We present a low computational cost bit allocation procedure dedicated to wavelet compression performed by entropy coded lattice vector quantization (ECLVQ). This approach is based on a previously proposed statistical model called multidimensional mixture of generalized Gaussian densities. Here, we focus on the distribution estimation step which requires to be as fast as possible. We show that the method of moments (MoM) can be used successfully as an alternative to Monte Carlo Markov chain approach (MCMC); this method allows not only to reduce the computational complexity but also to maintain a good estimation performance. Experimental results show the efficiency of our approach in terms of CPU time.

The design of a variable rate joint watermarking and compression (JWC) scheme is examined in this paper. The proposed approach, called modulated lattice vector quantization (MLVQ), is based on dither modulation quantization index modulation (DM-QIM) which allows for embedding information while maintaining good coding performance. In the first part, we propose a specific indexing method to make MLVQ with a multidimensional codebook feasible. Furthermore, a quantization parameter estimation was designed to ensure the invertibility of the embedding. The limitations of the compression performance of JWC schemes are studied in the second part. We show the existence of a coding rate lower bound which depends mainly on the codebook characteristics and dramatically decreases coding performance. To circumvent this drawback, two improved MLVQ schemes are proposed. In the first one, called arbitrary MLVQ, the embedding is performed on part of the signal to ensure a low embedding/coding ratio. In the second one, called deadzone MLVQ, the coding efficiency is further improved by maintaining the sparsity of the quantized host signal. It consists in excluding the sparse signal components from the embedding process then thresholding them. These schemes both applying wavelet coding demonstrate their efficiency as variable rate coders.

Entropy-coded lattice vector quantization (ECLVQ) with codebooks dedicated to independent identically distributed (i.i.d.) generalized Gaussian sources have proven their high coding performances in the wavelet domain. It is well known that wavelet coefficients with high magnitude (corresponding to edges and textures) tend to be clustered in a few amount of vectors. In this paper, we first show that this property has a major influence on the performances of ECLVQ schemes. Since this clustering property cannot be taken into account by the classical i.i.d. assumption, our first proposal is to model the joint distribution of vectors by a multidimensional mixture of generalized Gaussian (MMGG) densities. The main outcome of this MMGG model is to provide a theoretical framework to simply derive from i.i.d. R- D models, the corresponding MMGG R- D models. In a second part, a new codebook better suited to wavelet coding is proposed: the so-called dead zone lattice vector quantizers (DZLVQ). It consists of generalizing the scalar dead zone to vector quantization by thresholding vectors according to their energy. We show that DZLVQ improves the rate-distortion tradeoff. Experimental results are provided for the pyramidal LVQ scheme under the assumption of a multidimensional mixture of Laplacian (MML) densities. Results performed on a set of real life images show the precision of the analytical R- D curves and the efficiency of the DZLVQ scheme.

The problem of the design of a joint watermarking and compression (JWC) scheme allowing an efficient variable rate coding is addressed here. We have proposed in previous works a method, called modulated lattice vector quantization (MLVQ), based on dither modulation and lattice vector quantization (LVQ) and have shown experimentally its good performances. In this paper, we first show theoretically that the use of a multidimensional lattice codebook must be privileged in JWC. To benefit from this property, an indexing method dedicated to the MLVQ codebook is proposed. It is based on the use of the geometrical properties of the MLVQ codebook

In this paper, we present a new bit allocation procedure based on the approximation of the rate distortion (R-D) functions provided by our efficient lattice vector quantizer with pyramidal dead zone (DZLVQ). Here, we show that DZLVQ R-D functions can be efficiently fitted by an exponential model. This property leads to an analytical solution to the bit allocation problem which reduces significantly the complexity of our compression scheme. Furthermore, our method is highly parallelizable. Finally we show that it keeps the very good results in terms of visual quality of DZLQV, as it better preserves fine structures with respect to SPIHT and JPEG2000 at low rates.

To improve lattice vector quantization (LVQ) performance in image compression applications, we propose to design a new code book shape adapted thresholding. As for the scalar dead zone quantizer, the goal here is to remove non significant data, but in our case the n-dimensional dead zone permits the exploitation of the characteristics of source vectors (i.e. blocks of wavelet coefficients). A theoretical rate model is defined for this new truncation shape. It permits tuning of the dead zone parameters in order to reach a minimum of distortion of the quantized source at a given rate, and thus to outperform classical LVQ.

We propose a combined watermarking and compression algorithm based on the embedding of the mark in the quantized data domain. This approach offers a good tradeoff between robustness, invisibility, embedding capacity and compression bit rate. Furthermore, the extraction process does not need knowledge of the compression parameters, even for attacked images. This leads to a watermarking algorithm which is bit-rate independent.