On the structure of optimal entropy-constrained scalar quantizers

Entropy-constrained scalar quantizers (ECSQ) with the mean-squared error (MSE) distortion measure are widely used in the field of image compression. The design is based on the iterative Lloyd clustering algorithm, which ensures only a locally optimum quantizer and is highly dependent on the training parameters. We propose an alternative quasi-linear time (in the number of training samples) design that guarantees a near globally optimal ECSQ and we provide the upper bound on the loss of optimality. Our simulations, using l.l.d. Gaussian samples and a set of aerial images, indicate that our algorithm leads to a better rate-distortion performance than the Lloyd algorithm with a comparable design complexity.

This paper presents solutions to the entropy-constrained scalar quantizer (ECSQ) design problem for two sources commonly encountered in image and speech compression applications: sources having the exponential and Laplacian probability density functions. We use the memoryless property of the exponential distribution to develop a new noniterative algorithm for obtaining the optimal quantizer design. We show how to obtain the optimal ECSQ either with or without an additional constraint on the number of levels in the quantizer. In contrast to prior methods, which require a multidimensional iterative solution of a large number of nonlinear equations, the new method needs only a single sequence of solutions to one-dimensional nonlinear equations (in some Laplacian cases, one additional two-dimensional solution is needed). As a result, the new method is orders of magnitude faster than prior ones. We show that as the constraint on the number of levels in the quantizer is relaxed, the optimal ECSQ becomes a uniform threshold quantizer (UTQ) for exponential, but not for Laplacian sources. We then further examine the performance of the UTQ and optimal ECSQ, and also investigate some interesting alternatives to the UTQ, including a uniform-reconstruction quantizer (URQ) and a constant dead-zone ratio quantizer (CDZRQ).

A new coding scheme based on the scalar-vector quantizer (SVQ) is developed for compression of medical images. SVQ is a fixed-rate encoder and its rate-distortion performance is close to that of optimal entropy-constrained scalar quantizers (ECSQs) for memoryless sources. The use of a fixed-rate quantizer is expected to eliminate some of the complexity issues of using variable-length scalar quantizers. When transmission of images over noisy channels is considered, our coding scheme does not suffer from error propagation which is typical of coding schemes which use variable-length codes. For a set of magnetic resonance (MR) images, coding results obtained from SVQ and ECSQ at low bit-rates are indistinguishable. Furthermore, our encoded images are perceptually indistinguishable from the original, when displayed on a monitor. This makes our SVQ based coder an attractive compression scheme for picture archiving and communication systems (PACS), currently under consideration for an all digital radiology environment in hospitals, where reliable transmission, storage, and high fidelity reconstruction of images are desired.

The paper considers scheduling jobs online to minimize the objective $$\sum _{i \in [n]}w_ig(C_i-r_i)$$ , where $$w_i$$ is the weight of job i, $$r_i$$ is its release time, $$C_i$$ is its completion time and g is any non-decreasing convex function. It is known that the clairvoyant algorithm Highest-Density-First (HDF) is $$(2+\epsilon )$$ -speed O(1)-competitive for this objective on a single machine for any fixed $$ 0< \epsilon < 1$$ (Im et al., in: ACM-SIAM symposium on discrete algorithms, pp 1254–1265, 2012). In this paper, we give the first non-trivial results for this problem when g is a non-decreasing convex function and the algorithm must be non-clairvoyant. More specifically, our results include: The paper gives the first non-trivial upper-bound on multiple machines even if the algorithm is allowed to be clairvoyant. All performance guarantees above hold for all non-decreasing convex functions gsimultaneously. The positive results are supplemented by almost matching lower bounds. We show that any algorithm that is oblivious to g is not O(1)-competitive with speed augmentation less than 2 on a single machine. Further, any non-clairvoyent algorithm that knows the function g cannot be O(1)-competitive with speed augmentation less than $$\sqrt{2}$$ on a single machine or $$(2-\frac{1}{m})$$ on m identical machines.

Previous research in the area of active noise control has shown that the sound radiation from a vibrating body can be completely controlled using layers of simple and dipole sources located on a surface that completely surrounds the vibrating body. If complete control of the sound radiation could be achieved with simple sources only, the implementation of the control strategy will be considerably simplified because simple sources can be approximated with conventional loudspeakers. It is shown that complete control can be achieved when the simple source layer is very near the boundary surface of the vibrating body (assuming that the fluid coupling between the vibrating body and the surrounding fluid is negligible). When the simple source layer is not near the surface of the vibrating body, the control strategy fails at the resonance frequencies of the volume of fluid between the surface of the vibrating body and the layer of simple sources. The effect of approximating the continuous source distribution by a discrete source distribution is examined, and it is shown that the discrete sources can control the sound radiation effectively as long as there are an adequate number of sources distributed over the surface of the vibrating body.

Let A be an operator with the polar decomposition A = U|A|. The Aluthge transform of the operator A, denoted by ?, is defined as ? = |A|1/2U |A|1/2. In this paper, first we generalize the definition of Aluthge transformfor non-negative continuous functions f,g such that f(x)g(x) = x (x ? 0). Then, by using this definition, we get some numerical radius inequalities. Among other inequalities, it is shown that if A is bounded linear operator on a complex Hilbert space H, then h (w(A)) ? 1/4||h(g2 (|A|)) + h(f2(|A|)|| + 1/2h (w(? f,g)), where f,g are non-negative continuous functions such that f(x)g(x) = x (x ? 0), h is a non-negative and non-decreasing convex function on [0,?) and ? f,g = f (|A|)Ug(|A|).

We consider the problem of side-information scalable (SI-scalable) source coding, where the encoder constructs a two-layer description, such that the receiver with high quality side information will be able to use only the first layer to reconstruct the source in a lossy manner, while the receiver with low quality side information will have to receive both layers in order to decode. We provide inner and outer bounds to the rate-distortion (R-D) region for general discrete memoryless sources. The achievable region is tight when either one of the decoders requires a lossless reconstruction, and when the distortion measures are degraded and deterministic. Furthermore, the gap between the inner and the outer bounds can be bounded by certain constants when the squared error distortion measure is used. The notion of perfect scalability is introduced, for which necessary and sufficient conditions are given for sources satisfying a mild support condition. Using SI-scalable coding and successive refinement Wyner-Ziv coding as basic building blocks, we provide a complete characterization of the rate-distortion region for the important quadratic Gaussian source with multiple jointly Gaussian side informations, where the side information quality is not necessarily monotonic along the scalable coding order. A partial result is provided for the doubly symmetric binary source under the Hamming distortion measure when the worse side information is a constant, for which one of the outer bounds is strictly tighter than the other.

Optimal zero-delay coding (quantization) of a vector-valued Markov source driven by a noise process is considered. Using a stochastic control problem formulation, the existence and structure of optimal quantization policies are studied. For a finite-horizon problem with bounded per-stage distortion measure, the existence of an optimal zero-delay quantization policy is shown provided that the quantizers allowed are ones with convex codecells. The bounded distortion assumption is relaxed to cover cases that include the linear quadratic Gaussian problem. For the infinite horizon problem and a stationary Markov source the optimality of deterministic Markov coding policies is shown. The existence of optimal stationary Markov quantization policies is also shown provided randomization that is shared by the encoder and the decoder is allowed.

For a piecewise expanding unimodal interval map$f$with unique absolutely continuous invariant probability measure$\mu $, a perturbation$X$, and an observable$\varphi $, the susceptibility function is$\Psi _\varphi (z)= \sum _{k=0}^\infty z^k \int X(x) \varphi '( f^k)(x) (f^k)'(x) \, d\mu $. Combining previous results [V. Baladi, On the susceptibility function of piecewise expanding interval maps.Comm. Math. Phys.275(2007), 839–859; V. Baladi and D. Smania, Linear response for piecewise expanding unimodal maps.Nonlinearity21(2008), 677–711] (deduced from spectral properties of Ruelle transfer operators) with recent work of Breuer–Simon [Natural boundaries and spectral theory.Adv. Math.226(2011), 4902–4920] (based on techniques from the spectral theory of Jacobi matrices and a classical paper of Agmon [Sur les séries de Dirichlet.Ann. Sci. Éc. Norm. Supér.(3)66(1949), 263–310]), we show that density of the postcritical orbit (a generic condition) implies that$\Psi _\varphi (z)$has a strong natural boundary on the unit circle. The Breuer–Simon method provides uncountably many candidates for the outer functions of$\Psi _\varphi (z)$, associated with precritical orbits. If the perturbation$X$is horizontal, a generic condition (Birkhoff typicality of the postcritical orbit) implies that the non-tangential limit of$\Psi _\varphi (z)$as$z\to 1$exists and coincides with the derivative of the absolutely continuous invariant probability measure with respect to the map (‘linear response formula’). Applying the Wiener–Wintner theorem, we study the singularity type of non-tangential limits of$\Psi _\varphi (z)$as$z\to e^{i\omega }$for real$\omega $. An additional ‘law of the iterated logarithm’ typicality assumption on the postcritical orbit gives stronger results.

높은 전송률에서 엔트로피 제한 양자화를 수행 시 최적의 양자기는 격자(lattice) 형태의 부호책을 가지는데, 규칙적인 구조로 인하여 양자화 과정이 단순하며, 격자의 형태에 따라 여러 양자화 알고리듬이 제안되어있다. 이러한 격자 벡터 양자기(vector quantizer: VQ)는 표본 적응 프로덕트 양자기(sample-adaptive product quantizer: SAPQ)를 사용하여 구현이 가능하며, 그 출력도 단순하게 엔트로피 부호화가 가능하다. 본 논문에서는 SAPQ에 기초한 엔트로피 부호화 방법을 제안하고, 무기억성(memoryless) 가우시언 분포에 대하여 여러 제안한 격자 VQ를 구현하고 양자화 에러 곡선을 엔트로피에 대하여 구하여 그 성능을 비교하였다. 실험을 통하여 전송률이 증가하면서 균등 분포에 이론적으로 얻는 이득과 비슷한 이득을 무기억성 가우시언 분포에서도 SAPQ의 출력을 엔트로피 부호화함으로 얻을 수 있음을 확인하였다.

A practical suboptimal universal block source coding scheme, subject to a fidelity criterion, is proposed. The algorithm is an extension of the Lempel-Ziv algorithm and is based on string matching with distortion. It is shown that given average distortion D>0, the algorithm achieves a rate of exceeding R(D/2) for a large class of sources and distortion measures. Tighter bounds on the rate are derived for discrete memoryless sources and for memoryless Gaussian sources.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

We show two discrete memoryless source (DMS) rate-distortion (R-D) problems with continuous alphabets where the optimal reconstruction distributions are discrete. The first problem is a Gaussian distributed source with square error distortion but a peak-limited reconstruction alphabet. The second is a source distribution related to the generalized inverse Gaussian (GIG) distribution with a certain distortion measure.

We explore the effects of continuous number density measurement on atoms in an optical lattice. By integrating a master equation for quantum observables, we calculate how single particle correlations decay. We consider weakly- and strongly- interacting bosons and noninteracting fermions. Even in the Mott regime, such measurements destroy correlations and increase the average energy, as long as some hopping is allowed. We explore the role of spatial resolution, and find that the heating rate is proportional to the amount of information gained from such measurements.

Breast cancer is the most common cancer diagnosed among US women and the second leading cause of cancer death 1 . Breast density is an independent risk factor for breast cancer and more than 25 states mandate its reporting to patients as part of the lay mammogram report 2 . Recent publications have demonstrated that breast density measured from low-dose chest CT (LDCT) correlates well with that measured from mammograms and MRIs 3-4 , thereby providing valuable information for many women who have undergone LDCT but not recent mammograms. A fully automated framework for breast density assessment from LDCT is presented in this paper. The whole breast region is first segmented using an anatomy-orientated novel approach based on the propagation of muscle fronts for separating the fibroglandular tissue from the underlying muscles. The fibroglandular tissue regions are then identified from the segmented whole breast and the percentage density is calculated based on the volume ratio of the fibroglandular tissue to the local whole breast region. The breast region segmentation framework was validated with 1270 LDCT scans, with 96.1% satisfactory outcomes based on visual inspection. The density assessment was evaluated by comparing with BI-RADS density grades established by an experienced radiologist in 100 randomly selected LDCT scans of female subjects. The continuous breast density measurement was shown to be consistent with the reference subjective grading, with the Spearman’s rank correlation 0.91 (p-value < 0.001). After converting the continuous density to categorical grades, the automated density assessment was congruous with the radiologist’s reading in 91% cases.

The problem of asymptotic (i,e., low-distortion) behavior of the rate-distortion function of a random vector is investigated for a class of non-difference distortion measures. The main result is an asymptotically tight expression which parallels the Shannon lower bound for difference distortion measures. For example, for an input-weighted squared error distortion measure d(x,y)=/spl par/W(x)(y-x)/spl par//sup 2/,y,x/spl isin/R/sup n/, the asymptotic expression for the rate-distortion function of X/spl isin/R/sup n/ at distortion level D equals h(X)-/sub 2///sup n/log(2/spl pi/eD/n)+Elog|detW(X)| where h(X) is the differential entropy of X. Extensions to stationary sources and to high-resolution remote (noisy) source coding are also given.