Multiresolution Vector Quantization

Effros, Michelle; Dugatkin, Diego

doi:10.1109/tit.2004.838381

Cited by 21 publications

(23 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This is consistent with our intuition that performance degrades at higher resolutions for greedily designed codes and suggests that the performance penalty associated with using greedily grown TSVQs [20] rather than jointly optimized multiresolution vector quantizers [5], [6] may be large. Proving such a result would require a tight bound on the rate losses studied in Theorem 7.…”

Section: Resultssupporting

confidence: 86%

“…ECAUSE of their ability to satisfy varying bandwidth, computation, and performance constraints with a single code, multiresolution source codes (MRSCs) are playing an increasingly important role in research and in practice (e.g., [1]- [6] (b/s) to describe with distortion and then uses an additional b/s to refine the description to distortion , as shown in Fig. 1.…”

mentioning

confidence: 99%

See 1 more Smart Citation

Improved bounds for the rate loss of multi-resolution source codes

Feng

Effros²

Proceedings. 2001 IEEE International Symposium on Information Theory (IEEE Cat. No.01CH37252)

Self Cite

View full text Add to dashboard Cite

Abstract-In this paper, we present new bounds for the rate loss of multiresolution source codes (MRSCs). Considering an -resolution code, the rate loss at the th resolution with distortion is defined as = ( ), where is the rate achievable by the MRSC at stage . This rate loss describes the performance degradation of the MRSC compared to the best single-resolution code with the same distortion. For two-resolution source codes, there are three scenarios of particular interest: i) when both resolutions are equally important; ii) when the rate loss at the first resolution is 0 ( 1 = 0); iii) when the rate loss at the second resolution is 0 ( 2 = 0). The work of Lastras and Berger gives constant upper bounds for the rate loss of an arbitrary memoryless source in scenarios i) and ii) and an asymptotic bound for scenario iii) as 2 approaches 0. In this paper, we focus on the squared error distortion measure and a) prove that for scenario iii) 1 1 1610 for all 2 1 ; b) tighten the Lastras-Berger bound for scenario ii) from 2 1 to 2 0 7250; c) tighten the Lastras-Berger bound for scenario i) from 1 2 to 0 3802, 1 2 ; and d) generalize the bounds for scenarios ii) and iii) to -resolution codes with 2. We also present upper bounds for the rate losses of additive MRSCs (AMRSCs). An AMRSC is a special MRSC where each resolution describes an incremental reproduction and the th-resolution reconstruction equals the sum of the first incremental reproductions. We obtain two bounds on the rate loss of AMRSCs: one primarily good for low-rate coding and another which depends on the source entropy.Index Terms-Additive successive refinement code, progressive transmission, tree-structured vector quantizer.

show abstract

Section: Resultssupporting

confidence: 86%

mentioning

confidence: 99%

Improved bounds for the rate loss of multi-resolution source codes

Feng

Effros²

Proceedings. 2001 IEEE International Symposium on Information Theory (IEEE Cat. No.01CH37252)

Self Cite

View full text Add to dashboard Cite

show abstract

“…of MR coding include tree-structured VQ [15], [16], locally optimal fixed-rate multiresolution SQ (MRSQ) [17], variable-rate MRSQ [18], and locally optimal variable-rate MRSQ and multiresolution VQ (MRVQ) [19], [20]. Examples of MD coding include locally optimal two-description fixed-rate multiple-description VQ (MDVQ) [21] and locally optimal two-description fixed-rate [22] and entropy-constrained [23] MDSQ.…”

Section: B Network Vector Quantizers (Nvqs)mentioning

confidence: 99%

“…Following the entropy-constrained coding tradition (see, for example, [14], [20], [32], [33]), we describe lossy code design as quantization followed by entropy coding. The only loss of generality associated with the entropy-constrained approach is the restriction to solutions lying on the lower convex hull of achievable entropies and distortions.…”

Section: B Network Vector Quantizers (Nvqs)mentioning

confidence: 99%

Network Vector Quantization

Fleming

Zhao

Effros

2004

IEEE Trans. Inform. Theory

Self Cite

View full text Add to dashboard Cite

Abstract-We present an algorithm for designing locally optimal vector quantizers for general networks. We discuss the algorithm's implementation and compare the performance of the resulting "network vector quantizers" to traditional vector quantizers (VQs) and to rate-distortion (R-D) bounds where available. While some special cases of network codes (e.g., multiresolution (MR) and multiple description (MD) codes) have been studied in the literature, we here present a unifying approach that both includes these existing solutions as special cases and provides solutions to previously unsolved examples.

show abstract

“…Prior work on fixed-rate multiresolution scalar quantizer design likewise includes both iterative descent algorithms [7], [8], [9] and shortest path algorithms [2], [10]. 3 As in the corresponding multiple description codes, iterative descent techniques generalize easily to vector quantizer design, but they are susceptible to local optimality problems and their complexity is difficult bound.…”

Section: Introductionmentioning

confidence: 99%