROUND OF THE INVENTION
The present standards for compressing video data, for example ISO/IEC 11172-2 (MPEG1) or ISO/IEC 13818-2 (MPEG2), combine the principles of prediction and transformation coding to produce so-called hybrid coding. The prediction is carried out here with the aid of a so-called differential pulse code modulation loop (DPCM loop), which generates differential frames by subtracting predicted video data from the original video data to be coded. The local correlations remaining in these differential frames between adjacent pixels are utilized with the aid of suitable transformation, preferably with the is aid of discrete cosine transformation (DCT). The transformation coefficients produced in this case are subsequently subjected to quantization and transmitted after entropy coding. The fundamental mode of operation of this compression method is known to the person skilled in the art from various publications, for example from D. J. Le Gall, “The MPEG video compression algorithm”, Signal Processing: Image Communication 4 (1992) 129-140. International Standard ISO/IEC 11172-2: Coding of moving pictures and associated audio, ISO/MPEG, 1993(E). Draft International Standard ISO/IEC 13818-2: Generic coding of moving pictures and associated audio, Mar. 25, 1994.
In order to improve the quality of the prediction in those picture areas in which moving objects occur, use is made of so-called movement-compensated prediction. The principles of the movement estimation required for this purpose and their application for the movement-compensated prediction are known to the person skilled in the art, for example from M. Bierling, “Displacement estimation by hierarchical block matching”, 3rd SPIE Symp. on Visual Commun., Cambridge, Mass., November 1988. Draft International Standard ISO/IEC 13818-2: Generic coding of moving pictures and associated audio, Mar. 25, 1994. So-called movement-compensated interpolation is provided in addition to the movement-compensated prediction in the case of the said standardized methods for compressing video data. In connection with MPEG terminology, movement-compensated interpolation is also designated as bidirectional prediction. However, since this designation may easily be confused with movement-compensated prediction, the term movement-compensated interpolation is preferred in the context of this application. The picture quality is decisively improved by the use of movement-compensated interpolation, since it allows a satisfactory treatment of masked picture areas and contributes to improved noise suppression.
A distinction is made between three differently coded types of frames. So-called I-frames are transmitted without any chronological prediction, but rather are subjected only to intra-frame coding, preferably DCT coding with subsequent quantization of the transformation coefficients. In the context of this patent application, “intra-frame coding” is to be understood quite generally as any method which is suitable for treating local correlations in video data. The so-called P-frames are predicted with the aid of the DPCM loop from chronologically preceding I-frames or P-frames (forward prediction). The difference between the predicted frame and the actual frame is subjected to inter-frame coding, preferably DCT transformation with subsequent quantization of the transformation coefficients. So-called B-frames, which are also designated as “interpolated frames” in the context of the present patent application, are chronologically situated between an I-frame and a P-frame or between two P-frames. B-frames are determined by means of “bidirectional” movement-compensated interpolation from a chronologically preceding I- or P-frame and a chronologically succeeding I- or P-frame. In this case, the expressions “chronologically”, “succeeding” and “preceding” do not refer to the order in which these frames are transmitted in the data stream of compressed frames, rather they refer to the order in which these frames are recorded/reproduced. In the same way as P-frames, B-frames, too, are coded in the form of quantized transformation coefficients of a differential frame.
In the case of currently known implementations, the reconstruction of a B-frame by means of movement-compensated interpolation from a chronologically preceding I- or P-frame and a chronologically succeeding I- or P-frame necessitates the provision of the two reference frames (which are also occasionally designated as support frames in the literature) in fully decoded form. Therefore, two fully decoded reference frames (i- or P-frames) have to be stored in a frame store in the case of the methods belonging to the prior art for carrying out movement-compensated interpolation. The re-interlacing during the video output requires further storage capacity. The overall required memory is a decisive cost factor in the hardware used for decoding and encoding. A reduction in the storage capacity required is therefore desirable.
SUMMARY OF THE INVENTION
The invention is based on the object of specifying a method for decoding compressed video data with a reduced memory requirement. This object is achieved according to the invention by a method for decoding compressed video data with a reduced memory requirement In this case, only memory space taken up by one fully decoded first reference frame and one compressed second reference frame is required for storing the reference frame data when carrying out interpolation. A reduction in the memory requirement by, for example, at least 2.6 Mbits can be expected in a typical application (MPEG2/Main Level/Main Profile) with an assumed typical channel data rate of 10 Mbit/s. This reduction in the necessary storage capacity means that integration of the codec on a chip is substantially facilitated or even becomes possible in the first place, and the implementation becomes decisively more economical thereby. However, a prerequisite for this is an adequately fast hardware logic unit for the decoding/encoding or, if this is not the case, multiple replication of the corresponding hardware logic circuits.
The core of the invention is not to store the uncompressed data for the second reference frame but rather to keep in store only the compressed data in the (correspondingly enlarged) input buffer of the decoder. For the purpose of reconstructing a B-frame between two P- or I-frames (generally between two reference frames), only those picture areas which are instantaneously required for the video output are decoded from the compressed data stream. In the case of MPEG, these are, for example, those blocks of the second reference frame which overlap with the macroblock to be determined of the interpolated frame, taking into account a displacement corresponding to the movement compensation. In the example of frame prediction, 9 blocks each of 8 times 8 luminance pixels of the first reference frame are required for the luminance pixels of a macroblock to be decoded of the second reference frame.
The area of application of the invention includes not only purely decoding but also encoding video data, because an encoder which operates in accordance with the principle of prediction by DPCM coding always contains a decoder, which decodes the coded data again, as a result of which it becomes possible in the first place to calculate differential frames in a DPC loop D. J. Le Gall, “The MPEG video compression algorithm”, Signal Processing: Image Communication 4 (1992) 129-140.
An advantageous embodiment of the invention is one in which, for the purpose of reconstructing and outputting the second reference frame, after the interpolated frames are output, the memory space provided for the output of these frames is initially used to store the second reference frame in an uncompressed form.
It is further advantageous if
after this the store for the first reference frame is used to store the remainder of the second reference frame in an uncompressed form,
the remainder of the store for the first reference frame which is no longer required is used as an output memory for the
Geib Heribert
Hammer Bernard
Pandel Jürgen
Prange Stefan
Rao Andy
Schiff & Hardin & Waite
Siemens Aktiengesellschaft
LandOfFree
Method for decoding compressed video data with a reduced... 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 for decoding compressed video data with a reduced..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for decoding compressed video data with a reduced... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2981180