Adaptive lifting scheme with sparse criteria for image coding
 Mounir Kaaniche^{1}Email author,
 Béatrice PesquetPopescu^{1},
 Amel BenazzaBenyahia^{2} and
 JeanChristophe Pesquet^{3}
https://doi.org/10.1186/16876180201210
© Kaaniche et al; licensee Springer. 2012
Received: 30 June 2011
Accepted: 13 January 2012
Published: 13 January 2012
Abstract
Lifting schemes (LS) were found to be efficient tools for image coding purposes. Since LSbased decompositions depend on the choice of the prediction/update operators, many research efforts have been devoted to the design of adaptive structures. The most commonly used approaches optimize the prediction filters by minimizing the variance of the detail coefficients. In this article, we investigate techniques for optimizing sparsity criteria by focusing on the use of an ℓ_{1} criterion instead of an ℓ_{2} one. Since the output of a prediction filter may be used as an input for the other prediction filters, we then propose to optimize such a filter by minimizing a weighted ℓ_{1} criterion related to the global ratedistortion performance. More specifically, it will be shown that the optimization of the diagonal prediction filter depends on the optimization of the other prediction filters and viceversa. Related to this fact, we propose to jointly optimize the prediction filters by using an algorithm that alternates between the optimization of the filters and the computation of the weights. Experimental results show the benefits which can be drawn from the proposed optimization of the lifting operators.
1 Introduction
The discrete wavelet transform has been recognized to be an efficient tool in many image processing fields, including denoising [1] and compression [2]. Such a success of wavelets is due to their intrinsic features: multiresolution representation, good energy compaction, and decorrelation properties [3, 4]. In this respect, the second generation of wavelets provides very efficient transforms, based on the concept of lifting scheme (LS) developed by Sweldens [5]. It was shown that interesting properties are offered by such structures. In particular, LS guarantee a lossytolossless reconstruction required in some specific applications such as remote sensing imaging for which any distortion in the decoded image may lead to an erroneous interpretation of the image [6]. Besides, they are suitable tools for scalable reconstruction, which is a key issue for telebrowsing applications [7, 8].
Generally, LS are developed for the 1D case and then they are extended in a separable way to the 2D case by cascading vertical and horizontal 1D filtering operators. It is worth noting that a separable LS may not appear always very efficient to cope with the twodimensional characteristics of edges which are neither horizontal nor vertical [9]. To this respect, several research studies have been devoted to the design of non separable lifting schemes (NSLS) in order to better capture the actual twodimensional contents of the image. Indeed, instead of using samples from the same rows (resp. columns) while processing the image along the lines (resp. columns), 2D NSLS provide smarter choices in the selection of the samples by using horizontal, vertical and oblique directions at the prediction step [9]. For example, quincunx lifting schemes were found to be suitable for coding satellite images acquired on a quincunx sampling grid [10, 11]. In [12], a 2D wavelet decomposition comprising an adaptive update lifting step and three consecutive fixed prediction lifting steps was proposed. Another structure, which is composed of three prediction lifting steps followed by an update lifting step, has also been considered in the nonadaptive case [13, 14].
In parallel with these studies, other efforts have been devoted to the design of adaptive lifting schemes. Indeed, in a coding framework, the compactness of a LSbased multiresolution representation depends on the choice of its prediction and update operators. To the best of our knowledge, most existing studies have mainly focused on the optimization of the prediction stage. In general, the goal of these studies is to introduce spatial adaptivity by varying the direction of the prediction step [15–17], the length of the prediction filters [18, 19] and the coefficient values of the corresponding filters [9, 11, 15, 20, 21]. For instance, Gerek and Çetin [16] proposed a 2D edgeadaptive lifting scheme by considering three direction angles of prediction (0°, 45°, and 135°) and by selecting the orientation which leads to the smallest gradient. Recently, Ding et al. [17] have built an adaptive directional lifting structure with perfect reconstruction: the prediction is performed in local windows in the direction of high pixel correlation. A good directional resolution is achieved by employing fractional pixel precision level. A similar approach was also adopted in [22]. In [18], three separable prediction filters with different numbers of vanishing moments are employed, and then the best prediction is chosen according to the local features. In [19], a set of linear predictors of different lengths are defined based on a nonlinear function related to an edge detector. Another alternative strategy to achieve adaptivity aims at designing lifting filters by defining a given criterion. In this context, the prediction filters are often optimized by minimizing the detail signal variance through mean square criteria [15, 20]. In [9], the prediction filter coefficients are optimized with a least mean squares (LMS) type algorithm based on the prediction error. In addition to these adaptation techniques, the minimization of the detail signal entropy has also been investigated in [11, 21]. In [11], the approach is limited to a quincunx structure and the optimization is performed in an empirical manner using the NelderMead simplex algorithm due to the fact that the entropy is an implicit function of the prediction filter. However, such heuristic algorithms present the drawback that their convergence may be achieved at a local minimum of entropy. In [21], a generalized prediction step, viewed as a mapping function, is optimized by minimizing the detail signal energy given the pixel value probability conditioned to its neighbor pixel values. The authors show that the resulting mapping function also minimizes the output entropy. By assuming that the signal probability density function (pdf) is known, the benefit of this method has firstly been demonstrated for lossless image coding in [21]. Then, an extension of this study to sparse image representation and lossy coding contexts has been presented in [23]. Consequently, an estimation of the pdf must be available at the coder and the decoder side. Note that the main drawback of this method as well as those based on directional wavelet transforms [15, 17, 22, 24, 25] is that they require to transmit losslessly a side information to the decoder which may affect the whole compression performance especially at low bitrates. Furthermore, such adaptive methods lead to an increase of the computational load required for the selection of the best direction of prediction.
It is worth pointing out that, in practical implementations of compression systems, the sparsity of a signal, where a portion of the signal samples are set to zero, has a great impact on the ultimate ratedistortion performance. For example, embedded waveletbased image coders can spend the major part of their bit budget to encode the significance map needed to locate nonzero coefficients within the wavelet domain. To this end, sparsitypromoting techniques have already been investigated in the literature. Indeed, geometric wavelet transforms such as curvelets [26] and contourlets [27] have been proposed to provide sparse representations of the images. One difficulty of such transforms is their redundancy: they usually produce a number of coefficients that is larger than the number of pixels in the original image. This can be a main obstacle for achieving efficient coding schemes. To control this redundancy, a mixed contourlet and wavelet transform was proposed in [28] where a contourlet transform was used at fine scales and the wavelet transform was employed at coarse scales. Later, bandlet transforms that aim at developing sparse geometric representations of the images have been introduced and studied in the context of image coding and image denoising [29]. Unlike contourlets and curvelets which are fixed transforms, bandelet transforms require an edge detection stage, followed by an adaptive decomposition. Furthermore, the directional selectivity of the 2D complex dualtree discrete wavelet transforms [30] has been exploited in the context of image [31] and video coding [32]. Since such a transform is redundant, Fowler et al. applied a noiseshaping process [33] to increase the sparsity of the wavelet coefficients.
With the ultimate goal of promoting sparsity in a transform domain, we investigate in this article techniques for optimizing sparsity criteria, which can be used for the design of all the filters defined in a non separable lifting structure. We should note that sparsest wavelet coefficients could be obtained by minimizing an ℓ_{0} criterion. However, such a problem is inherently nonconvex and NPhard [34]. Thus, unlike previous studies where prediction has been separately optimized by minimizing an ℓ_{2} criterion (i.e., the detail signal variance), we focus on the minimization of an ℓ_{1} criterion. Since the output of a prediction filter may be used as an input for other prediction filters, we then propose to optimize such a filter by minimizing a weighted ℓ_{1} criterion related to the global prediction error. We also propose to jointly optimize the prediction filters by using an algorithm that alternates between filter optimization and weight computation. While the minimization of an ℓ_{1} criterion is often considered in the signal processing literature such as in the compressed sensing field [35], it is worth pointing out that, to the best of our knowledge, the use of such a criterion for lifting operator design has not been previously investigated.
The rest of this article is organized as follows. In Section 2, we recall our recent study for the design of all the operators involved in a 2D non separable lifting structure [36, 37]. In Section 3, the motivation for using an ℓ_{1} criterion in the design of optimal lifting structures is firstly discussed. Then, the iterative algorithm for minimizing this criterion is described. In Section 4, we present a weighted ℓ_{1} criterion which aims at minimizing the global prediction error. In Section 5, we propose to jointly optimize the prediction filters by using an algorithm that alternates between optimizing all the filters and redefining the weights. Finally, in Section 6, experimental results are given and then some conclusions are drawn in Section 7.
2 2D lifting structure and optimization methods
2.1 Principle of the considered 2D NSLS structure
Let x denote the digital image to be coded. At each resolution level j and each pixel location (m, n), its approximation coefficient is denoted by x_{ j }(m, n) and the associated four polyphase components by x_{0,j}(m, n) = x_{ j }(2m,2n), x_{1,j}(m,n) = x_{ j }(2m,2n+1), x_{2,j}(m,n) = x_{ j }(2m+1,2n), and x_{3,j}(m,n) = x_{ j }(2m + 1, 2n + 1). Furthermore, we denote by ${\mathbf{P}}_{j}^{\left(HH\right)}$, ${\mathbf{P}}_{j}^{\left(LH\right)}$, ${\mathbf{P}}_{j}^{\left(HL\right)}$, and U_{ j }the three prediction and update filters employed to generate the detail coefficients ${x}_{j+1}^{\left(HH\right)}$ oriented diagonally, ${x}_{j+1}^{\left(LH\right)}$ oriented vertically, ${x}_{j+1}^{\left(HL\right)}$ oriented horizontally, and the approximation coefficients x_{j+1}. In accordance with Figure 1, let us introduce the following notation:

For the first prediction step, the prediction multiple input, single output (MISO) filter ${\mathbf{P}}_{j}^{\left(HH\right)}$ can be seen as a sum of three single input, single output (SISO) filters ${\mathbf{P}}_{0,j}^{\left(HH\right)}$, ${\mathbf{P}}_{1,j}^{\left(HH\right)}$, and ${\mathbf{P}}_{2,j}^{\left(HH\right)}$ whose respective inputs are the components x_{0,j}, x_{1,j}and x_{2,j}.

For the second (resp. third) prediction step, the prediction MISO filter ${\mathbf{P}}_{j}^{\left(LH\right)}$ (resp. ${\mathbf{P}}_{j}^{\left(HL\right)}$) can be seen as a sum of two SISO filters ${\mathbf{P}}_{0,j}^{\left(LH\right)}$ and ${\mathbf{P}}_{1,j}^{\left(LH\right)}$ (resp. ${\mathbf{P}}_{0,j}^{\left(HL\right)}$ and ${\mathbf{P}}_{1,j}^{\left(HL\right)}$) whose respective inputs are the components x_{2,j}and ${x}_{j+1}^{\left(HH\right)}$ (resp. x_{1,j}and ${x}_{j+1}^{\left(HH\right)}$).

For the update step, the update MISO filter U_{ j }can be seen as a sum of three SISO filters ${\mathbf{U}}_{j}^{\left(HL\right)}$, ${\mathbf{U}}_{j}^{\left(LH\right)}$, and ${\mathbf{U}}_{j}^{\left(HH\right)}$ whose respective inputs are the detail coefficients ${x}_{j+1}^{\left(HL\right)}$, ${x}_{j+1}^{\left(LH\right)}$, and ${x}_{j+1}^{\left(HH\right)}$.
The set ${\mathcal{P}}_{i,j}^{\left(o\right)}$ (resp. ${\mathcal{U}}_{j}^{\left(o\right)}$) and the coefficients ${p}_{i,j}^{\left(o\right)}\left(k,l\right)$ (resp. ${u}_{j}^{\left(o\right)}\left(k,l\right)$) denote the support and the weights of the three prediction filters (resp. of the update filter). Note that in Equations (1)(4), we have introduced the rounding operations ⌊.⌋ in order to allow lossytolossless encoding of the coefficients [7]. Once the considered NSLS structure has been defined, we will focus now on the optimization of its lifting operators.
2.2 Optimization methods
where

${\mathbf{p}}_{j}^{\left(HH\right)}={\left({\mathbf{p}}_{0,j}^{\left(HH\right)},{\mathbf{p}}_{1,j}^{\left(HH\right)},{\mathbf{p}}_{2,j}^{\left(HH\right)}\right)}^{\mathsf{\text{T}}}$ is the prediction vector, and, for every i ∈ {0, 1, 2},${p}_{i,j}^{\left(HH\right)}={\left({p}_{i,j}^{\left(HH\right)}\left(k,l\right)\right)}_{\left(k,l\right)\in {\mathcal{P}}_{i,j}^{\left(HH\right)}},$

${\stackrel{\u0303}{\mathbf{x}}}_{j}^{\left(HH\right)}\left(m,n\right)={\left({\mathbf{x}}_{0,j}^{\left(HH\right)}\left(m,n\right),{\mathbf{x}}_{1,j}^{\left(HH\right)}\left(m,n\right),{\mathbf{x}}_{2,j}^{\left(HH\right)}\left(m,n\right)\right)}^{\mathsf{\text{T}}}$ is the reference vector with${\mathbf{x}}_{i,j}^{\left(HH\right)}\left(m,n\right)={\left({x}_{i,j}\left(mk,nl\right)\right)}_{\left(k,l\right)\in {\mathcal{P}}_{i,j}^{\left(HH\right)}}.$
The other optimal prediction filters ${\mathbf{p}}_{j}^{\left(HL\right)}$ and ${\mathbf{p}}_{j}^{\left(LH\right)}$ are obtained in a similar way.
Where

${\mathbf{u}}_{j}={\left({u}_{j}^{\left(o\right)}\left(k,l\right)\right)}_{\left(k,l\right)\in {\mathcal{U}}_{j}^{\left(o\right)},o\in \left\{HL,LH,HH\right\}}^{\mathsf{\text{T}}}$is the update weight vector,

${\mathbf{x}}_{j+1}\left(m,n\right)={\left({x}_{j+1}^{\left(o\right)}\left(mk,nl\right)\right)}_{\left(k,l\right)\in {\mathcal{P}}_{i,j}^{\left(o\right)},o\in \left\{HL,LH,HH\right\}}^{\mathsf{\text{T}}}$ is the reference vector containing the detail signals previously computed at the j th resolution level.
Now, we will introduce a novel twist in the optimization of the different filters: the use of an ℓ_{1}based criterion in place of the usual ℓ_{2}based measure.
3 From ℓ_{2}to ℓ_{1}minimization
3.1 Motivation
where $\Gamma \left(z\right)={\int}_{0}^{+\infty}{t}^{z1}{e}^{t}dt$ is the Gamma function, α > 0 is the scale parameter, and β > 0 is the shape parameter. We should note that in the particular case when β = 2 (resp. β = 1), the GGD corresponds to the Gaussian distribution (resp. the Laplace one). The parameters α and β can be easily estimated by using the maximum likelihood technique [42].
where (M_{ j },N_{ j }) corresponds to the dimensions of the subband ${x}_{j+1}^{\left(o\right)}$.
This shows that there exists a close link between the minimization of the entropy of the detail wavelet coefficients and the minimization of their ${\ell}_{{{\beta}_{j+1}^{\left(o\right)}}_{}}$norm. This suggests in particular that most of the existing studies minimizing the ℓ_{2}norm of the detail signals aim at minimizing their entropy by assuming a Gaussian model.
It is important to note that the shape parameters of the resulting detail subbands are closer to β = 1 than to β = 2. Further experiments performed on a large dataset of images^{b} have shown that the average of β values are closer to 1 (typical values range from 0.5 to 1.5). These observations suggest that minimizing the ℓ_{1}norm may be more appropriate than ℓ_{2} minimization. In addition, the former approach has the advantage of producing sparse representations.
3.2 ℓ_{1}minimization technique
where x_{ i,j }(m,n) is the (i + 1)^{ th }polyphase component to be predicted, ${\stackrel{\u0303}{\mathbf{x}}}_{j}^{\left(o\right)}\left(m,n\right)$ is the reference vector containing the samples used in the prediction step, ${\mathbf{p}}_{j}^{\left(o\right)}$ is the prediction operator vector to be optimized (L will subsequently designate its length). Although the criterion in (11) is convex, a major difficulty that arises in solving this problem stems from the fact that the function to be minimized is not differentiable. Recently, several optimization algorithms have been proposed to solve nonsmooth minimization problems like (11). These problems have been traditionally addressed with linear programming [45]. Alternatively, a flexible class of proximal optimization algorithms has been developed and successfully employed in a number of applications. A survey on these proximal methods can be found in [46]. These methods are also closely related to augmented Lagrangian methods [47]. In our context, we have employed the DouglasRachford algorithm which is an efficient optimization tool for this problem [48].
3.2.1 The DouglasRachford algorithm
For minimizing the ℓ_{1} criterion, we will resort to the concept of proximity operators [49], which has been recognized as a fundamental tool in the recent convex optimization literature [50, 51]. The necessary background on convex analysis and proximity operators [52, 53] is given in Appendix A.
In this case, the DouglasRachford algorithm can be applied to provide an appealing numerical solution to Problem (13) (see Appendix B).
Once the different terms involved in the iterative algorithm (33) are defined, this one can be applied and further extended to optimize all the prediction filters.
4 Global prediction error minimization technique
4.1 Motivation
Up to now, each prediction filter ${\mathbf{p}}_{j}^{\left(o\right)}\left(o\in \left\{HL,LH,HH\right\}\right)$ has been separately optimized by minimizing the ℓ_{1}norm of the corresponding detail signal ${x}_{j+1}^{\left(o\right)}$ which seems appropriate to determine ${\mathbf{p}}_{j}^{\left(LH\right)}$ and ${\mathbf{p}}_{j}^{\left(HL\right)}$. However, it can be noticed from Figure 1 that the diagonal detail signal ${x}_{j+1}^{\left(HH\right)}$ is also used through the second and the third prediction steps to compute the vertical and the horizontal detail signals respectively. Therefore, the solution ${\mathbf{p}}_{j}^{\left(HH\right)}$ resulting from the previous optimization method may be suboptimal. As a result, we propose to optimize the prediction filter ${\mathbf{p}}_{j}^{\left(HH\right)}$ by minimizing the global prediction error, as described in detail in the next section.
4.2 Optimization of the prediction filter ${\mathbf{p}}_{j}^{\left(HH\right)}$
where ${\kappa}_{j+1}^{\left(o\right)}$, o ∈ {HL, LH, HH}, are strictly positive weighting terms.
Before focusing on the method employed to minimize the proposed criterion, we should first express ${\mathcal{J}}_{w{\ell}_{1}}$ as a function of the filter ${\mathbf{p}}_{j}^{\left(HH\right)}$ to be optimized.
where ${h}_{i,j}^{\left(o,1\right)}$ is a filter which depends on the prediction coefficients of ${\mathbf{p}}_{j}^{\left(LH\right)}$ and ${\mathbf{p}}_{j}^{\left(HL\right)}$.
It is worth noting that in practice, the determination of ${y}_{j}^{\left(o,1\right)}\left(m,n\right)$ and ${\mathbf{x}}_{j}^{\left(o,1\right)}\left(m,n\right)$ does not require to find the explicit expressions of ${h}_{i,j}^{\left(o,1\right)}$ and these signals can be determined numerically as follows:

The first term (resp. the second one) in the expression of ${y}_{j}^{\left(o,1\right)}\left(m,n\right)$ in Equation (20) can be found by computing ${x}_{j+1}^{\left(o\right)}\left(m,n\right)$ from the components ${\left({x}_{i,j}^{\left(1\right)}\left(m,n\right)\right)}_{i\in \left\{0,1,2,3\right\}}$ while setting ${x}_{3,j}^{\left(1\right)}\left(m,n\right)=0$ (resp. while setting ${x}_{i,j}^{\left(1\right)}\left(m,n\right)=0$ for i ∈ {0,1,2} and ${x}_{3,j}^{\left(1\right)}\left(m,n\right)={x}_{3,j}\left(m,n\right)$).

The vector ${\mathbf{x}}_{j}^{\left(o,1\right)}\left(m,n\right)$ in Equation (21) can be found as follows. For each i ∈ {0,1,2}, the computation of its component ${\sum}_{k,l}{h}_{3,j}^{\left(o,1\right)}\left(k,l\right){x}_{i,j}\left(mk,nl\right)$ requires to compute ${x}_{j+1}^{\left(o\right)}\left(m,n\right)$ by setting ${x}_{3,j}^{\left(1\right)}\left(m,n\right)={x}_{i,j}\left(m,n\right)$ and ${x}_{{i}^{\prime},j}^{\left(1\right)}\left(m,n\right)=0$ for i' ∈ {0,1,2}. The result of this operation has to be considered for different shift values (r, s) (as can be seen in Equation (21)).
Once the different terms involved in the proposed weighted criterion in Equation (22) are defined (the constant values ${\kappa}_{j+1}^{\left(o\right)}$ are supposed to be known), we will focus now on its minimization. Indeed, unlike the previous criterion (Equation 11), which consists only of an ℓ_{1} term, the proposed criterion is a sum of three ℓ_{1} terms. To minimize such a criterion (22), one can still use the DouglasRachford algorithm through a formulation in a product space [46, 54].
4.2.1 DouglasRachford algorithm in a product space
where ${\kappa}_{j+1}^{\left(o\right)}$, o ∈ {HL,LH,HH}, are positive weights.
We are thus back to a problem involving two functions in a larger space, which is the product space ${\mathbb{H}}_{j}$. So, the DouglasRachford algorithm can be applied to solve our minimization problem (see Appendix C). Finally, once the prediction filter ${\mathbf{p}}_{j}^{\left(HH\right)}$ is optimized and fixed, it can be noticed that the other prediction filters ${\mathbf{p}}_{j}^{\left(HL\right)}$ and ${\mathbf{p}}_{j}^{\left(LH\right)}$ can be separately optimized by minimizing ${\mathcal{J}}_{{\ell}_{1}}\left({\mathbf{p}}_{j}^{\left(HL\right)}\right)$ and ${\mathcal{J}}_{{\ell}_{1}}\left({\mathbf{p}}_{j}^{\left(LH\right)}\right)$ as explained in Section 3. This is justified by the fact that the inputs of the filter ${\mathbf{p}}_{j}^{\left(HL\right)}$ (resp. ${\mathbf{p}}_{j}^{\left(LH\right)}$) are independent of the output of the filter ${\mathbf{p}}_{j}^{\left(LH\right)}$ (resp. ${\mathbf{p}}_{j}^{\left(HL\right)}$).
5 Joint optimization method
5.1 Motivation
From Equations (20) and (21), it can be observed that ${y}_{j}^{\left(o,1\right)}$ and ${\mathbf{x}}_{j}^{\left(o,1\right)}$, which are used to optimize ${\mathbf{p}}_{j}^{\left(HH\right)}$, depend on the coefficients of the prediction filters ${\mathbf{p}}_{j}^{\left(HL\right)}$ and ${\mathbf{p}}_{j}^{\left(LH\right)}$. On the other hand, since ${\mathbf{p}}_{j}^{\left(HL\right)}$ and ${\mathbf{p}}_{j}^{\left(LH\right)}$ use ${x}_{j+1}^{\left(HH\right)}$ as reference signal in the second and the third prediction steps, their optimal values will depend on the optimal prediction filter ${\mathbf{p}}_{j}^{\left(HH\right)}$. Thus, we conclude that the optimization of the filters (${\mathbf{p}}_{j}^{\left(HL\right)}$, ${\mathbf{p}}_{j}^{\left(LH\right)}$) depends on the optimization of the filter ${\mathbf{p}}_{j}^{\left(HH\right)}$ and viceversa.
A joint optimization method can therefore be proposed which iteratively optimizes the prediction filters ${\mathbf{p}}_{j}^{\left(HH\right)}$, ${\mathbf{p}}_{j}^{\left(HL\right)}$, and ${\mathbf{p}}_{j}^{\left(LH\right)}$.
5.2 Proposed algorithms
While the optimization of the prediction filters ${\mathbf{p}}_{j}^{\left(HL\right)}$ and ${\mathbf{p}}_{j}^{\left(LH\right)}$ is simple, the optimization of the prediction filter ${\mathbf{p}}_{j}^{\left(HH\right)}$ is less obvious. Indeed, if we examine the criterion ${\mathcal{J}}_{w{\ell}_{1}}$, the immediate question that arises is: which values of the weighting parameters will produce the sparsest decomposition?
A simple solution consists of setting all the weights ${\kappa}_{j+1}^{\left(o\right)}$ to one. Then, we are considering the particular case of the unweighted ℓ_{1} criterion, which simply represents the sum of the ℓ_{1}norm of the three details subbands ${x}_{j+1}^{\left(o\right)}$. In this case, the joint optimization problem is solved by applying the following simple iterative algorithm at each resolution level j.
5.2.1 First proposed algorithm
➀ Initialize the iteration number it to 0.

Optimize separately the three prediction filters as explained in Section 3. The resulting filters will be denoted respectively by ${\mathbf{p}}_{j}^{\left(HH,0\right)}$, ${\mathbf{p}}_{j}^{\left(LH,0\right)}$, and ${\mathbf{p}}_{j}^{\left(HL,0\right)}$.

Compute the resulting global unweighted prediction error (i.e., the sum of the ℓ_{1}norm of the three resulting details subbands).
➁ for it = 1,2,3,

Set ${\mathbf{p}}_{j}^{\left(LH\right)}={\mathbf{p}}_{j}^{\left(LH,it1\right)},{\mathbf{p}}_{j}^{\left(HL\right)}={\mathbf{p}}_{j}^{\left(HL,it1\right)}$, and optimize ${\mathbf{P}}_{j}^{\left(HH\right)}$ by minimizing ${\mathcal{J}}_{w{\ell}_{1}}\left({\mathbf{p}}_{j}^{\left(HH\right)}\right)$ (while setting ${\kappa}_{j+1}^{\left(o\right)}=1$). Let ${\mathbf{p}}_{j}^{\left(HH,it\right)}$ be the new optimal filter at iteration it.

Set ${\mathbf{p}}_{j}^{\left(HH\right)}={\mathbf{p}}_{j}^{\left(HH,it\right)}$, and optimize ${\mathbf{P}}_{j}^{\left(LH\right)}$ by minimizing ${\mathcal{J}}_{{\ell}_{1}}\left({\mathbf{p}}_{0}^{\left(LH\right)}\right)$. Let ${\mathbf{p}}_{j}^{\left(LH,it\right)}$ be the new optimal filter.

Set ${\mathbf{p}}_{j}^{\left(HH\right)}={\mathbf{p}}_{j}^{\left(HH,it\right)}$, and optimize ${\mathbf{P}}_{j}^{\left(HL\right)}$ by minimizing ${\mathcal{J}}_{{\ell}_{1}}\left({\mathbf{p}}_{j}^{\left(HL\right)}\right)$. Let ${\mathbf{p}}_{j}^{\left(HL,it\right)}$ be the new optimal filter.
Once the prediction filters are optimized, the update filter is finally optimized as explained in Section 2. However, in practice, once all the filters are optimized and the decomposition is performed, the different generated wavelet subbands ${x}_{j+1}^{\left(o\right)}$ are weighted before the entropy encoding (using JPEG2000 encoder) in order to obtain a distortion in the spatial domain which is very close to the distortion in the wavelet domain.
This weighting step is very important since standard bit allocation algorithms assume that the quadratic distortion in the wavelet domain is equal to that in the spatial domain, which is not true in the case of biorthogonal wavelets [55]. Therefore, the filters resulting from the first choice of ${\kappa}_{j+1}^{\left(o\right)}$ are suboptimal in the sense that they do not take into account the weighting procedure. For this reason, it has been noticed on some experiments (as it can be seen in Section 6) that the basic optimization technique does not achieve the best coding performances.
Since the resulting entropy of each subband uses weights which also depend on the prediction filters (as mentioned above), we propose an iterative algorithm that alternates between optimizing all the filters and redefining the weights. This algorithm, which is performed for each resolution level j, is as follows.
5.2.2 Second proposed algorithm
➀ Initialize the iteration number it to 0.

Optimize separately the three prediction filters as explained in Section 3. The resulting filters will be denoted respectively by ${\mathbf{p}}_{j}^{\left(HH,0\right)}$, ${\mathbf{p}}_{j}^{\left(LH,0\right)}$, and ${\mathbf{p}}_{j}^{\left(HL,0\right)}$.

Optimize the update filter (as explained in Section 2).

Compute the weights ${w}_{j+1}^{\left(o,0\right)}$ of each detail subband as well as the constant values ${\kappa}_{j+1}^{\left(o,0\right)}$.
➁ for it = 1,2,3,...

Set ${\mathbf{p}}_{j}^{\left(LH\right)}={\mathbf{p}}_{j}^{\left(LH,it1\right)},{\mathbf{p}}_{j}^{\left(HL\right)}={\mathbf{p}}_{j}^{\left(HL,it1\right)}$, and optimize ${\mathbf{P}}_{j}^{\left(HH\right)}$ by minimizing ${\mathcal{J}}_{w{\ell}_{1}}\left({\mathbf{p}}_{j}^{\left(HH\right)}\right)$. Let ${\mathbf{p}}_{j}^{\left(HH,it\right)}$ be the new optimal filter.

Set ${\mathbf{p}}_{j}^{\left(HH\right)}={\mathbf{p}}_{j}^{\left(HH,it\right)}$, and optimize ${\mathbf{P}}_{j}^{\left(LH\right)}$ by minimizing ${\mathcal{J}}_{{\ell}_{1}}\left({\mathbf{p}}_{j}^{\left(LH\right)}\right)$. Let ${\mathbf{p}}_{j}^{\left(LH,it\right)}$ be the new optimal filter.

Set ${\mathbf{p}}_{j}^{\left(HH\right)}={\mathbf{p}}_{j}^{\left(HH,it\right)}$, and optimize ${\mathbf{P}}_{j}^{\left(HL\right)}$ by minimizing ${\mathcal{J}}_{{\ell}_{1}}\left({\mathbf{p}}_{j}^{\left(HL\right)}\right)$. Let ${\mathbf{p}}_{j}^{\left(HL,it\right)}$ be the new optimal filter.

Optimize the update filter (as explained in Section 2).

Compute the new weights ${w}_{j+1}^{\left(o,it\right)}$ as well as ${\kappa}_{j+1}^{\left(o,it\right)}$.
6 Experimental results
Simulations were carried out on two kinds of still images originally quantized over 8 bpp which are either single views or stereoscopic ones. A large dataset composed of 50 still images^{b} and 50 stereo images^{c} has been considered. The gain related to the optimization of the NSLS operators, using different minimization criteria, was evaluated in these contexts. In order to show the benefits of the proposed ℓ_{1} optimization criterion, we provide the results for the following decompositions carried out over three resolution levels:

The first one is the LS corresponding to the 5/3 transform, also known as the (2,2) wavelet transform [7]. In the following, this method will be designated by NSLS(2,2).

The second method consists of optimizing the prediction and update filters as proposed in [20, 38]. More precisely, the prediction filters are optimized by minimizing the ℓ_{2}norm of the detail coefficients whereas the update filter is optimized by minimizing the reconstruction error. This optimization method will be designated by NSLS(2,2)OPTGM.

The third approach corresponds to our previous method presented recently in [37]. While the prediction filters are optimized in the same way as the second method, the update filter is optimized by minimizing the difference between the approximation signal and the decimated version of the output of an ideal lowpass filter. We emphasize here that the prediction filters are optimized separately. This method will be denoted by NSLS(2,2)OPTL2.

The fourth method modifies the optimization stage of the prediction filters by using the ℓ_{1}norm instead of the ℓ_{2}norm. The optimization of the update filter is similar to the technique used in the third method. In what follows, this method will be designated by NSLS(2,2)OPTL1.

The fifth method consists of jointly optimizing the prediction filters by using the proposed weighted ℓ_{2} minimization technique where the weights ${\kappa}_{j+1}^{\left(o\right)}$ are set to $\frac{1}{{\alpha}_{j+1}^{\left(o\right)}}$. The optimization of the update filter is similar to the technique used in the third and fourth methods. This optimization method will be designated by NSLS(2,2)OPTWL1. We have also tested this optimization method when the weights ${\kappa}_{j+1}^{\left(o\right)}$ are set to 1. In this case, the method will be denoted by NSLS(2,2)OPTWL1 (${\kappa}_{j+1}^{\left(o\right)}=1$).
These plots show that NSLS(2,2)OPTL2 outperforms NSLS(2,2) by 0.10.5 dB. It can also be noticed that NSLS(2,2)OPTL2 and NSLS(2,2)OPTGM perform similarly in terms of quality of reconstruction. An improvement of 0.10.3 dB is obtained by using the ℓ_{1} minimization technique instead of the ℓ_{2} one. Finally, the joint optimization technique (NSLS(2,2)OPTWL1) outperforms the separate optimization technique (NSLS(2,2)OPTL1) and improves the PSNR by 0.10.2 dB. The gain becomes more important (up to 0.55 dB) when compared with NSLS(2,2)OPTL2. It is important to note here that setting the weights ${\kappa}_{j+1}^{\left(o\right)}$ to 1 (NSLS(2,2)OPTWL1 (${\kappa}_{j+1}^{\left(o\right)}=1$)) can yield to a degradation of about 0.10.25 dB compared with NSLS(2,2)OPTWL1 on some images.
Performance of the proposed method vs the 9/7 transform
0.05 bpp  0.1 bpp  0.15 bpp  0.2 bpp  

NSLS(2,2)OPTWL1  9/7  NSLS (2,2)OPTWL1  9/7  NSLS (2,2)OPTWL1  9/7  NSLS (2,2)OPTWL1  9/7  
PSNR  27.85  27.75  30.25  30.31  31.23  31.35  31.76  31.92  
elaine  SSIM  0.669  0.659  0.716  0.715  0.739  0.739  0.754  0.756 
VSNR  18.44  18.09  23.10  23.05  25.60  25.50  27.28  27.42  
PSNR  25.10  25.09  27.08  27.18  28.36  28.51  29.51  29.58  
castle  SSIM  0.725  0.712  0.790  0.780  0.825  0.821  0.855  0.851 
VSNR  17.54  17.22  21.55  21.10  23.74  23.40  25.80  25.32  
PSNR  27.51  27.58  29.12  29.24  29.92  30.12  30.50  30.70  
einst  SSIM  0.603  0.601  0.654  0.655  0.687  0.689  0.710  0.715 
VSNR  15.33  15.25  18.62  18.71  20.37  20.47  21.59  21.94  
PSNR  26.70  26.68  29.59  29.56  31.25  31.47  32.70  32.90  
lena  SSIM  0.747  0.734  0.818  0.808  0.851  0.850  0.871  0.873 
VSNR  15.94  15.73  20.56  20.18  24.06  23.95  26.12  26.15  
PSNR  26.51  26.43  29.81  30.33  31.84  32.63  33.61  34.44  
cameraman  SSIM  0.783  0.774  0.847  0.842  0.887  0.892  0.914  0.915 
VSNR  16.74  16.34  21.73  21.66  24.94  25.70  27.75  28.34  
PSNR  24.65  24.55  26.82  26.86  28.43  28.54  29.52  29.74  
boat  SSIM  0.675  0.661  0.753  0.746  0.806  0.802  0.837  0.836 
VSNR  13.41  13.03  17.14  16.89  20.24  19.76  22.19  21.89  
PSNR  25.75  25.50  29.24  29.17  30.88  31.16  31.12  32.38  
peppers  SSIM  0.720  0.705  0.789  0.778  0.818  0.815  0.834  0.832 
VSNR  16.00  15.51  21.87  21.19  25.18  25.00  27.22  27.09  
PSNR  24.19  23.84  30.66  29.88  33.99  33.10  36.13  35.82  
plane  SSIM  0.809  0.754  0.890  0.871  0.917  0.903  0.931  0.921 
VSNR  9.48  7.72  17.73  15.51  21.28  20.30  24.68  24.12  
PSNR  24.88  24.72  27.67  27.73  29.24  29.46  30.45  30.65  
average  SSIM  0.647  0.633  0.727  0.720  0.773  0.771  0.803  0.802 
VSNR  14.50  13.98  18.90  18.62  21.77  21.71  23.90  23.85 
Computation time (s) of the sparse optimization methods for the design of each prediction filter
Plane  Girl  Boat  Cameraman  

256 × 256  256 × 256  512 × 512  512 × 512  
it  Time (s)  it  time(s)  it  time(s)  it  time(s)  
ℓ_{1} criterion: ${\mathbf{p}}_{0}^{\left(HL\right)}$  22  0.09  27  0.09  30  0.38  60  0.81 
ℓ_{1} criterion: ${\mathbf{p}}_{0}^{\left(LH\right)}$  55  0.15  28  0.09  31  0.39  100  1.13 
weighted ℓ_{1} criterion: ${\mathbf{p}}_{0}^{\left(HH\right)}$  30  0.42  35  0.49  49  3.08  30  2.01 
7 Conclusion
In this article, we have studied different optimization techniques for the design of filters in a NSLS structure. A new criterion has been presented for the optimization of the prediction filters in this context. The idea consists of jointly optimizing these filters by minimizing iteratively a weighted ℓ_{1} criterion. Experimental results carried out on still images and stereo images pair have illustrated the benefits which can be drawn from the proposed optimization technique. In future study, we plan to extend this optimization method to LS with more than two stages like the PUP and PUPU structures.
Appendix
A Some background on convex optimization
The main definitions which will be useful to understand our optimization algorithms are briefly summarized below:

ℝ^{ K }is the usual Kdimensional Euclidean space with norm ..

The distance function to a nonempty set C ⊂ ℝ^{ K }is defined by$\forall \mathbf{x}\in {\mathbb{R}}^{K},\phantom{\rule{1em}{0ex}}{d}_{C}\left(\mathbf{x}\right)=\underset{\mathbf{y}\in C}{\text{inf}}\left\right\mathbf{x}\mathbf{y}\left\right.$

The projection of x ∈ ℝ^{ K }onto a nonempty closed convex set C ⊂ ℝ^{ K }is the unique point P_{ C }(x) ∈ C such that d_{ C }(x) = x  P_{ C }(x).

The indicator function of C is given by$\forall \mathbf{x}\in {\mathbb{R}}^{K,}\phantom{\rule{1em}{0ex}}{\iota}_{C}\left(\mathbf{x}\right)=\left\{\begin{array}{cc}\hfill 0\hfill & \hfill \mathsf{\text{if}}\mathbf{x}\phantom{\rule{2.77695pt}{0ex}}\in C,\hfill \\ \hfill +\infty \hfill & \hfill \mathsf{\text{otherwise}}.\hfill \end{array}\right.$(32)

Γ_{0}(ℝ^{ K }) is the class of functions from ℝ^{ K }to ]  ∞, + ∞] which are lower semicontinuous, convex, and not identically equal to + ∞.

The proximity operator of f ∈ Γ_{0}(ℝ^{ K }) is $\mathsf{\text{pro}}{\mathsf{\text{x}}}_{f}:{\mathbb{R}}^{K}\to {\mathbb{R}}^{K}:\mathbf{x}\mapsto \text{arg}\underset{\mathbf{y}\in {\mathbb{R}}^{K}}{\text{min}}f\left(\mathbf{y}\right)+\frac{1}{2}\left\right\mathbf{x}\mathbf{y}{}^{2}$. It is important to note that the proximity operator generalizes the notion of a projection operator onto a closed convex set C in the sense that $\mathsf{\text{pro}}{\mathsf{\text{x}}}_{{\iota}_{C}}={P}_{C}$, and it moreover possesses most of its attractive properties [49] that make it particularly wellsuited for designing iterative minimization algorithms.
B The Douglas Rachford algorithm
where ${\mathbf{p}}_{j,k}^{\left(o\right)}={\left({\sum}_{m,n}{\stackrel{\u0303}{\mathbf{x}}}_{j}^{\left(o\right)}\left(m,n\right){\left({\stackrel{\u0303}{\mathbf{x}}}_{j}^{\left(o\right)}\left(m,n\right)\right)}^{\mathsf{\text{T}}}\right)}^{1}{\sum}_{m,n}{\stackrel{\u0303}{\mathbf{x}}}_{j}^{\left(o\right)}\left(m,n\right){t}_{j,j}^{\left(o\right)}\left(m,n\right).$.
Finally it is important to note that it has been shown (see [62] and references therein) that every sequence ${\left({\mathbf{z}}_{j,k}^{\left(o\right)}\right)}_{k\in \mathbb{N}}$ generated by the DouglasRachford algorithm (33) converges to a solution to problem (13) provided that the parameters γ and λ are fixed as indicated.
C The DouglasRachford algorithm in a product space
Endnotes
^{a}The ztransform of a signal x will be denoted in capital letters by X. ^{b}http://sipi.usc.edu/database. ^{c}http://vasc.ri.cmu.edu/idb/html/stereo/index.html, http://vasc.ri.cmu.edu/idb/html/jisct/index.html and http://cat.middlebury.edu/stereo/data.html.
Declarations
Acknowledgements
Part of this study has been presented in [63].
Authors’ Affiliations
References
 Donoho DL, Johnstone IM: Ideal spatial adaptation by wavelet shrinkage. Biometrika 1994, 81(3):425455. 10.1093/biomet/81.3.425MathSciNetView ArticleGoogle Scholar
 Antonini M, Barlaud M, Mathieu P, Daubechies I: Image coding using wavelet transform. IEEE Trans Image Process 1992, 1(2):205220. 10.1109/83.136597View ArticleGoogle Scholar
 Woods JW: Subband Image Coding. Kluwer Academic Publishers, Norwell, MA, USA; 1990.Google Scholar
 Mallat S: A Wavelet Tour of Signal Processing. Academic Press, San Diego; 1998.Google Scholar
 Sweldens W: The lifting scheme: a customdesign construction of biorthogonal wavelets. Volume 3. Appl Comput Harmonic Anal; 1996:186200.Google Scholar
 Arai K: Preliminary study on information lossy and lossless coding data compression for the archiving of ADEOS data. IEEE Trans Geosci Remote Sens 1990, 28: 732734. 10.1109/TGRS.1990.573001View ArticleGoogle Scholar
 Calderbank AR, Daubechies I, Sweldens W, Yeo BL: Wavelet transforms that map integers to integers. Appl Comput Harmonic Anal 1998, 5(3):332369. 10.1006/acha.1997.0238MathSciNetView ArticleGoogle Scholar
 Taubman D, Marcellin M: JPEG2000: Image Compression Fundamentals, Standards and Practice. Kluwer Academic Publishers, Norwell, MA, USA; 2001.Google Scholar
 Gerek ON, Çetin AE: Adaptive polyphase subband decomposition structures for image compression. IEEE Trans Image Process 2000, 9(10):16491660. 10.1109/83.869176View ArticleGoogle Scholar
 Gouze A, Antonini M, Barlaud M, Macq B: Optimized lifting scheme for twodimensional quincunx sampling images. In IEEE International Conference on Image Processing. Volume 2. Thessaloniki, Greece; 2001:253258.Google Scholar
 BenazzaBenyahia A, Pesquet JC, Hattay J, Masmoudi H: Blockbased adaptive vector lifting schemes for multichannel image coding. EURASIP Int J Image Video Process 2007, 10. (2007)Google Scholar
 Heijmans H, Piella G, PesquetPopescu B: Building adaptive 2D wavelet decompositions by update lifting. In IEEE International Conference on Image Processing. Volume 1. Rochester, New York, USA; 2002:397400.View ArticleGoogle Scholar
 Chokchaitam S: A nonseparable twodimensional LWT for an image compression and its theoretical analysis. Thammasat Internat J Sci Technol 2004, 9: 3543.Google Scholar
 Sun YK: A twodimensional lifting scheme of integer wavelet transform for lossless image compression. In International Conference on Image Processing. Volume 1. Singapore; 2004:497500.Google Scholar
 Chappelier V, Guillemot C: Oriented wavelet transform for image compression and denoising. IEEE Trans Image Process 2006, 15(10):28922903.View ArticleGoogle Scholar
 Gerek ON, Çetin AE: A 2D orientationadaptive prediction filter in lifting structures for image coding. IEEE Trans Image Process 2006, 15: 106111.View ArticleGoogle Scholar
 Ding W, Wu F, Wu X, Li S, Li H: Adaptive directional liftingbased wavelet transform for image coding. IEEE Trans Image Process 2007, 10(2):416427.MathSciNetView ArticleGoogle Scholar
 Boulgouris NV, Strintzis MG: Reversible multiresolution image coding based on adaptive lifting. In IEEE International Conference on Image Processing. Volume 3. Kobe, Japan; 1999:546550.Google Scholar
 Claypoole RL, Davis G, Sweldens W, Baraniuk RG: Nonlinear wavelet transforms for image coding. the 31st Asilomar Conference on Signals, Systems and Computers 1997, 1: 662667.Google Scholar
 Gouze A, Antonini M, Barlaud M, Macq B: Design of signaladapted multidimensional lifting schemes for lossy coding. IEEE Trans Image Process 2004, 13(12):15891603. 10.1109/TIP.2004.837556View ArticleGoogle Scholar
 Solé J, Salembier P: Generalized lifting prediction optimization applied to lossless image compression. IEEE Signal Process Lett 2007, 14(10):695698.View ArticleGoogle Scholar
 Chang CL, Girod B: Direction Adaptive discrete wavelet transform for image compression. IEEE Trans Image Process 2007, 16(5):12891302.MathSciNetView ArticleGoogle Scholar
 Rolon JC, Salembier P: Generalized lifting for sparse image representation and coding. In Picture Coding Symposium. Lisbon, Portugal; 2007.Google Scholar
 Liu Y, Ngan KN: Weighted adaptive liftingbased wavelet transform for image coding. IEEE Trans Image Process 2008, 17(4):500511.MathSciNetView ArticleGoogle Scholar
 Mallat S: Geometrical grouplets. Appl Comput Harmonic Anal 2009, 26(2):161180. 10.1016/j.acha.2008.03.004MathSciNetView ArticleGoogle Scholar
 Candes JE, Donoho LD: New tight frames of curvelets and optimal representations of objects with piecewise C^{2}singularities. Commun Pure Appl Math 2004, 57(2):219266. 10.1002/cpa.10116MathSciNetView ArticleGoogle Scholar
 Do MN, Vetterli M: The contourlet transform: an efficient directional multiresolution image representation. IEEE Trans Image Process 2005, 14(12):20912106.MathSciNetView ArticleGoogle Scholar
 Chappelier V, Guillemot C, Marinkovic S: Image coding with iterated contourlet and wavelet transforms. In International Conference on Image Processing. Volume 5. Singapore; 2004:31573160.Google Scholar
 Pennec EL, Mallat S: Sparse geometric image representations with bandelets. IEEE Trans Image Process 2005, 14(4):423438.MathSciNetView ArticleGoogle Scholar
 Kingsbury NG: Complex wavelets for shift invariant analysis and filtering of signals. J Appl Comput Harmonic Anal 2001, 10: 234253. 10.1006/acha.2000.0343MathSciNetView ArticleGoogle Scholar
 Fowler JE, Boettcher JB, PesquetPopescu B: Image coding using a complex dualtree wavelet transform. In the European Signal Processing Conference. Poznan, Poland; 2007:994998.Google Scholar
 Boettcher JB, Fowler JE: Video coding using a complex wavelet transform and set partitioning. IEEE Signal Process Lett 2007, 14(9):633636.View ArticleGoogle Scholar
 Reeves TH, Kingsbury NG: Overcomplete image coding using iterative projectionbased noise shaping. In International Conference on Image Processing. Volume 3. Rochester, NY; 2007:597600.Google Scholar
 Natarajan BK: Sparse approximate solutions to linear systems. SIAM J Comput 1995, 24(2):227234. 10.1137/S0097539792240406MathSciNetView ArticleGoogle Scholar
 Donoho D: Compressed Sensing. IEEE Trans Inf Theory 2006, 52(4):12891306.MathSciNetView ArticleGoogle Scholar
 Kaaniche M, Pesquet JC, BenazzaBenyahia A, PesquetPopescu B: Twodimensional non separable adaptive lifting scheme for still and stereo image coding. In IEEE International Conference on Acoustics, Speech and Signal Processing. Dallas, Texas, USA; 2010:12981301.Google Scholar
 Kaaniche M, BenazzaBenyahia A, PesquetPopescu B, Pesquet JC: Non separable lifting scheme with adaptive update step for still and stereo image coding. Elsevier Signal Processing: Special issue on Advances in Multirate Filter Bank Structures and Multiscale Representations 2011, 91(12):27672782.View ArticleGoogle Scholar
 PesquetPopescu B: Twostage adaptive filter bank. first filling date 1999/07/27, official filling number 99401919.8, European patent number EP1119911. 1999.Google Scholar
 LoPresto SM, Ramchandran K, Orchard MT: Image coding based on mixture modeling of wavelet coefficients and a fast estimation quantization framework. In Data Compression Conference. Snowbird, USA; 1997:221230.Google Scholar
 Mallat S: A theory for multiresolution signal decomposition. IEEE Trans Pattern Anal Mach Intell 1989, 11: 674693. 10.1109/34.192463View ArticleGoogle Scholar
 Payan F, Antonini M: An efficient bit allocation for compression normal meshes whith an error driven quantization. Comput Aid Geometr Design (Special Issue On Geometric Mesh Processing) 2005, 22(5):466486.MathSciNetView ArticleGoogle Scholar
 Do MN, Vetterli M: Waveletbased texture retrieval using generalized Gaussian density and KullbackLeibler distance. IEEE Trans Image Process 2002, 11(2):146158. 10.1109/83.982822MathSciNetView ArticleGoogle Scholar
 Gish H, Pierce JN: Asymptotically efficient quantizing. IEEE Trans Inf Theory 1969, 14(5):676683.View ArticleGoogle Scholar
 Petrisor T, PesquetPopescu B, Pesquet JC: A Compressed Sensing Approach to FrameBased Multiple Description Coding. In IEEE International Conference on Acoustics, Speech and Signal Processing. Honolulu, HI; 2007:709712.Google Scholar
 Chen S, Doniho D, Saunders M: Atomic decomposition by basis pursuit. SIAM Rev 2001, 43: 129159. 10.1137/S003614450037906XMathSciNetView ArticleGoogle Scholar
 Combettes PL, Pesquet JC: Proximal splitting methods in signal processing. In FixedPoint Algorithms for Inverse Problems in Science and Engineering. Edited by: Bauschke HH, Burachik, R, Combettes, PL, Elser, V, Luke, DR, Wolkowicz, H. SpringerVerlag, New York; 2010.Google Scholar
 Afonso M, BioucasDias JM, Figueiredo MAT: An augmented Lagrangian approach to the constrained optimization formulation of imaging inverse problems. IEEE Trans Image Process 2011, 20(3):681695.MathSciNetView ArticleGoogle Scholar
 Eckstein J, Bertsekas DP: On the DouglasRachford splitting methods and the proximal point algorithm for maximal monotone operators. Math Programm 1992, 55: 293318. 10.1007/BF01581204MathSciNetView ArticleGoogle Scholar
 Moreau JJ: Proximité et dualité dans un espace hilbertien. Bulletin de la Societé Mathématique de France 1965, 93: 273288.MathSciNetGoogle Scholar
 Chaux C, Combettes P, Pesquet JC, Wajs V: A variational formulation for frame based inverse problems. Inverse Probl 2007, 23(4):14951518. 10.1088/02665611/23/4/008MathSciNetView ArticleGoogle Scholar
 Combettes PL, Wajs VR: Signal Recovery by Proximal ForwardBackward Splitting. Multiscale Model Simul 2005, 4(4):11681200. 10.1137/050626090MathSciNetView ArticleGoogle Scholar
 HiriartUrruty JB, Lemaréchal C: Convex Analysis and Minimization Algorithms. SpringerVerlag, Berlin, London; 1993.View ArticleGoogle Scholar
 Rockafellar RT: Convex Analysis. Volume 28. Princeton University Press, N.J; 1970.View ArticleGoogle Scholar
 BricenoArias LM, Combettes PL, Pesquet JC, Pustelnik N: Proximal algorithms for multicomponent image recovery problems. J Math Imag Vision 2010, 41(12):322.MathSciNetView ArticleGoogle Scholar
 Usevitch B: Optimal bit allocation for biorthogonal wavelet coding. In Data Compression Conference. Snowbird, USA; 1996:387395.Google Scholar
 Parrilli S, Cagnazzo M, PesquetPopescu B: Distortion evaluation in transform domain for adaptive lifting schemes. In International Workshop on Multimedia Signal Processing. Cairns, Queensland, Australia; 2008:200205.Google Scholar
 Chandler DM, Hemami SS: VSNR: A waveletbased Visual SignaltoNoise Ratio for natural images. IEEE Trans Image Process 2007, 16(9):22842298.MathSciNetView ArticleGoogle Scholar
 Boulgouris NV, Strintzis MG: A family of waveletbased stereo image coders. IEEE Trans. Circuits and Syst. Video Technol 2002, 12(10):898903.Google Scholar
 Kaaniche M, BenazzaBenyahia A, PesquetPopescu B, Pesquet JC: Vector lifting schemes for stereo image coding. IEEE Trans Image Process 2009, 18(11):24632475.MathSciNetView ArticleGoogle Scholar
 Frajka T, Zeger K: Residual image coding for stereo image compression. Opt Eng 2003, 42: 182189. 10.1117/1.1526492View ArticleGoogle Scholar
 Moellenhoff MS, Maier MW: Characteristics of disparitycompensated stereo image pair residuals. Signal Process: Image Commun 1998, 14: 4955.Google Scholar
 Combettes PL, Pesquet JC: A DouglasRachford splitting approach to nonsmooth convex variational signal recovery. IEEE J Sel Top Signal Process 2007, 1: 564574.View ArticleGoogle Scholar
 Kaaniche M, Pesquet JC, BenazzaBenyahia A, PesquetPopescu B: Schémas de lifting adaptatifs via des critères parcimonieux. In Colloque GRETSI. Bordeaux, France; 2011:4.Google Scholar
Copyright
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.