Television – Bandwidth reduction system – Data rate reduction
Patent
1991-07-23
1994-03-01
Kostak, Victor R.
Television
Bandwidth reduction system
Data rate reduction
375 27, H04N 713
Patent
active
052912841
DESCRIPTION:
BRIEF SUMMARY
BACKGROUND
I. Field of the Invention
The invention relates to predictive coding and decoding and particularly, but not exclusively, to an interframe video coding system employing transform coding.
II. Prior Art and Other Considerations
In a predictive encoding scheme, the receiver constructs a prediction of what the present value of a signal will be based upon its past values. Such schemes therefore generally employ a delay or store stage termed a predictor, which supplies past signal values (possibly weighted) to form a prediction of present signal values.
At the transmitter, a similar prediction is performed, and the difference between the predicted signal values and the actual signal values is encoded and transmitted as a difference signal to the receiver where it is added to the prediction to reconstitute an approximation of the actual signal.
Such systems may be analogue or digital; an example of a digital system is a digital video encoding system.
Video data is highly redundant, and it is therefore known to transmit video data employing an interframe predictor, which may also compensate for motion between frames. At the transmitter end of such a system, the difference signal values may be coded, in blocks of, for example, 8.times.8 picture elements (pixels), employing a spatial transform, and at the receiver the inverse transform is applied to decode the difference signal.
A commonly used transform is the Discrete Cosine Transform. The mathematical operations needed to perform it cannot be expressed exactly with a finite number of bits. Therefore practical implementations with different internal computing architectures produced, for example, by different manufacturers, can give slightly different results.
An inverse transform stage is located in both the coder and decoder. If these two do not produce exactly the same difference signal values at their outputs then the predictions at the encoder and decoder begin to diverge, and the resulting error becomes visible on the decoded picture.
This problem is tackled by specifying the maximum allowable error with respect to a (notional) very accurate version of the inverse transform.
Tests indicate that the two mean error parameters (the mean error on any pixel, and the mean error overall) are very important as they reflect bias, the presence of which results in a cumulatively increasing divergence between the pictures at the coder and decoder. The overall mean error thus becomes visible, as a steadily increasing or decreasing brightness of a picture block over many frames, or as an increasing colour cast.
Analogous problems occur in coding systems for other types of data (e.g. speech).
SUMMARY
According to one aspect of the invention, there is provided a predictive decoder comprising means for supplying a difference signal, means for supplying a prediction signal, and means for combining the said two signals to form an output signal dependent thereon, the prediction signal being derived from the output signal so as to depend upon previous portions of the difference signal, characterised in that means are provided for periodically alternating the operation of the decoder between first and second states, in which the cumulative effect of any time-averaged bias present in the difference signal is, respectively, to increase and to deplete the prediction signal, the respective lengths of the periods spent in the first and second states being such as to reduce the tendency of such time-averaged bias error to accumulate in the prediction signal.
This has the advantages of reducing long-term perceptually significant errors in the decoded signal, and thereby requiring lower precision, or granting greater error-immunity in the decoder.
According to other aspects, defined in the sub-claims, corresponding transmitters and systems, and a digital signal processing device useful therein, are provided.
In a preferred embodiment, particularly useful in block transform coding video systems, this invention reduces the mean errors by inverting the polarity of the data into
REFERENCES:
patent: 2820181 (1958-01-01), Bowman et al.
patent: 4317208 (1982-02-01), Araseki et al.
patent: 4377759 (1983-03-01), Ohhata et al.
patent: 4542516 (1985-09-01), Weirich
patent: 4549304 (1985-10-01), Weirich et al.
patent: 4593398 (1986-06-01), Millar
patent: 4727422 (1988-02-01), Hinman
patent: 4859883 (1989-08-01), Bradinal
patent: 4950929 (1990-08-01), Yan
patent: 5030858 (1991-07-01), Sekiya
IEEE Transactions on Communications, vol. COM-31, No. 3, (Mar. 1983), IEEE, (New York, US), R. Wilson et al. "Anisotropic nonstationary image estimation and its applications: Part II-Predictive image coding", pp. 398-406.
IEEE Transactions on Communications, vol. COM-21, No. 6, (Jun. 1973), IEEE, (New York, US), J. P. Agrawal et al. "Low bit rate differential PCM for monochrome television signals", pp. 706-714.
Proceedings of the Instituion of Electrical Engineers, vol. 124, No. 9, (Sep. 1977), J. A. Kitchen et al. "Subjective effects of d.p.c.m. errors and predictor leak on visual telephone pictures", pp. 737-741.
IEEE Transactions on Communications, vol. COM-29, No. 12, (Dec. 1981), IEEE, (New York, US), D. E. Troxel "Applicatin of pseudorandom noise to DPCM", pp. 1763-1767.
Carr Michael D.
Morrison David G.
British Telecommunications
Kostak Victor R.
LandOfFree
Predictive coding and decoding with error drift reduction does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Predictive coding and decoding with error drift reduction, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Predictive coding and decoding with error drift reduction will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-582306