DNA exhibits remarkable potential as a data storage solution due to its impressive storage density and long-term stability, stemming from its inherent biomolecular structure. However, developing this novel medium comes with its own set of challenges, particularly in addressing errors arising from storage and biological manipulations. These challenges are further conditioned by the structural constraints of DNA sequences and cost considerations. In response to these limitations, we have pioneered a novel compression scheme and a cutting-edge Multiple Description Coding (MDC) technique utilizing neural networks for DNA data storage. Our MDC method introduces an innovative approach to encoding data into DNA, specifically designed to withstand errors effectively. Notably, our new compression scheme overperforms classic image compression methods for DNA-data storage. Furthermore, our approach exhibits superiority over conventional MDC methods reliant on auto-encoders. Its distinctive strengths lie in its ability to bypass the need for extensive model training and its enhanced adaptability for fine-tuning redundancy levels. Experimental results demonstrate that our solution competes favorably with the latest DNA data storage methods in the field, offering superior compression rates and robust noise resilience.

Recently, convolutional auto-encoders (CAE) were introduced for image coding. They achieved performance improvements over the state-of-the-art JPEG2000 method. However, these performances were obtained using massive CAEs featuring a large number of parameters and whose training required heavy computational power.In this paper, we address the problem of lossy image compression using a CAE with a small memory footprint and low computational power usage.In this work, we propose a constrained approach and a new structured sparse learning method. We design an algorithm and test it on three constraints: the classical ℓ1constraint, the ℓ1,∞and the new ℓ1,1constraint. Experimental results show that the ℓ1,1constraint provides the best structured sparsity, resulting in a high reduction of memory ( 82 %) and computational cost reduction (25 %), with similar rate-distortion performance as with dense networks.

Remote control vehicles require the transmission of large amounts of data, and video is one of the most important sources for the driver. To ensure reliable video transmission, the encoded video stream is transmitted simultaneously over multiple channels. However, this solution incurs a high transmission cost. To address this issue, it is necessary to use more efficient video encoding methods that can make the video stream robust to noise. Moreover it should have a less complexity to adapt to the real time requirement. In this paper, we propose a low-complexity, low-latency 2-channel Multiple Description Coding (MDC) solution with an adaptive Instantaneous Decoder Refresh (IDR) frame period, which is compatible with the HEVC standard with adaptive redundancy adjustment. This method shows a better resistance to high packet loss rates with lower complexity.

Over the past years, the ever-growing trend on data storage demand, more specifically for "cold" data (i.e. rarely accessed), has motivated research for alternative systems of data storage. Because of its biochemical characteristics, synthetic DNA molecules are now considered as serious candidates for this new kind of storage. This paper introduces a novel arithmetic coder for DNA data storage, and presents some results on a lossy JPEG 2000 based image compression method adapted for DNA data storage that uses this novel coder.The DNA coding algorithms presented here have been designed to efficiently compress images, encode them into a quaternary code, and finally store them into synthetic DNA molecules. This work also aims at making the compression models better fit the problematic that we encounter when storing data into DNA, namely the fact that the DNA writing, storing and reading methods are error prone processes.The main take away of this work is our arithmetic coder and it's integration into a performant image codec.

Multiple Description Coding (MDC) is an error-resilient source coding method designed for transmission over noisy channels. We present a novel MDC scheme employing a neural network based on implicit neural representation. This involves overfitting the neural representation for images. Each description is transmitted along with model parameters and its respective latent spaces. Our method has advantages over traditional MDC that utilizes auto-encoders, such as eliminating the need for model training and offering high flexibility in redundancy adjustment. Experiments demonstrate that our solution is competitive with autoencoder-based MDC and classic MDC based on HEVC, delivering superior visual quality.

Over the past years, the ever-growing trend on data storage demand, more specifically for “cold” data (rarely accessed data), has motivated research for alternative systems of data storage. Because of its biochemical characteristics, synthetic DNA molecules are now considered as serious candidates for this new kind of storage. This paper presents some results on lossy image compression methods based on convolutional autoencoders adapted to DNA data storage, with synthetic DNA-adapted entropic and fixed-length codes. The model architectures presented here have been designed to efficiently compress images, encode them into a quaternary code, and finally store them into synthetic DNA molecules. This work also aims at making the compression models better fit the problematics that we encounter when storing data into DNA, namely the fact that the DNA writing, storing and reading methods are error prone processes. The main take aways of this kind of compressive autoencoder are our latent space quantization and the different DNA adapted entropy coders used to encode the quantized latent space, which are an improvement over the fixed length DNA adapted coders that were previously used.

The goal of this work is to propose a simple yet efficient way to dynamically transform a sequence of images according to the functional properties of the visual system. To achieve this goal, we extend to video sequences the Retina-Inspired Filter (RIF), which we have recently proposed for still images. Under the assumption that the input signal remains constant for a given time, the RIF decomposition was proven to be invertible, meaning that the image could be perfectly recovered. In this paper, we relax this assumption into a piece-wise constant input and we prove that RIF can be applied to a Group Of Pictures (GOP). Under the condition that a GOP consists of frames without strong pixel motion, we mathematically prove and experimentally show that when RIF is applied to GOP, whatever the size of the GOP is, we are still able to perfectly recover the video frames and at the same time simplify the complexity of the whole process. In addition, we show that while the GOP size increases, the memory cost required to store this amount of frames is sufficiently reduced.

We present an original workflow for structuring a point cloud generated from several scans. Our representation is based on a set of local graphs. Each graph is constructed from the depth map provided by each scan. The graphs are then connected together via the overlapping areas, and careful consideration of the redundant points in these regions leads to a piecewise and globally consistent structure for the underlying surface sampled by the point cloud. The proposed workflow allows structuring aggregated point clouds, scan after scan, whatever the number of acquisitions and the number of points per acquisition, even on computers with very limited memory capacities. To show that our structure can be highly relevant for the community, where the gigantic amount of data represents a real scientific challenge per se, we present an algorithm based on this structure capable of resampling billions of points on standard computers. This application is particularly attractive for simplifying and visualizing gigantic point clouds representing very large-scale scenes (buildings, urban scenes, historical sites...), which often require a prohibitive number of points to describe them accurately.

Digital media explosion has led to an exponential increase of the amount of data generated worldwide and the need for new means of storage able to keep up with the current growth of digital information has become a critical challenge. During the last decade, DNA has been proven to be a potential candidate thanks to its biological properties allowing to store information at high density (215 petabytes in 1 gram) for centuries. In previous works we have presented an end-to-end storage workflow specifically designed for the efficient storage of images onto synthetic DNA and proven its feasibility in a wet-lab experiment in which sequencing was performed using the Illumina machine. In this work we are studying the sequencing using rather the MinION sequencer on the same data after being stored in a sealed capsule for two years. MinION is a very promising sequencer although introducing a much higher error rate in the process of reading. In this paper, we propose a solution to deal with the MinION sequencing noise allowing to recover the original stored data.

The exponential increase of digital data and the limited capacity of current storage devices have made clear the need for exploring new storage solutions. Thanks to its biological properties, DNA has proven to be a potential candidate for this task, allowing the storage of information at a high density for hundreds or even thousands of years. With the release of nanopore sequencing technologies, DNA data storage is one step closer to become a reality. Many works have proposed solutions for the simulation of this sequencing step, aiming to ease the development of algorithms addressing nanopore-sequenced reads. However, these simulators target the sequencing of complete genomes, whose characteristics differ from the ones of synthetic DNA. This work presents a nanopore sequencing simulator targeting synthetic DNA on the context of DNA data storage.

This paper introduces a novel coding/decoding mechanism that mimics one of the most important properties of the human visual system: its ability to enhance the visual perception quality in time. In other words, the brain takes advantage of time to process and clarify the details of the visual scene. This characteristic is yet to be considered by the state-of-the-art quantization mechanisms that process the visual information regardless the duration of time it appears in the visual scene. We propose a compression architecture built of neuroscience models; it first uses the leaky integrate-and-fire (LIF) model to transform the visual stimulus into a spike train and then it combines two different kinds of spike interpretation mechanisms (SIM), the time-SIM and the rate-SIM for the encoding of the spike train. The time-SIM allows a high quality interpretation of the neural code and the rate-SIM allows a simple decoding mechanism by counting the spikes. For that reason, the proposed mechanisms is called Dual-SIM quantizer (Dual-SIMQ). We show that (i) the time-dependency of Dual-SIMQ automatically controls the reconstruction accuracy of the visual stimulus, (ii) the numerical comparison of Dual-SIMQ to the state-of-the-art shows that the performance of the proposed algorithm is similar to the uniform quantization schema while it approximates the optimal behavior of the non-uniform quantization schema and (iii) from the perceptual point of view the reconstruction quality using the Dual-SIMQ is higher than the state-of-the-art.

Rapid technological advances and the increasing use of social media has caused a tremendous increase in the generation of digital data, a fact that imposes nowadays a great challenge for the field of digital data storage due to the short-term reliability of conventional storage devices. Hard disks, fiash, tape or even optical storage have a durability of 5 to 20 years while running data centers also require huge amounts of energy. An alternative to hard drives is the use of DNA, which is life's information-storage material, as a means of digital data storage. Recent works have proven that storing digital data into DNA is not only feasible but also very promising as the DNA's biological properties allow the storage of a great amount of information into an extraordinary small volume for centuries or even longer with no loss of information. In this work we present an extended end-to-end storage workflow specially designed for the efficient storage of images onto synthetic DNA. This workflow uses a new encoding algorithm which serves the needs of image compression while also being robust to the biological errors which may corrupt the encoding.

The exponential increase of digital data that is being generated every year along with the capacity and durability limits of conventional storage devices are raising one of the greatest challenges for the field of data storage. The use of DNA for digital data archiving is a very promising alternative as the biological properties of the DNA molecule allow the storage of a huge amount of information into a very limited volume while also promising data longevity for centuries or even longer. In this paper we present a comparative study of our work with the state of the art solutions, and show that our solution is competitive.

The aggressive growth of digital data threatens to exceed the capacity of conventional storage devices. The need for new means to store digital information has brought great interest in novel solutions as it is DNA, whose biological properties allow the storage of information at a high density and preserve it without any information loss for hundreds of years when stored under specific conditions. Despite being a promising solution, DNA storage faces two major obstacles: the large cost of synthesis and the high error rate introduced during sequencing. While most of the works focus on adding redundancy aiming for effective error correction, this work combines noise resistance to minimize the impact of the errors in the decoded data and post-processing to further improve the quality of the decoding.

The exponential growth in the generation of digital information creates a big challenge for data storage given the capacity limitations of conventional storage devices. Recent works have proposed DNA as a means of digital data storage proposing a novel solution for long-term storage. Although having many advantages, DNA storage is a challenging topic due to the error-prone process of DNA sequencing (reading). To deal with this error most existing works focus on the introduction of error-correction methods. However, most of those methods introduce important redundancy without promising full error correction for the widely used sequencing method using the Nanopore sequencer. This work focuses on noise resistance rather than error-correction proposing a new algorithm for optimally assigning VQ indices to DNA codewords while reducing the visual impact of substitution errors that are caused during sequencing.

This interdisciplinary work focuses on the interest of a new proximal algorithm (ADRS ) for supervised classification of live cell populations growing in a thermostated imaging station and acquired by a Quantitative Phase Imaging (QPI) camera. This type of camera produces interferograms that have to be processed in order to extract features derived from quantitative linear retardance and birefringence measurements. We monitor cell cycling in different populations with both a classical fluorescent DNA intercalating agent imaging on the one hand and QPI without any cellular manipulation nor treatment on the other hand. We show that the accuracy of the classification of these cells in different phases of the cell cycle is equivalent, if not better, when using QPI features as compared to fluorescence imaging features. This is a very important finding since we demonstrate that it is now possible to very precisely follow cell growth under regular culture conditions without any bias. No dye or any kind of markers are necessary for this live monitoring, thus the cells normal physiology is not at all affected by this non invasive procedure. Any studies requiring analysis of cell growth or cellular response to any kind of treatment could benefit from this new approach.

This paper aims to build an image coding system based on a model of the mammalian retina. The retina is the light-sensitive layer of tissue located on the inner coat of the eye and it is responsible for vision. Inspired by the way the retina handles and compresses visual information and based on previous studies we aim to build and analytically study a retinal-inspired image quantizer, based on the Leaky Integrate-and-Fire (LIF) model, a neural model approximating the behavior of the ganglion cells of the Ganglionic retinal layer that is responsible for visual data compression. In order to have a more concrete view of the encoder's behavior, in our experiments, we make use of the spatiotemporal decomposition layers provided by extensive studies on a previous retinal layer, the Outer Plexiform Layer (OPL). The decomposition layers produced by the OPL, are being encoded using our LIF image encoder and then, they are reconstructed to observe the encoder's efficiency.

This paper presents a novel neuro-inspired quantization model which is the extension of the recently released perfect-Leaky Integrate and Fire (LIF) model. We propose that the LIF, which is a very efficient neuromathematical model that describes the spike generation neural mechanism, can lead to a groundbreaking and above all dynamic compression algorithm which is called LIF encoder/decoder. We also prove that under some assumptions, there is a link between the novel LIF encoder/decoder and the conventional Uniform Deadzone Quantizer (UDQ).

This paper introduces a novel filter, which is inspired by the human retina. The human retina consists of three different layers: the Outer Plexiform Layer (OPL), the inner plexiform layer, and the ganglionic layer. Our inspiration is the linear transform which takes place in the OPL and has been mathematically described by the neuroscientific model "virtual retina." This model is the cornerstone to derive the non-separable spatio-temporal OPL retina-inspired filter, briefly renamed retina-inspired filter, studied in this paper. This filter is connected to the dynamic behavior of the retina, which enables the retina to increase the sharpness of the visual stimulus during filtering before its transmission to the brain. We establish that this retina-inspired transform forms a group of spatio-temporal Weighted Difference of Gaussian (WDoG) filters when it is applied to a still image visible for a given time. We analyze the spatial frequency bandwidth of the retina-inspired filter with respect to time. It is shown that the WDoG spectrum varies from a lowpass filter to a bandpass filter. Therefore, while time increases, the retina-inspired filter enables to extract different kinds of information from the input image. Finally, we discuss the benefits of using the retina-inspired filter in image processing applications such as edge detection and compression.

We propose a new compression method devoted to large structured hexahedral meshes having discontinuities. It is dedicated to applications such as visualization or physical simulations whose management by any workstation or mobile device with limited memory and bandwidth is critical. Our method relies on a multiresolution analysis that generates a hierarchy of meshes at increasing resolutions. Our technique also uses a discontinuity tracking feature for their preservation, whatever the resolution, and consequently maintains a coherent geometry with respect to the original mesh. Experimental results emphasize the quality of our compression, in terms of both geometrical distortion and compression ratio.

This paper considers the Softcast joint source-channel video coding scheme for data transmission over parallel channels with different power constraints and noise characteristics, typical in DSL or PLT channels. To minimize the mean square error at receiver, an optimal precoding matrix design problem has to be solved, which requires the solution of an inverse eigenvalue problem. Such solution is taken from the MIMO channel precoder design literature. Alternative suboptimal precoding matrices are also proposed and analyzed, showing the efficiency of the optimal precoding matrix within Softcast, which provides gains increasing with the encoded video quality.

In this paper, we aim to propose a video codec based on the novel retina-inspired filter and retina-inspired quantizer which both perform according to the early visual system. The recently released non-separable spatiotemporal OPL retina-inspired filter enables to progressively extract different kind of information from the input signal which is the sequence of pictures of a video stream. This retina inspired transform has been proven to be a redundant frame which ensures a perfect reconstruction when no quantization appears. The reduction of this redundancy is achieved by a quantization which is inspired by the spike generation mechanism of ganglion cells. This mechanism has been approximated by the Rank Order Coder (ROC) and the Leaky-Integrate and Fire (LIF) models. The ROC model encodes the rank of the spikes and it has been proposed as a complete and very efficient codec for still-images. However, its limitations concerning the reconstruction method forced us to focus our attention on LIF which encodes the spike delays. We approximate the LIF by a scalar quantizer with a dead-zone. This is the first attempt to build a complete retina-inspired video codec which gives promising reconstruction results at low bitrate and high reconstruction quality.

This paper introduces a novel non-Separable sPAtioteMporal filter (non-SPAM) which enables the spatiotemporal decomposition of a still-image. The construction of this filter is inspired by the model of the retina which is able to selectively transmit information to the brain. The non-SPAM filter mimics the retinal-way to extract necessary information for a dynamic encoding/decoding system. We applied the non-SPAM filter on a still image which is flashed for a long time. We prove that the non-SPAM filter decomposes the still image over a set of time-varying difference of Gaussians, which form a frame. We simulate the analysis and synthesis system based on this frame. This system results in a progressive reconstruction of the input image. Both the theoretical and numerical results show that the quality of the reconstruction improves while the time increases.

The coding efficiency of the new video coding standard, High Efficiency Video Coding (HEVC), is strongly associated with better use of spatio-temporal redundancies thanks to an increased number of competing coding modes. However, this competition involves a massive increase in signaling bitrate which becomes a possible limit for the next generation of encoder. This paper proposes a new coding scheme that breaks with conventional approaches. It exploits a more complex decoder able to reproduce the choice of the encoder based on causal references, eliminating thus the need to signal coding modes and associated parameters. The general outline of this new codec and a proposed implementation are described in this paper. Experimental results under common test conditions report an average bitrate saving of 1.7% at the same quality compared to HEVC for a wide range of video sequences.

The paper addresses random field time series analysis and simulation. The analysis constrains a spatial isotropic fractional Brownian field to a dynamic temporal behavior from separable time varying Hurst parameters. The constrained dynamic applies by embedding the wavelet packet spectrum of the input random field into different spectra associated with the same random family (exponential spectrum decay). The paper highlights the relevance of the approach for representing and simulating isotropic light source and cloud dynamics.

Asynchronous event-based cameras use time encoding to code the pixel intensity values. A time encoding of a random valued pixel is a representation of the intensity of this pixel as a random sequence of strictly increasing times. The goal of this paper is the estimation of the pixel mean value from asynchronous samples given by the integrate and fire time encoding. The optimal uniformly minimum variance unbiased estimator is calculated and its statistical performance is compared with a conventional frame-based estimator which exploits regular samples of the pixel intensity. Time encoding significantly reduces the mean number of bits needed to minimize the mean square error of the estimate. Hence, time encoding saves power compared to regular sampling.

This paper deals with stego-image steganalysis to detect hidden information in natural images. Hidden bits are embedded by using the Least Significant Bit (LSB) replacement mechanism. We address the problem of learning the weights which characterize the structure and the performance of the standard Weighted Stego-image (WS) detector. In this paper we propose a new Hybrid Weighted Stego-detection (HWS) algorithm. We assume that the WS weights are related to the image pixels variance through an unknown function which is decomposed onto a set of known basis functions. This yields a linear detector which consists of a linear combination of parametric features derived from the structure of the standard WS detector. The coefficients of the linear combination are learnt by minimizing calibrated losses using stochastic gradient descent or a more efficient stochastic Newton descent approach. Thus, the HWS algorithm benefits from two fundamental advantages: the posterior probability of detection is well estimated and the numerical complexity of the algorithm is linear with the number of samples and the dimension of the features. The benchmark on real images shows that HWS method outperforms standard WS baseline method.

We propose in this paper a robust simplification technique, which preserves geometric features such as sharp edges or corners from original surfaces. To achieve this goal, our simplification process relies on a detection tool that enables to preserve the sharp features during the three subsequent steps: a Poisson disk sampling that intelligently reduces the number of vertices of the initial mesh; the meshing of the samples that aligns the edges along the feature lines; and a constrained relaxation step that improves the shape of the triangles of our final simplified mesh. Experimental results show that our method always produces valid meshes without aliasing artifacts, and without giving up the shape fidelity and quality of the mesh elements.

Our objective is to include in stereoscopic 3D acquisition systems new technologies to automatically detect deformations on aircraft fuselages. We propose in this paper a semiregular mesh reconstruction dedicated to stereoscopic scanners, combined to a multiresolution analysis tool that detects dents on smooth surfaces. The proposed technique for reconstruction is based on a coarse-to-fine approach, and creates semiregular meshes directly from the stereoscopic images. The output of our scanner is thus a structured mesh, by the way well suited for many applications unlike most of scanners that generate only point clouds. Local distances are then calculated between this semiregular mesh and a smooth version of it, in order to automatically detect dents on the scanned surface. The smooth version is obtained via a technique based on multiresolution analysis. Experimental results show the reliability of our contributions on scanned aircraft fuselages.

Our goal is to revisit rank order coding by proposing an original exact decoding procedure for it. Rank order coding was proposed by Thorpe et al. who stated that the order in which the retina cells are activated encodes for the visual stimulus. Based on this idea, the authors proposed in [1] a rank order coder/decoder associated to a retinal model. Though, it appeared that the decoding procedure employed yields reconstruction errors that limit the model bit-cost/quality performances when used as an image codec. The attempts made in the literature to overcome this issue are time consuming and alter the coding procedure, or are lacking mathematical support and feasibility for standard size images. Here we solve this problem in an original fashion by using the frames theory, where a frame of a vector space designates an extension for the notion of basis. Our contribution is twofold. First, we prove that the analyzing filter bank considered is a frame, and then we define the corresponding dual frame that is necessary for the exact image reconstruction. Second, to deal with the problem of memory overhead, we design a recursive out-of-core blockwise algorithm for the computation of this dual frame. Our work provides a mathematical formalism for the retinal model under study and defines a simple and exact reverse transform for it with over than 265 dB of increase in the peak signal-to-noise ratio quality compared to [1]. Furthermore, the framework presented here can be extended to several models of the visual cortical areas using redundant representations.

The motion compensation is one of the most important mechanisms in the context of video coding, allowing the efficient coding of temporal information. Video coding standards have been using entropy coding for motion information which is more prone to transmission errors than fixed-length coding. In this paper, we propose a joint source-channel decoding (JSCD) scheme for robustly decoding motion vectors, enabling the traditional usage of motion compensation in noisy channel scenarios while still maintaining a good coding efficiency. At the encoder, motion vectors are simply coded using scalar quantization enabling scalable video coding setups. The decoder then tries to recover corrupted motion vectors by means of a maximum-a-posteriori (MAP) estimation. This scheme works in open-loop mode, not requiring feedback from the decoder, which is equally important for noisy channel environments. Because some of the coding complexity is brought from the encoder to the decoder, this scheme can be employed in nontraditional applications like video streaming from low power mobile peers.

In this paper, we propose an optimization of the lifted Butterfly scheme for semi-regular meshes. This optimization consists in adapting prediction and update steps at each level of resolution for a given semi-regular mesh. The motivation is the improvement of the multiresolution analysis in order to increase the compression performances of the subsequent geometry coder. We first compute an optimized prediction filter that minimizes the L1-norm of the wavelet coefficients for each level of resolution, independently. We then compute the update filter in order to preserve the data average (0thmoment) at the lower resolution. Experimental results show that our technique globally reduces the entropy of the wavelet coefficients of any semi-regular mesh. Consequently our contribution also improves the compression performances of the zerotree coder PGC.

The rate allocation tradeoff between motion vectors and transform coefficients has a major importance when it comes to efficient video compression. This paper introduces a new coding mode for an H.264/AVC-like video coder, which improves the management of this resource allocation. The proposed technique can be used within any hybrid video encoder allowing a different coding mode for any macroblock. The key tool of the new mode is the lossy coding of motion vectors, obtained via quantization: while the transformed motion-compensated residual is computed with a high-precision motion vector, the motion vector itself is quantized before being sent to the decoder, in a rate/distortion optimized way. Several problems have to be faced with in order to get an efficient implementation of the coding mode, especially the coding and prediction of the quantized motion vectors, and the selection and encoding of the quantization steps. This new coding mode improves the performance of the hybrid video encoder over several sequences at different resolutions.

New standardization activities have been recently launched by the JCT-VC experts group in order to challenge the current video compression standard H.264/AVC. Several improvements of this standard, previously integrated in the JM key technical area software, are already known and gathered in the high efficiency video coding test model. In particular, competition-based motion vector prediction has proved its efficiency. However, the targeted 50% bitrate saving for equivalent quality is not yet achieved. In this context, this paper proposes to reduce the signaling information resulting from this motion vector competition, by using data hiding techniques. As data hiding and video compression traditionally have contradictory goals, an advanced study of data hiding schemes is first performed. Then, an original way of using data hiding for video compression is proposed. The main idea of this paper is to hide the competition index into appropriately selected chroma and luma transform coefficients. To minimize the prediction errors, the transform coefficients modification is performed via a rate-distortion optimization. The proposed scheme is evaluated on several low and high resolution sequences. Objective improvements (up to 2.40% bitrate saving) and subjective assessment of the chroma loss are reported.

We present a novel bio-inspired static image compression scheme. Our model is a combination of a simplified spiking retina model and well known data compression techniques. The fundamental hypothesis behind this work is that the mammalian retina generates an efficient neural code associated to the visual flux. The main novelty of this work is to show how this neural code can be exploited in the context of still image compression. Our model has three main stages. The first stage is the bio-inspired retina model proposed by Thorpe et al, which transforms an image into a wave of spikes. This transform is based on the so-called rank order coding. In the second stage, we show how this wave of spikes can be expressed using a 4-ary dictionary alphabet, through a stack run coder. The third stage consists of applying a first order arithmetic coder to the stack run coded signal. We compare our results to JPEG standards and we show that our model has comparable performance for lower computational cost under strong bit rate restrictions when data is highly contaminated with noise. In addition, our model offers scalability for monitoring data transmission flow. The subject matter presented highlights a variety of important issues in the conception of novel bio-inspired compression schemes and additionally presents many potential avenues for future research efforts.

New activities have been recently launched in order to challenge the H.264/AVC standard. Several improvements of this standard are already known, however the targeted 50% bitrate saving for equivalent quality is not yet achieved. In this context, a previous work proposes to use data hiding techniques to reduce the signaling information resulting from an improvement of Inter-coding. The main idea is to hide the indices into appropriately selected chroma and luma transform coefficients. To minimize the prediction errors, the modification is performed via a rate-distortion optimization. In this paper, the scheme is extended for Intra-coding and 4:4:4 sequences in order to explore the limits highlighted by the previous study and tackle some remaining issues. A different rate-distortion optimization built on the Pareto theory is also proposed. Resulting improvements (1.5% in average) for several sequences are reported and analyzed.

Wireless channels are more prone to transmission errors than wired communications counterparts, leading to excessive packet retransmission that is very inconvenient on media streaming. The multiple-description-coding (MDC) has proven to be very effective against transmission errors, thus providing a solution for that problem. What is presented is a method for MDC using the highly optimized and scalable JPEG 2000. Descriptions are encoded by our modified version of JPEG 2000 which produces compatible codestreams provided with key-error detection registers. The multiple-description JPEG 2000 decoder is then capable of precisely detecting transmission errors and to efficiently choose between available description-information, achieved by a clever exploitation of the EBCOT system. To test the potential of the proposed method, it is integrated as a spatial MD-Coder in a state-of-the-art joint-source channel video coder framework capable of an efficient bit-allocation between descriptions. A comprehensive set of experimental results is presented.

We investigate, in this paper, the processing of stimuli in the mammalians retina, and raise the analogy between the biological mechanisms involved and already existing analog-to-digital converters functioning. Besides, we propose a possible decoding procedure for the retina neural code under the restrictions of the model presented. The coder/decoder, we describe here, focuses on the temporal behavior of the three last retina layers. As time goes, our system gradually changes from a quasi-uniform quantizer to a highly non-linear one. Besides, high magnitude stimuli are well refined, while small magnitudes are coarsely approximated. This yields an original bioinspired quantization system, the behavior of which evolves dynamically during the time interval of stimuli observation. Here, we present a biologically realistic retina model adapted to a temporal signal. Then, we explore the input/output map of the system and its ability to recover the original signal. Further, we make the parallel between this bioinspired system and well known compandor/quantizer systems used for analog-to-digital converters. Finally, we compare the performance of our quantizer to the dead zone uniform scalar quantizer used in JPEG2000, and show a slightly better behavior for low rate transmissions.

We propose an adaptive semi-regular remeshing algorithm for surface meshes. Our algorithm uses Voronoi tessellations during both simplification and refinement stages. During simplification, the algorithm constructs a first centroidal Voronoi tessellation of the vertices of the input mesh. The sites of the Voronoi cells are the vertices of the base mesh of the semi-regular output. During refinement, the new vertices added at each resolution level by regular subdivision are considered as new Voronoi sites. We then use the Lloyd relaxation algorithm to update their position, and finally we obtain uniform semi-regular meshes. Our algorithm also enables adaptive remeshing by tuning a threshold based on the mass probability of the Voronoi sites added by subdivision. Experimentation shows that our technique produces semi-regular meshes of high quality, with significantly less triangles than state of the art techniques.

Our proposal is a method for computing the distance between two surfaces modeled by massive triangle meshes which can not be both loaded entirely in memory. The method consists in loading at each step a small part of the two meshes and computing the symmetrical distance for these areas. These areas are chosen in such a way as the orthogonal projection, used to compute this distance, have to be in it. For this, one of the two meshes is simplified and then a correspondence between the simplified mesh and the triangles of the input meshes is done. The experiments show that the proposed method is very efficient in terms of memory cost, while producing results comparable to the existent tools for the small and medium size meshes. Moreover, the proposed method enables us to compute the distance for massive meshes.

Since the finalization of the H.264/AVC standard and in order to meet the target set by both ITU-T and MPEG to define a new standard that reaches 50% bit rate reduction compared to H.264/AVC, many tools have efficiently improved the texture coding and the motion compensation accuracy. These improvements have resulted in increasing the proportion of bit rate allocated to motion information. Thus, the bit rate reduction of this information becomes a key subject of research. This paper proposes a method for motion vector coding based on an adaptive redistribution of motion vector residuals before entropy coding. Motion information is gathered to forecast a list of motion vector residuals which are redistributed to unexpected residuals of lower coding cost. Compared to H.264/AVC, this scheme provides systematic gain on tested sequences, and 2.3% in average, reaching up to 4.9% for a given sequence.

2010 appears to be the launching date for new compression activities intended to challenge the current video compression standard H.264/AVC. Several improvements of this standard are already known like competition-based motion vector prediction. However the targeted 50% bitrate saving for equivalent quality is not yet achieved. In this context, this paper proposes to reduce the signaling information resulting from this vector competition, by using data hiding techniques. As data hiding and video compression traditionally have contradictory goals, a study of data hiding is first performed. Then, an efficient way of using data hiding for video compression is proposed. The main idea is to hide the indices into appropriately selected chroma and luma transform coefficients. To minimize the prediction errors, the modification is performed via a rate-distortion optimization. Objective improvements (up to 2.3% bitrate saving) and subjective assess ment of chroma loss are reported and analyzed for several sequences.

This paper proposes a new bit plane coding method for signed integer sequences. This method consists in mapping successive bit planes onto quinary symbols (+, -, 0, 1, EoP), where the symbol ldquoEoPrdquo stands for ldquoEnd of Planerdquo, and applying arithmetic coding. Sign bits are efficiently coded in combination with the corresponding most significant bit of non-zero integers. Moreover, bit planes are scanned and coded in a non-sequential manner to exploit the correlation between successive planes. Results for conversational transform coding of wideband speech and audio signals - sampled at 16 kHz - show that the performance/complexity of the proposed bitplane coder is near equivalent to non-embedded coding (stack-run coding), while offering additional flexibility (bitstream scalability).

The problem of efficient video transmission over noisy channels involves high compression rates and robustness to channel errors. In the framework of multiple description coding (MDC), we focus on the direct estimation of the source from two noisy descriptions, without trying to estimate the single descriptions. The challenge is to reconstruct a central signal with distortion as small as possible using the knowledge of the two noisy descriptions. We propose in this paper a maximum a posteriori (MAP) estimator for the decoding of the central description, using the knowledge of the probability density function (pdf) of the different subband descriptions. The balanced MDC scheme used for application is a scan-based wavelet transform video coding scheme, and includes an efficient bit allocation procedure that dispatches the source video redundancy between the different descriptions, depending on the characteristics of the channel. Simulation results show a good robustness of the proposed decoding scheme against transmission errors, with an improvement of 2 dB in PSNR compared to a maximum likelihood (ML) technique.

A view-dependent geometry coding of 3D scenes defined by sets of semi-regular meshes is presented. The objective is to reduce the quantity of significant data to store when visualizing static 3D scenes from a specific point of view. The proposed coding scheme combines a segmentation for determining the visible regions, and an allocation process for improving the visual quality of the encoded scene.

The coding resources used for motion vectors (MVs) can attain quite high ratios even in the case of efficient video coders like H.264, and this can easily lead to suboptimal rate-distortion performance. In a previous paper, we proposed a new coding mode for H.264 based on the quantization of motion vectors (QMV). We only considered the case of 16 times 16 partitions for motion estimation and compensation. That method allowed us to obtain an improved trade-off in the resource allocation between vectors and coefficients, and to achieve better rate-distortion performances with respect to H.264. In this paper, we build on the proposed QMV coding mode, extending it to the case of macroblock partition into smaller blocks. This issue requires solving some problems mainly related to the motion vector coding. We show how this task can be performed efficiently in our framework, obtaining further improvements over the standard coding technique.

This paper introduces an efficient method to compute the wavelet transform for huge 3D volume data like seismic data or 3D medical images with minimum resources. This method consists in a local data processing while reducing considerably the memory requirements. The resulting wavelet transform is identical to the one obtained if the whole 3D object was stored in memory. Experimental results show that the proposed method permits to reduce the memory requirements to the minimum with a low computation time due to the low complexity algorithm.

In this paper, we are interested in wavelet-based coding of 3D semi-regular meshes in order to ensure the progressiveness of the reconstruction. The contribution of this work relies on the adaptation procedure of the lifting scheme operators carried out at each resolution level. More precisely, we propose to firstly segment the original mesh into nonoverlapping regions. The involved predictors are then optimally computed for each region.

In this work we present an efficient coding scheme suitable for lossy image compression using a lattice vector quantizer (LVQ) based on statistically independent data projections. The independence of these components guarantees the optimality of the quantizer. However, this introduces an overload in coding since the projection matrix rendering the components independent needs to be transmitted to the decoder. This issue is tackled by modeling the data such that the projection matrix can be recovered at the decoder side based solely on the model parameters. The original data can thus be recovered based on a reduced descriptive data model and the statistically independent components. Results show that the coding of independent components with a lattice vector quantizer is highly efficient compared with scalar or simple LVQ. Furthermore, the independent data obtained by a model-based projection shows better efficiency without the penalizing coding load of the projection matrix.

This paper proposes a new model-based method for transform coding of audio signals. The input signal is mapped in "perceptual" domain by linear-predictive weighting filter followed by modified discrete cosine transform (MDCT). To provide bitstream scalability, model-based bit plane coding is then applied with respect to the mean square error (MSE) criterion. We present methods to estimate the symbol probability in bit planes assuming a generalized Gaussian model for the distribution of MDCT coefficients. We compare the performance of the proposed bitstream scalable coder with stack-run coding and ITU-T G.722.1. Objective and subjective quality results are presented. The proposed coder is equivalent to or slightly worse than reference coders, but presents the nice advantage of being scalable. Performance penalty due to bitstream scalability is evident at low bitrates.

In this paper, we present a new framework to analyse and process 3D animations (defined by sequences of triangular meshes sharing the same connectivity at any frame). Our idea is to develop a spatio-temporal wavelet filtering for such data, leading to a relevant multiresolution decomposition. In geometry processing, the most efficient spatial wavelets are based on a semiregular sampling. Since the 3D animations generally have an irregular sampling, one of our contribution is a remeshing technique, transforming an animation in a sequence of semiregular meshes, which presents regularity in time but also in space. This new sampling has the advantage to improve the quality of the multiresolution decomposition in the spatial dimension. To show the contribution of such a spatio-temporal filtering in animation processing, we present some experimental results in data compression.

Many compression algorithms consist of quantizing the coefficients of an image in a linear basis. This introduces compression noise that often look like ringing. Recently some authors proposed variational methods to reduce those artifacts. They consists of minimizing a regularizing functional in the set of antecedents of the compressed image. In this paper we propose a fast algorithm to solve that problem. Our experiments lead us to the conclusion that these algorithms effectively reduce oscillations but also reduce contrasts locally. To handle that problem, we propose a fast contrast enhancement procedure. Experiments on a large dataset suggest that this procedure effectively improves the image quality at low bitrates.

The problem of efficient video transmission over noisy channels involves good compression rates and robustness in presence of channel failures. The proposed framework of joint source-channel (JSC) coding is a balanced multiple description coding (MDC) scheme for scan-based wavelet transform video coding. This MDC scheme includes an efficient bit allocation procedure that dispatches the source video redundancy between the different descriptions, depending of the characteristics of the channel. The paper focuses on the joint decoding of two descriptions received at decoder and corrupted by noise. The challenge is to reconstruct a 'central' signal with central distortion as small as possible using the knowledge of the two descriptions. We propose in this paper a maximum a posteriori (MAP) algorithm for the decoding, using the knowledge of the probability density function (pdf) of the different subband descriptions. Promising experimental results are presented.

The MPEG-4 standard adopted a novel technology for compression of dynamic 3D meshes with constant connectivity and time-varying geometry, referred to as FAMC - frame-based animated mesh compression. In this paper, we propose two optimizations of the FAMC approach, aiming at improving the compression efficiency. The first one is based on a PCA (principal component analysis) decomposition of the motion compensation error residuals. The second improves the bi-orthogonal (4-2) wavelet coding approach supported by the standard, by introducing an optimal bit allocation procedure, combined with an adapted quantization of wavelet coefficients. Experimental results show that both optimizations lead to significant gains in compression rate (about 20-30%) at low bitrates.

In this paper we present a new model-based method to code the transform coefficients of audio signals. The histogram of transform coefficients is approximated by a generalized Gaussian model for efficient model-based bit allocation and the spectrum is coded by scalar quantization followed by arithmetic coding. An example coder operating at 16 kHz and using predictive modified discrete cosine transform (MDCT) coding is described. We compare the performance of the proposed coder with ITU-T G.722.1. Objective and subjective quality results are presented. The proposed coder is better than ITU-T G.722.1 at 24 kbit/s and equivalent at 32 kbit/s.

In this paper, We propose an efficient compression method to encode the geometry of 3D mesh sequences of objects sharing the same connectivity. Our approach is based on the clustering of the input mesh geometry into groups of vertices following the same affine motion. The proposed algorithm uses a scan-based temporal wavelet filtering geometrically compensated. The wavelet coefficients are encoded by an efficient coding scheme that includes a bit allocation process, whereas the displacement vectors are lossless entropy encoded. Simulation results provides good compression performances compared to some state of the art coders.

LVQ is a simple but powerful tool for vector quantization and can be viewed as a vector generalization of uniform scalar quantization. Like VQ, LVQ is able to take into account spatial dependencies between adjacent pixels as well as to take advantage of the n-dimensional space filling gain. However, the design of a lattice vector quantizer is not trivial particularly when one wants to use vectors with high dimensions. Indeed, using high dimensions involves lattice codebooks with a huge population that makes indexing difficult. On the other hand, in the framework of wavelet transform, a bit allocation across the subbands must be done in an optimal way. The use of VQ and the lack of non asymptotical distortion-rate models for this kind of quantizers make this operation difficult. In this work we focus on the problem of efficient indexing and optimal bit allocation and propose efficient solutions.

Indexing lattice vectors is a key problem in lattice quantization applications. In this paper we propose a solution to this problem using the lattice leaders and the framework of the theory of partitions. Our solution works for generalized Gaussian distributions sources and allows the use of product codes. It also permits to index high dimensional vectors.

Wavelet-based coding is emerging as a promising framework for efficient and scalable compression of video. Nevertheless, a number of basic tools currently employed in this field have been conceived for hybrid block-based transform coding. This is the case of motion estimation, which generally aims to minimize the energy or the absolute sum of prediction error. However, as wavelet video coders do not employ predictive coding, this is no longer an optimal approach. In this paper we study the problem of the theoretical optimal criterion for wavelet-based video coders, using coding gain as merit figure. A simple solution has been found for a peculiar but useful class of temporal filters. Experiments confirm that the optimally estimated vectors increase the coding gain as well as the performance of a complete video coder, but at the cost of an augmented complexity.

Quality criteria for image coding are often based on mean square error. However, this is not always a relevant measure of visual quality at low bit rates. Here, we investigate the properties of a distortion measure based on the conditional differential entropy of the input signal given its quantized value. The proposed measure appears to be a correct representation of the amount of information lost by quantization. An adaptive bit allocation algorithm is proposed in order to take advantage of this criterion. Experimental results illustrate the behavior of the proposed distortion measure and exhibit interesting visual properties for low bit-rate subband image coding.

In motion-compensated wavelet based video coders, a very precise motion estimation is necessary. However, a motion vectors field of high precision is expensive in binary resources and requires a great place in the bitstream compared to the wavelet coefficients. Thus, we need to reduce the cost of the motion information. To this end, we propose an approach based on a scalable lossy coding of high-precision motion vectors. It allows to optimize the trade-off between motion bit-rate and wavelet coefficients bit-rate, strongly reduces the motion cost, and thus, increases the coder performances at low bit-rate. Obviously, this lossy motion coding has an impact on the decoded sequence. In this paper, we evaluate this impact by establishing a theoretical distortion model for the motion coding error. This model allows the realization of an optimal model-based bit-rate allocation between wavelet subbands and motion vectors. The experimental validation of the model gives satisfactory results

Multiresolution representation of surface meshes is known to be a powerful tool for modeling complex 3D objects. Among the existing schemes, normal meshes have proven to be very attractive for multiresolution and wavelet coding. However, most of the coding methods proposed in the literature are based on scalar quantization despite that vector quantization is known to be more efficient. In this work we propose a novel compression scheme based on a lattice vector quantizer which exploits the correlation inside each geometry subband in a multiresolution framework. Furthermore, we developed a model-based bit allocation algorithm able to work whatever the quantizer is and especially with vector quantization. The proposed scheme allows an improvement up to 1 dB compared to the best state-of-the-art method

In motion-compensated wavelet based video coders (MCWT), it is known that a precise motion estimation is necessary to minimize the wavelet coefficients energy. However, a motion vectors field of high precision is expensive in binary resources compared to the wavelet subbands. In order to reduce the quantity of bits required to represent these vectors, we proposed in a previous work to quantize the motion vectors using a scalable and open-loop lossy coder, while controlling the rate-distortion trade-off. Obviously, this lossy motion coding has an impact on the decoded sequence. In this paper, we propose to evaluate this impact by establishing a theoretical distortion model of motion coding error, including also the subbands quantization noise. This model will allow to realize an optimal model-based bit-rate allocation between wavelet coefficients and motion information. Experimental validation of the model gives interesting results.

Classical quality criteria for image coding are based on the mean square error. We investigate here the properties of a distortion measure based on differential entropy of the error signal. The proposed measure leads to an interesting alternative code design criterion. An adapted bit allocation algorithm is proposed in order to take advantage of this criterion. Experimental results illustrate the behavior of the proposed distortion measure and exhibit interesting psycho-visual properties.

The objective of this paper is to propose an efficient model-based bit allocation process optimizing the performances of a wavelet coder for semiregular meshes. More precisely, this process should compute the best quantizers for the wavelet coefficient subbands that minimize the reconstructed mean square error for one specific target bitrate. In order to design a fast and low complex allocation process, we propose an approximation of the reconstructed mean square error relative to the coding of semiregular mesh geometry. This error is expressed directly from the quantization errors of each coefficient subband. For that purpose, we have to take into account the influence of the wavelet filters on the quantized coefficients. Furthermore, we propose a specific approximation for wavelet transforms based on lifting schemes. Experimentally, we show that, in comparison with a "naive" approximation (depending on the subband levels), using the proposed approximation as distortion criterion during the model-based allocation process improves the performances of a wavelet-based coder for any model, any bitrate, and any lifting scheme.

Motion-compensated lifting schemes have become a reference for the temporal filtering of video data. However, block-based motion estimation and compensation produce annoying blocking artifacts. In this paper, we propose a new lifted temporal filtering method, based on joint segmentation and motion estimation in macroblocks. This method consists in assigning locally the motion information locally to regions instead of blocks, hence the name "puzzle temporal lifting". We first present the puzzle filtering algorithm and we state the conditions of its invertibility. Then, we propose a method to extract regions of occlusion from the motion and segmentation information. The resulting regions are finally exploited within the proposed puzzle filtering with occlusion management. Preliminary experimental results show that the blocking artifacts can be removed, which results in an improvement of visual quality.

Motion compensation has been widely used in both DCT- and wavelet-based video coders for years. The recent success of the temporal wavelet transform based on motion-compensated lifting suggests that a high-performance, scalable wavelet video coder may soon outperform the best DCT-based coders. However, motion-compensated lifting does not implement exactly its transversal equivalent unless certain conditions on motion are satisfied. We review those conditions, and we discuss their importance. We derive a new class of temporal transforms, the so-called 1-N transversal or (N,0) lifting transforms, that are particularly interesting if those conditions on motion are not satisfied. We compare experimentally the 1-3 and 5-3 motion compensated wavelet transforms for the ubiquitous block-motion model used in all video compression standards. For this model, the 1-3 transform outperforms the 5-3 transform due to the need to transmit additional motion information in the latter case. This interesting result, however, does not extend to motion models satisfying the transversal/lifting equivalence conditions.

We present a highly scalable wavelet-based video coder, featuring a scan-based motion-compensated temporal wavelet transform (WT) with lifting schemes which have been specially designed for video. Output bitstream is compatible with JPEG2000, as it is used to compress temporal subbands (SBs). Rate allocation among SBs is done by means of an optimal algorithm, which requires SBs rate-distortion (RD) curves. We propose a model-based approach allowing us to compute these curves with a considerable reduction in complexity. The use of temporal WT and JPEG2000 guarantees high scalability.

In this paper, we analyze the scalability properties of the JPEG2000-compatible video encoder presented in M. Cagnazzo et al., (2004), and we improve its performances by presenting a new technique for an efficient motion vectors (MVs) encoding, producing a motion bitstream also compatible with JPEG2000. Our study shows that, thanks to our encoding strategy and to our peculiar temporal filters, scalably encoded sequences have the same or almost the same quality than non-scalably encoded ones: this is what we call smooth scalability. We also compared our encoder performances with the recent H.264 standard, showing comparable or sometimes better performances.

In this paper, we propose a powerful bit allocation that optimizes the quantization of the normal mesh geometry. This bit allocation aims to minimize the surface-to-surface distance [P. Cignoni et al., 1998] between the original irregular mesh and the quantized normal one, according to a target bitrate. Moreover, to provide a fast bit allocation, we approximate this surface-to-surface distance with a simple criterion depending on the wavelet coefficient distributions, and we use theoretical models. This provides a fast and low-complex model-based bit allocation yielding results better than the recent state-of-the-art methods like [A. Khodakovsky and I. Guskov, 2002].

This paper proposes a new method for the design of lifting filters to compute a multidimensional nonseparable wavelet transform. Our approach is stated in the general case, and is illustrated for the 2-D separable and for the quincunx images. Results are shown for the JPEG2000 database and for satellite images acquired on a quincunx sampling grid. The design of efficient quincunx filters is a difficult challenge which has already been addressed for specific cases. Our approach enables the design of less expensive filters adapted to the signal statistics to enhance the compression efficiency in a more general case. It is based on a two-step lifting scheme and joins the lifting theory with Wiener's optimization. The prediction step is designed in order to minimize the variance of the signal, and the update step is designed in order to minimize a reconstruction error. Application for lossy compression shows the performances of the method.

Summary form only given. The problem of efficient image/video transmission over wireless channels is considered. Such a problem involves good compression rates and effectiveness in presence of channel failures. The proposed work includes a balanced multiple description-coding (MDC) schemes for still image based on the discrete wavelet transform (DWT) developed. This MDC includes an efficient bit allocation procedure that dispatches the source image redundancy between different channels. Extended version for video was presented, and uses the 3D scan-based DWT that involves scan-based MDC with rate or quality control. To solve this problem, redundancy between the descriptors in function of channel model and state (BER) is proposed.

In this paper we address the problem of quincunx sampled images compression. Our objective is to define an efficient bit allocation method adapted to quincunx sampled images. We first estimate the subband optimal weightings for a global distortion estimation. These weightings are derived from the filters used to perform the quincunx wavelet transform. Then, we use our weightings in a model-based bit allocation procedure. Our method uses generalized Gaussians to approximate the probability density function of the wavelet coefficients in each subband. The bit allocation method which is proposed provides both low complexity and high performance.

In this paper, we propose a 3D geometry compression technique for densely sampled surface meshes. Based on a 3D multiresolution analysis (performed by a 3D Discrete wavelet transform for semiregular meshes), this scheme includes a model-based bit allocation process across the wavelet subbands and an efficient surface adapted weighted criterion for 3D wavelet coefficient coordinates. This permits to highly improve the visual quality of quantized meshes obtained by classical bit allocation based on MSE distortion. Moreover, the coefficients are encoded with an original 3D context-based bitplane arithmetic coder. The main contribution of this paper is the introduction of 3D multiresolution contexts adapted to 3D semiregular mesh geometry information.

We present a system for video streaming well adapted to the unpredictable and varying nature of Internet. The proposed system uses a superposition of several multiple description coding (MDC) schemes, each with N = 2 descriptions, to reach rate scalability and adaptability to varying channel conditions. Each MDC (N = 2 descriptions), that we will call base MDC has a bit rate and a redundancy associated. The superposition of several base MDC results on a MDC scheme for N > 2 descriptions with different bit rates, allowing rate scalability, for different redundancies. The proposed scheme is well adapted to varying channel conditions. In the proposed method multiple descriptions are generated by the coder and downloaded in the server, leaving to the server the only task to choose sending out the right description at the right time depending of channel conditions (bandwidth and loss rate).

The goal of this correspondence is to propose a low-complexity enumeration algorithm for lattice vectors, based on a geometrical interpretation and valid for different source distributions, i.e., for different L/sub p/-norms in the range 0

In this paper, we introduce a new decoding algorithm for DCT-based video encoders, such as Motion JPEG (M-JPEG), H26x, or MPEG. This algorithm considers not only the compression artifacts but also the ones due to transmission, acquisition or storage of the video. The novelty of our approach is to jointly tackle these two problems, using a variational approach. The resulting decoder is object-based, allowing independent and adaptive processing of objects and backgrounds, and considers available information provided by the bitstream, such as quantization steps, and motion vectors. Several experiments demonstrate the efficiency of the proposed method. Objective and subjective quality assessment methods are used to evaluate the improvement upon standard algorithms, such as the deblocking and deringing filters included in MPEG-4 postprocessing.

It is well known that compression of very large images (e.g. medical imaging, microscopy, satellite images) requires stripe-based or tiling processing. In some applications, the transmission of compressed data is performed through a rate constrained channel. Thus, rate allocation and control procedures have to be used to fit the channel characteristics. On the other hand, most applications require high quality image coding without any real time rate constraint (e.g. off-line compression for storage or for broadcasting over IP, ADSL, HDTV, etc.). Therefore, we propose a new stripe-based compression algorithm based on quality control. Our method computes first an optimal subband MSE allocation and then, the corresponding quantization steps. The proposed algorithm provides, both accurate local MSE control and a global rate-distortion improvement when compared to a rate constrained compression scheme. Furthermore, it performs better than JPEG2000.

In this paper, we propose an efficient low complexity compression scheme for densely sampled irregular 3D meshes. This scheme is based on 3D multiresolution analysis (3D discrete wavelet transform) and includes a model-based bit allocation process across the wavelet subbands. Coordinates of 3D wavelet coefficients are processed separately and statistically modeled by a generalized Gaussian distribution. This permits an efficient allocation even at a low bitrate and with a very low complexity. We introduce a predictive geometry coding of LF subbands and topology coding is made by using an original edge-based method. The main idea of our approach is the model-based bit allocation adapted to 3D wavelet coefficients and the use of EBCOT coder to efficiently encode the quantized coefficients. Experimental results show compression ratio improvement for similar reconstruction quality compared to the well-known PGC method.

A challenge of image communication over unreliable channels is to achieve good compression rates and be effective in presence of channel failures. In this work we use the multiple description coding (MDC) techniques, based on wavelet transforms, that have been shown to be powerful against channel failures. We propose a bit allocation procedure that dispatch redundancy between the different channels when compressing to a target bit rate with a bounded side distortion. In this way we develop a MDC scheme well adapted to channel noise.

In Pereira et al. (2002) we proposed a balanced multiple description coding (MDC) scheme based on the discrete wavelet transform (DWT). This MDC includes an efficient bit allocation procedure that dispatches the source image redundancy between different channels. The amount of redundancy is controlled according to the bit error rate of the different channels. Here, we propose to extend this approach for low bit rate video transmission. The proposed method uses the 3D scan-based DWT of Parisot et al. (2000) and involves scan-based MDC with rate or quality control.

This paper introduces the problem of designing a quincunx lifting scheme well-adapted to lossy compression applications. The main goal is to design quincunx filters adapted to the statistical properties of the input signal in order to perform a better signal reconstruction after deterioration due to a lossy coding. To perform such a filtering, the basic idea is to combine optimization methods and a lifting scheme. The latter is an attractive tool to perform wavelet transforms.

This paper considers the estimation scenario where an original image has undergone blurring, noise corruption and compression. The authors consider these unwanted effects jointly in the estimation formalism they propose. The contribution of the paper aims to address the joint coding/decoding formulation, involving a priori assumptions on the solution and knowledge of the imaging systems to account for effects due to acquisition noise and compression noise. Most of the authors' construction is based on well-known techniques drawn from a variety of areas in modern signal processing, including optimization theory, wavelet decomposition, optimal bit allocation, inverse problem and bounded noise assumption.

Wavelet coding has been shown to be better than DCT coding. This method outperforms the DCT JPEG codec and moreover allows scalability. The 2D DWT can be easily extended to 3D and thus applied to video coding. However 3D subband coding of video suffers from two drawbacks. The first one is the memory complexity required for coding 3D blocks. The second one is the lack of temporal quality resulting in temporal blocking artifacts or flickering when the DWT is performed on temporal blocks. We propose a new temporal scan-based wavelet transform method for video coding combining the usual advantages of wavelet coding (performance, scalability), with acceptable reduced memory requirements and no additional CPU complexity.

Quincunx sampling is of large interest for image coding applications. Recent remote sensors of satellites return quincunx sampled images. Moreover, a quincunx sampling allows the decomposition of the image into two channels and a twice as accurate multiresolution analysis as the dyadic one. This paper introduces a new construction of the quincunx wavelet transform. This new transform is a bidimensional extension of the factorization of a wavelet transform into a lifting scheme for finite and symmetrical low pass filters. The aim of this method is to deal with quincunx images with appropriate transforms while using advantages offered by the lifting scheme. This method allows us to find new efficient quincunx wavelet filters.

Efficient compression algorithms generally use wavelet transforms. They try to exploit all the signal dependencies that can appear inside and across the different sub-bands of the decomposition. This provides highly complex algorithms that can't generally he implemented for real-time purposes. However, efficiency of a coding scheme highly depends on bit allocation. In this paper, we present a new wavelet based image coder EBWIC (efficient bit allocation wavelet image coder). This algorithm is based on an accurate modelisation of the distortion-rate curve even at low bit rate. It results that the efficiency of our method is very close to JPEG 2000 with a very low complexity and possible parallelization of the encoding process. The method proposed below has a complexity less than 60 arithmetic operations per pixel (against about 300 for JPEG 2000). Our method can be applied whatever the wavelet transform (quincunx, dyadic...) and the entropy coder are. It can be used for strip-based processing.

We propose a new multiresolution variational joint denoising/deblurring approach, involving a priori assumptions on the solution and knowledge of the imaging systems to account for effects due to acquisition noise (edge preservation, degradation noise modeling, bounded noise assumption and spectral control of noise level-whiteness and stationarity). The techniques used are drawn from a variety of areas of modern signal processing, including optimization theory, inverse problem, wavelet packets decomposition and bounded noise assumption.

The increasing demand for real-time applications requires the use of variable-rate quantizers having good performance in the low bit rate domain. In order to minimize the complexity of quantization, as well as maintaining a reasonably high PSNR ratio, we propose to use an entropy-coded lattice vector quantizer (ECLVQ). These quantizers have proven to outperform the well-known EZW algorithm's performance in terms of rate-distortion tradeoff. In this paper, we focus our attention on the modeling of the mean squared error (MSE) distortion and the prefix code rate for ECLVQ. First, we generalize the distortion model of Jeong and Gibson (1993) on fixed-rate cubic quantizers to lattices under a high rate assumption. Second, we derive new rate models for ECLVQ, efficient at low bit rates without any high rate assumptions. Simulation results prove the precision of our models.

Imaging systems involves blur and the coder reduces the binary rate for transmission or storage. These operations remove pertinent information contained by the image, and introduce annoying artifacts. Removing these artifacts allows higher visual quality for the reconstructed data. Unlike usual techniques, which made separately decoding and post-processing, we propose an optimal joint decoding/deblurring method for image reconstruction The goal of this work is to overcome the introduction of these negative effects by taking into account all the acquisition chain model.

This paper introduces an optimal decoding scheme for the baseline Joint Photographers Expert Group (JPEG) standard. In particular, it deals with the minimization of a half-quadratic criterion which takes into account observed data, a priori knowledge of the solution, and precise spatial location of blocking artifacts. The method considers, at the same time, information in the original spatial domain, in the DCT domain, and in the spatio-frequency (DWT) domain. A model of the quantizer is also included. The method reduces blocking and ringing artifacts, resulting in improved peak signal-to-noise ratio performance as well as greater visual quality.

This paper examines the relationship between quantization noise removal and the variational problem. Traditional transformed and quantized image restoration techniques cannot prevent parasitic effects due to quantization noise. We propose a new method, involving a priori assumptions on the solution and knowledge of the coder (transformation and quantization) to account for effects due to quantization noise. This technique, called MORPHE, can be viewed as an inverse problem with optimization of the transform/quantization/decoding structure. This leads to the study of different ways to solve the constrained optimization problem. Experiments using this nonlinear inverse dynamic filtering demonstrate PSNR gains over standard linear inverse filtering as well as appreciable visual improvements.

This paper examines the problem of compressing multispectral images using the wavelet transform and stack-run entropy coding. Our goal is to explore various ways of coding the wavelet coefficients in order to see which techniques can best exploit the correlation between the multispectral bands to produce an efficient coding algorithm. The results of our study indicate that applying the KLT to each subband followed by lattice VQ on Z/sup n/ and subsequent independent entropy coding of each of the lattice dimensions is an effective and fairly simple coding technique. We also demonstrate the importance of proper bit-allocation or, equivalently, the correct choice of the lattice scale parameter for each subband.

Code vector indexing is a key problem in lattice quantization applications. In order to solve this problem, we propose a method based on the idea of a coding table encompassing a set of points as small as possible. Our method works for both spherical and pyramidal code books. It provides a good tradeoff between computational complexity and storage requirements.

Indexing the codevectors is a key problem in lattice quantization applications. We propose a method based on the idea of a coding table encompassing a set of points chosen as small as possible and called the fundamental region. Our method works for both spherical and pyramidal codebooks. It provides a good tradeoff between computational complexity and storage requirements. Finally, it improves of 3 dB on the signal-to-noise ratio of the LBG algorithm at a given memory cost.

The problem of quantizing sub-images of a multiresolution image decomposition while preserving edges is considered. For this purpose, we propose a coding algorithm which exploits the spatial locations of coefficients within scales. This algorithm, we call edge adaptive quantization, preserves edges while smoothing elsewhere once the different significant coefficients are determined. Spatial adaptation is introduced by considering a new spatially constrained scalar quantizer based on Markov random fields. A criterion based on new spatial models is developed as well as an optimization procedure selecting the optimal quantizer in the distortion-spatial constraints sense. This algorithm is dedicated to low bit rate coding.

Presents a new scheme for fractal image compression based on adaptive Delaunay triangulation. Such a partition is computed on an initial set of points obtained with a split and merge algorithm in a grey level dependent way. The triangulation is thus fully flexible and returns a limited number of blocks allowing good compression ratios. Moreover, a second original approach is the integration of a classification step based on a modified version of the Lloyd algorithm (vector quantization) in order to reduce the encoding complexity. The vector quantization algorithm is implemented on pixel histograms directly generated from the triangulation. The aim is to reduce the number of comparisons between the two sets of blocks involved in fractal image compression by keeping only the best representative triangles in the domain blocks set. Quality coding results are achieved at rates between 0.25-0.5 b/pixel depending on the nature of the original image and on the number of triangles retained.

In most of the quantization applications, we need variable rate vector quantizers. In 1988, Chou, Lookabaugh and Gray designed vector quantizers having minimum distortion subject to an entropy constraint. For this purpose, they used a generalization of Lloyd algorithm to n dimensions, called the ECVQ algorithm. We propose to use lattices in order to design entropy constrained lattice vector quantizers (ECLVQ). Low resolution (nonasymptotical) distortion and rate approximation models are given and a generalization of the distortion formula to entropy constraint is formulated. These works generalize those of Gibson (see IEEE Trans. on Inform. Theory, vol.IT-39, no.3, p.786-804, 1993) on fixed rate quantizers, to any cubic lattice Z/sup n/ subject to entropy constraint.

Introduces a new image coding scheme using lattice vector quantization. The proposed method involves two steps: biorthogonal wavelet transform of the image, and lattice vector quantization of wavelet coefficients. In order to obtain a compromise between minimum distortion and bit rate, we must truncate and scale the lattice suitably. To meet this goal, we need to know how many lattice points lie within the truncated area. We investigate the case of Laplacian sources where surfaces of equal probability are spheres for the L(1) metric (pyramids) for arbitrary lattices. We give explicit generating functions for the codebook sizes for the most useful lattices like Z(n), D(n), E(s), wedge(16).

The image coding scheme involves two steps: biorthogonal wavelet transform of the image and pyramidal lattice vector quantization of wavelet coefficients. In order to obtain a compromise between minimum distortion and bit rate, one must truncate and scale the lattice suitably. To meet this goal, one needs to know how many lattice points lie within the truncated area. The case of Laplacian sources where surfaces of equal probability are spheres for the L/sub 1/ metric (pyramids) for arbitrary lattices is investigated. Explicit generating functions for the codebook sizes of the most useful lattices are given.>

A scheme for image compression that takes into account psychovisual features both in the space and frequency domains is proposed. This method involves two steps. First, a wavelet transform used in order to obtain a set of biorthogonal subclasses of images: the original image is decomposed at different scales using a pyramidal algorithm architecture. The decomposition is along the vertical and horizontal directions and maintains constant the number of pixels required to describe the image. Second, according to Shannon's rate distortion theory, the wavelet coefficients are vector quantized using a multiresolution codebook. To encode the wavelet coefficients, a noise shaping bit allocation procedure which assumes that details at high resolution are less visible to the human eye is proposed. In order to allow the receiver to recognize a picture as quickly as possible at minimum cost, a progressive transmission scheme is presented. It is shown that the wavelet transform is particularly well adapted to progressive transmission.

An image coding scheme has been introduced by the authors (see IEEE ICASSP, p.2297, 1990). This scheme involves two steps. A biorthogonal wavelet transform is applied to the original image, and wavelet coefficients are then vector quantized using the LBG (Linde, Buzo and Gray, 1980) method. The purpose of this work is to propose a new scheme for vector quantization of wavelet coefficients. The proposed method is based on lattice vector quantization. The application of the D/sub 4/, E/sub 8/ and Barnes-Wall Lambda /sub 16/ lattices is investigated. These lattices are used to encode wavelet coefficients whose PDFs are close to Laplacian. A variable-length coding method is applied and the trade-off between distortion and optimal rate is investigated. Experimental results on the Lena image using the Lambda /sub 16/ lattice leads to a peak signal-to-noise ratio (PSNR) of 31.14 dB at 0.08 bpp. This result outperforms, to the authors knowledge, all other methods. Edges which are most of interest for image analysis are particularly sharp without any smoothing artefacts.>

A new method is proposed for image coding involving two steps. First, the authors use a dual recursive wavelet transform in order to obtain a set of subclasses of images with better characteristics than the original image (lower entropy, edges discrimination, etc.). Secondly, according to Shannon's rate distortion theory, the wavelet coefficients are vector quantized. The purpose of this work is to present and study new recursive filter banks with perfect reconstruction as an alternative to FIR (finite impulse response) filters which are commonly used in wavelet analysis. The authors present two kinds of experimental results: coding of the well known Lena image with only two multiplications per pixel and then with an optimized IIR (infinite impulse response) filter.>

A two-step scheme for image compression that takes into account psychovisual features in space and frequency domains is proposed. A wavelet transform is first used in order to obtain a set of orthonormal subclasses of images; the original image is decomposed at different scales using a pyramidal algorithm architecture. The decomposition is along the vertical and horizontal directions and maintains the number of pixels required to describe the image at a constant. Second, according to Shannon's rate-distortion theory, the wavelet coefficients are vector quantized using a multiresolution codebook. To encode the wavelet coefficients, a noise-shaping bit-allocation procedure which assumes that details at high resolution are less visible to the human eye is proposed. In order to allow the receiver to recognize a picture as quickly as possible at minimum cost, a progressive transmission scheme is presented. The wavelet transform is particularly well adapted to progressive transmission.>

This paper proposes a new method for image coding involving two steps. First, we use a 'Dual Recursive Wavelet' Transform in order to obtain a set of subclasses of images with better characteristics than the original image (lower entropy, edges discrimination, ... ). Second, according to Shannon's rate distortion theory, the wavelet coefficients are vector quantized using the Kohonen Self-Organizing Feature Maps. We compare this training method with the well known LBG algorithm.