Multiple description trellis coded quantization

Details Multiple description trellis coded quantization Multiple description trellis coded quantization

1998-05-06

2001-11-27

Chin, Stephen (Department: 2634)

Pulse or digital communications

Systems using alternating or pulsating current

Plural channels for transmission of a single pulse train

C375S340000, C714S792000

Reexamination Certificate

active

06324218

ABSTRACT:

BACKGROUND OF THE INVENTION
This invention relates to quantizers and, more particularly, to trellis-coded quantizers.
In recent years rate-scalable source coders have received growing attention. By selecting different sub-streams of the output of such coders, various levels of encoding rate and distortion can be achieved.
The ability to select different sub-streams is important in various applications. One such application, for example, may relate to receivers that are adapted to operate at one data rate under normal conditions, and adapted to accept a lower data rate when transmission conditions are adverse, while still producing a bona fide output, albeit, of lower fidelity.
A powerful source coding scheme for memoryless sources is trellis coded quantization. See, for example, M. W. Marcellin and T. R. Fischer, “Trellis coded quantization of memoryless and Gauss-Markov sources,”
IEEE Trans. Comm.
, vol. 38, pp. 82-93, January 1990. It has been shown that for a memoryless uniform source, trellis coded quantizers (TCQs) provide mean squared errors (MSEs) within 0.21 dB of the theoretical distortion bounds (for given rates). The performance of a trellis-coded quantization (TCQ) arrangement is much better than that of the best scalar quantizer (Lloyd-Max quantizer) at the same rate.
Rate scalability can be achieved with successive refinement, as well as with multiple descriptions. Successive refinement refers to the notion of a transmitter sending one stream of data which can decode the desired signal, albeit with lower fidelity, and one or more additional streams of data that refined the decoded output. Although until now, it has been thought that trellis coding does not lend itself to successive refinability, the aforementioned co-pending application discloses a successively refinable trellis quantizer.
Multiple description refers to the notion of a transmitter sending more than one description of a given sequence. A receiver accepting one of the descriptions can reproduce the signal with a certain fidelity, and a receiver accepting both descriptions can reproduce the signal with a higher fidelity. Unlike with successive refinement, either one of the multiple descriptions can be used to decode the signal.
Multiple description scenarios have a well-established history, but no present publications exist that disclose the use of multiple description coding in the context of trellis quantization. The challenge is to realize simple and effective multiple description arrangements for trellis coding.
SUMMARY
A multiple description TCQ arrangement is achieved by employing a trellis graph that is the tensor product, such as T
1
{circle around (X)} T
2
, of two trellis graphs, such as T
1
and T
2
. The codevectors of the tensor-product trellis, c
i
, are incorporated within trellis graph T
1
and also within trellis graph T
2
. The incorporation within trellis graph T
1
is effected by assigning the c
i
codevectors to sets, and by deriving therefrom codevectors for trellis graphs T
1
and T
2
. The actual values that these codevectors take on are arranged to insure certain distortion results. Specifically, the distortion measure of the tensor-product trellis is minimized, subject to the condition that the distortion measures of approximations made by decoders operating with the T
1
and T
2
trellises do not exceed a predetermined value. Consequently, an improved arrangement is realized in which two encoders are cooperatively generating separate trellis-coded descriptions of the input sequence. Three different fidelity levels can thus be achieved: a first when only a first description is employed, a second when the second description is employed, and a third (and highest level of fidelity) when both descriptions are employed. This allows for use of receivers that are responsive to different rates, or the use of receivers that have adaptable rates.


REFERENCES:
patent: 4677625 (1987-06-01), Betts et al.
patent: 4922507 (1990-05-01), Simon et al.
patent: 5052000 (1991-09-01), Wang et al.
H. A. Aksu and M. Salehi, “Multi-Stage Trellis Coded Quantization (MS-TCQ),”Proc. Conf. Inform. Sciences&Systems, Baltimore, Maryland, Mar. 1995.
M. W. Marcellin and T. R. Rischer, “Trellis Coded Quantization of Memoryless and Gauss-Markov Sources,”IEEE Trans. on Communications, vol. 38, No. 1, Jan. 1990, pp. 82-93.
P. J. Sementilli et al., “Progressive Transmission in Trellis Coded Quantization-Based Image Coders,”Conf. Image Processing, Santa Barbara, California, Oct. 1997.
H. Jafarkhani et al., “Entropy-Constrained Successively Refinable Scalar Quantization,”Proc. IEEE Data Compression Conference, Mar. 1997, pp. 337-346.
V. A. Vaishampayan, “Design of Multiple Description Scalar Quantizers,”IEEE Trans. on Information Theory, vol. 39, No. 3, May 1993; pp. 821-834.
V. A. Vaishampayan and J. Domaszewicz, “Design of Entropy-Constrained Multiple-Description Scalar Quantizers,”IEEE Trans. on Information Theory, vol. 40, No. 1, Jan. 1994, pp. 245-250.

Affiliated with

Also associated with

Rating

Multiple description trellis coded quantization does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Multiple description trellis coded quantization, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Multiple description trellis coded quantization will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-2616455