The recent advances in deep neural networks have led us to revisit the safety of current mechanisms for the authentification of people’s identities and validating the provenance of traded goods. This paper studies the problem of identifying counterfeits of Copy Detection Patterns (CDPs). CDPs are used on product packaging to identify counterfeiting by relying on the information loss principle. Current verification techniques rely on a supervised machine learning paradigm, which needs examples from real and fake CDPs to learn a model capable of differentiating both classes of CDPs. This paper proposes an unsupervised forensic approach that is capable of training an Energy-Based Model for detecting out-of-distribution (OOD) images (fake CDP examples in our case) using only original CDP images. This paper also introduces a novel thresholding technique that only requires original CDPs for threshold value selection. We validate our approach on the Indigo dataset, with results demonstrating comparable counterfeit detection capabilities to prior work but using a method trained on less data.

Navigation has been classically solved in robotics through the combination of SLAM and planning. More recently, beyond waypoint planning, problems involving significant components of (visual) high-level reasoning have been explored in simulated environments, mostly addressed with large-scale machine learning, in particular RL, offline-RL or imitation learning. These methods require the agent to learn various skills like local planning, mapping objects and querying the learned spatial representations. In contrast to simpler tasks like waypoint planning (PointGoal), for these more complex tasks the current state-of-the-art models have been thoroughly evaluated in simulation but, to our best knowledge, not yet in real environments. In this work we focus on sim2real transfer. We target the challenging Multi-Object Navigation (Multi-ON) task [41] and port it to a physical environment containing real replicas of the originally virtual Multi-ON objects. We introduce a hybrid navigation method, which decomposes the problem into two different skills: (1) waypoint navigation is addressed with classical SLAM combined with a symbolic planner, whereas (2) exploration, semantic mapping and goal retrieval are dealt with deep neural networks trained with a combination of supervised learning and RL. We show the advantages of this approach compared to end-to-end methods both in simulation and a real environment and outperform the SOTA for this task [28].

Person re- identification is an important task in video surveillance systems. It consists in matching an image of a probe person among a gallery image set of people detected from a network of surveillance cameras with non-overlapping fields of view. The main challenge of person re- identification is to find image representations that are discriminating the persons' identities and that are robust to the viewpoint, body pose, illumination changes and partial occlusions. In this paper, we proposed a metric learning approach based on a deep neural network using a novel loss function which we call the Rank- Triplet loss. This proposed loss function is based on the predicted and ground truth ranking of a list of instances instead of pairs or triplets and takes into account the improvement of evaluation measures during training. Through our experiments on two person re- identification datasets, we show that the new loss outperforms other common loss functions and that our approach achieves state-of-the-art results on these two datasets.

We propose a new method for human pose estimation which leverages information from multiple views to impose a strong prior on articulated pose. The novelty of the method concerns the types of coherence modelled. Consistency is maximised over the different views through different terms modelling classical geometric information (coherence of the resulting poses) as well as appearance information which is modelled as latent variables in the global energy function. Moreover, adequacy of each view is assessed and their contributions are adjusted accordingly. Experiments on the HumanEva and Utrecht multi‐person motion datasets show that the proposed method significantly decreases the estimation error compared to single‐view results.

This paper presents a study of metric learning systems on pairwise identity verification, including pairwise face verification and pairwise speaker verification, respectively. These problems are challenging because the individuals in training and testing are mutually exclusive, and also due to the probable setting of limited training data. For such pairwise verification problems, we present a general framework of metric learning systems and employ the stochastic gradient descent algorithm as the optimization solution. We have studied both similarity metric learning and distance metric learning systems, of either a linear or shallow nonlinear model under both restricted and unrestricted training settings. Extensive experiments demonstrate that with limited training pairs, learning a linear system on similar pairs only is preferable due to its simplicity and superiority, i.e., it generally achieves competitive performance on both the labeled faces in the wild face dataset and the NIST speaker dataset. It is also found that a pretrained deep nonlinear model helps to improve the face verification results significantly.

In this paper, we propose a pedestrian attribute recognition approach and a CNN-based person re-identification framework enhanced by pedestrian attributes. The knowledge of person attributes can help video surveillance tasks like person re-identification as well as person search, semantic video indexing and retrieval to overcome viewpoint changes with their robustness to the inherent visual appearance variations. Compared to previous approaches, our attribute recognition method using Local Maximal Occurrence (LOMO) features and a Multi-Label Multi-Layer Perceptron (MLMLP) classifier proves to be more robust to different view points and is computationally more efficient. The experiments on three public benchmarks show that the proposed method improves the state-of-the art on attribute recognition. Furthermore, we integrate our attribute recognition algorithm into a triplet CNN similarity learning framework for person re-identification fusing both learned CNN features and attributes. This fusion leads to an overall improvement, and we achieve state-of-the-art results on person re-identification.

Objective visual quality assessment of 3D models is a fundamental issue in computer graphics. Quality assessment metrics may allow a wide range of processes to be guided and evaluated, such as level of detail creation, compression, filtering, and so on. Most computer graphics assets are composed of geometric surfaces on which several texture images can be mapped to make the rendering more realistic. While some quality assessment metrics exist for geometric surfaces, almost no research has been conducted on the evaluation of texture-mapped 3D models. In this context, we present a new subjective study to evaluate the perceptual quality of textured meshes, based on a paired comparison protocol. We introduce both texture and geometry distortions on a set of 5 reference models to produce a database of 136 distorted models, evaluated using two rendering protocols. Based on analysis of the results, we propose two new metrics for visual quality assessment of textured mesh, as optimized linear combinations of accurate geometry and texture quality measurements. These proposed perceptual metrics outperform their counterparts in terms of correlation with human opinion. The database, along with the associated subjective scores, will be made publicly available online.

Several perceptually-based quality metrics have been introduced to predict the global impact of geometric artifacts on the visual appearance of a 3D model. They usually produce a single score that reflects the global level of annoyance caused by the distortions. However, beside this global information, it is also important in many applications to obtain information about the local visibility of the artifacts (i.e. estimating a localized distortion measure). In this work we present a psychophysical experiment where observers are asked to mark areas of 3D meshes that contain noticeable distortions. The collected per-vertex distortion maps are first used to illustrate several perceptual mechanisms of the human visual system. They then serve as ground-truth to evaluate the performance of well-known geometric attributes and metrics for predicting the visibility of artifacts. Results show that curvature-based attributes demonstrate excellent performance. As expected, the Hausdorff distance is a poor predictor of the perceived local distortion while the recent perceptually-based metrics provide the best results.

In this paper, we present a novel framework for combining several independent on-line trackers using visual scene context. The aim of our method is to decide automatically at each point in time which specific tracking algorithm works best under the given scene or acquisition conditions. To this end, we define a set of generic global context features computed on each frame of a set of training videos. At the same time, we record the performance of each individual tracker on these videos in terms of object bounding box overlap with the ground truth. Then a classifier is trained to estimate which tracker gives the best result given the global scene context in a particular frame. We experimentally show that such a classifier can predict the best tracker with a precision of over 80% in unknown videos with unknown environments. The proposed tracking method further filters the classifier responses temporarily using a Hidden Markov Model in order to avoid rapid oscillations between different trackers. Finally, we evaluated the overall tracking system and showed that this scene context-based tracker selection considerably improves the overall robustness and compares favourably with the state-of-the-art.

This paper presents a new method for similarity metric learning, called Logistic Similarity Metric Learning (LSML), where the cost is formulated as the logistic loss function, which gives a probability estimation of a pair of faces being similar. Especially, we propose to shift the similarity decision boundary gaining significant performance improvement. We test the proposed method on the face verification problem using four single face descriptors: LBP, OCLBP, SIFT and Gabor wavelets. Extensive experimental results on the LFW-a data set demonstrate that the proposed method achieves competitive state-of-the-art performance on the problem of face verification.

We propose a new method for activity recognition based on a view independent representation of human motion. Robust 3D volume motion templates (VMTs) are calculated from tracklets. View independence is achieved through a rotation with respect to a canonical orientation. From this volumes, features based on 3D gradients are extracted, projected to a codebook and pooled into a bags-of-words model classified with an SVM classifier. Experiments show that the method outperforms the original HoG3D method.

International audience

In this paper we tackle the problem of detecting individual human actions in video sequences. While the most successful methods are based on local features, which proved that they can deal with changes in background, scale and illumination, most existing methods have two main shortcomings: first, they are mainly based on the individual power of spatio-temporal interest points (STIP), and therefore ignore the spatio-temporal relationships between them. Second, these methods mainly focus on direct classification techniques to classify the human activities, as opposed to detection and localization. In order to overcome these limitations, we propose a new approach, which is based on a graph matching algorithm for activity recognition. In contrast to most previous methods which classify entire video sequences, we design a video matching method from two sets of ST-points for human activity recognition. First, points are extracted, and a hyper graphs are constructed from them, i.e. graphs with edges involving more than 2 nodes (3 in our case). The activity recognition problem is then transformed into a problem of finding instances of model graphs in the scene graph. By matching local features instead of classifying entire sequences, our method is able to detect multiple different activities which occur simultaneously in a video sequence. Experiments on two standard datasets demonstrate that our method is comparable to the existing techniques on classification, and that it can, additionally, detect and localize activities.

We present a fast and robust graph matching approach for 2D specific object recognition in images. From a small number of training images, a model graph of the object to learn is automatically built. It contains its local key points as well as their spatial proximity relationships. Training is based on a selection of the most efficient subgraphs using the mutual information. The detection uses dynamic programming with a lattice and thus is very fast. Experiments demonstrate that the proposed method outperforms the specific object detectors of the state-of-the-art in realistic noise conditions.

Existing action recognition approaches mainly rely on the discriminative power of individual local descriptors extracted from spatio-temporal interest points (STIP), while the geometric relationships among the local features are ignored. This paper presents new features, called pairwise features (PWF), which encode both the appearance and the spatio-temporal relations of the local features for action recognition. First STIPs are extracted, then PWFs are constructed by grouping pairs of STIPs which are both close in space and close in time. We propose a combination of two codebooks for video representation. Experiments on two standard human action datasets: the KTH dataset and the Weizmann dataset show that the proposed approach outperforms most existing methods.

This paper presents a benchmarking system for the evaluation of robust mesh watermarking methods. The proposed benchmark has three different components: a ''standard'' mesh model collection, a software tool and two application-oriented evaluation protocols. The software tool integrates both geometric and perceptual measurements of the distortion induced by watermark embedding, and also the implementation of a variety of attacks on watermarked meshes. The two evaluation protocols define the main steps to follow when conducting the evaluation experiments. The efficiency of the benchmark is demonstrated through the evaluation and comparison of two recent robust algorithms.

In this paper we address the problem of generative object categorization in computer vision. We propose a Bayesian model using hierarchical Dirichlet processes mixing AdaBoost learning. Although previous methods trained HDP model for one or two latent themes, our proposed approach uses small-patch-independent-words of appearance-based descriptor and shape information to train a set of intermediate components which are the mixture of visualwords. We then employ AdaBoost weaker learner to find the most related components for classification to handle the variance in intraclass and inter-class information. We show that it performs well for Caltech datasets and with the potential to connect the visual concepts with semantic concepts.

Zernike moments constitute a powerful shape descriptor in terms of robustness and description capability. However the classical way of comparing two Zernike descriptors only takes into account the magnitude of the moments and loses the phase information. The novelty of our approach is to take advantage of the phase information in the comparison process while still preserving the invariance to rotation. This new Zernike comparator provides a more accurate similarity measure together with the optimal rotation angle between the patterns, while keeping the same complexity as the classical approach. This angle information is particularly of interest for many applications, including 3D scene understanding through images. Experiments demonstrate that our comparator outperforms the classical one in terms of similarity measure. In particular the robustness of the retrieval against noise and geometric deformation is greatly improved. Moreover, the rotation angle estimation is also more accurate than state-of-the-art algorithms.

This paper describes a simple and efficient means for statistical detection of global and local mirror symmetry. The symmetry planes are selected from all the planes obtained after matching and pairing features extracted on 3D meshes. The original part of the proposed method is the way to extract significant symmetry planes with a 3D Hough transform built from the planes parameters. This method generates faster results compared to existing statistical methods for symmetry detection.

This paper presents a hierarchical watermarking framework for semiregular meshes. Three blind watermarks are inserted in a semiregular mesh with different purposes: a geometrically robust watermark for copyright protection, a high-capacity watermark for carrying a large amount of auxiliary information, and a fragile watermark for content authentication. The proposed framework is based on wavelet transform of the semiregular mesh. More precisely, the three watermarks are inserted in different appropriate resolution levels obtained by wavelet decomposition of the mesh: the robust watermark is inserted by modifying the norms of the wavelet coefficient vectors associated with the lowest resolution level; the fragile watermark is embedded in the high resolution level obtained just after one wavelet decomposition by modifying the orientations and norms of the wavelet coefficient vectors; the high-capacity watermark is inserted in one or several intermediate levels by considering groups of wavelet coefficient vector norms as watermarking primitives. Experimental results demonstrate the effectiveness of the proposed framework: the robust watermark is able to resist all common geometric attacks even with a relatively strong amplitude; the fragile watermark is robust to content-preserving operations, while being sensitive to other attacks of which it can also provide the precise location; the payload of the high-capacity watermark increases rapidly along with the number of watermarking primitives.

Three-dimensional (3-D) meshes have been used more and more in industrial, medical and entertainment applications during the last decade. Many researchers, from both the academic and the industrial sectors, have become aware of their intellectual property protection and authentication problems arising with their increasing use. This paper gives a comprehensive survey on 3-D mesh watermarking, which is considered an effective solution to the above two emerging problems. Our survey covers an introduction to the relevant state of the art, an attack-centric investigation, and a list of existing problems and potential solutions. First, the particular difficulties encountered while applying watermarking on 3-D meshes are discussed. Then we give a presentation and an analysis of the existing algorithms by distinguishing them between fragile techniques and robust techniques. Since attacks play an important role in the design of 3-D mesh watermarking algorithms, we also provide an attack-centric viewpoint of this state of the art. Finally, some future working directions are pointed out especially on the ways of devising robust and blind algorithms and on some new probably promising watermarking feature spaces.

An original hierarchical watermarking scheme for three-dimensional triangular meshes is proposed in this paper. A geometrically robust watermark and a high-capacity watermark are inserted in different resolution levels of the wavelet decomposition of a semi-regular mesh by modifying the norms of wavelet coefficients. Both watermarks are blind and invariant to similarity transformations. The robustness of the first watermark is achieved by synchronizing and quantizing watermark primitives according to edges lengths of the coarsest level, which are quite insensible to geometrical attacks. The high capacity of the second watermark is obtained by considering the permutation of the norms of a group of wavelet coefficients. Experiments have proven the high robustness of the first watermark under common geometrical attacks. To our knowledge, the capacity of the second method, which can attain the factorial of the candidate coefficients number, is the highest for 3D meshes in the literature.

During the last decades, the three-dimensional objects have begun to compete with traditional multimedia (images, sounds and videos) and have been used by more and more applications. The common model used to represent them is a surfacic mesh due to its intrinsic simplicity and efficacity. In this paper, we present a new algorithm for the segmentation of semi-regular triangle meshes, via multiresolution analysis. Our method uses several measures which reflect the roughness of the surface for all meshes resulting from the decomposition of the initial model into different fine-to-coarse multiresolution meshes. The geometric data decomposition is based on the lifting scheme. Using that formulation, we have compared various interpolant prediction operators, associated or not with an update step. For each resolution level, the resulting approximation mesh is then partitioned into classes having almost constant roughness thanks to a clustering algorithm. Resulting classes gather regions having the same visual appearance in term of roughness. The last step consists in decomposing the mesh into connex groups of triangles using region growing ang merging algorithms. These connex surface patches are of particular interest for adaptive mesh compression, visualisation, indexation or watermarking.

Abstract In this paper we present a new framework for subdivision surface approximation of three‐dimensional models represented by polygonal meshes. Our approach, particularly suited for mechanical or Computer Aided Design (CAD) parts, produces a mixed quadrangle‐triangle control mesh, optimized in terms of face and vertex numbers while remaining independent of the connectivity of the input mesh. Our algorithm begins with a decomposition of the object into surface patches. The main idea is to approximate the region boundaries first and then the interior data. Thus, for each patch, a first step approximates the boundaries with subdivision curves (associated with control polygons) and creates an initial subdivision surface by linking the boundary control points with respect to the lines of curvature of the target surface. Then, a second step optimizes the initial subdivision surface by iteratively moving control points and enriching regions according to the error distribution. The final control mesh defining the whole model is then created assembling every local subdivision control meshes. This control polyhedron is much more compact than the original mesh and visually represents the same shape after several subdivision steps, hence it is particularly suitable for compression and visualization tasks. Experiments conducted on several mechanical models have proven the coherency and the efficiency of our algorithm, compared with existing methods.

This work presents a new and efficient algorithm for the decomposition of 3D arbitrary triangle meshes into surface patches. The algorithm is based on the curvature tensor field analysis and presents two distinct complementary steps: a region based segmentation, which is an improvement of that presented by [G. Lavoue et al., (2004)] and which decomposes the object into known and near constant curvature patches, and a boundary rectification based on curvature tensor directions, which corrects boundaries by suppressing their artifacts or discontinuities. Experiments were conducted on various models including both CAD and natural objects, results are satisfactory. Resulting segmented patches, by virtue of their properties (known curvature, clean boundaries) are particularly adapted to computer graphics tasks like parametric or subdivision surface fitting in an adaptive compression objective.

Content based shape image retrieval is an important problem which gained the attention of the community. The challenge is to map the shape into compact and robust descriptor. This study presents a generalization of the angular radial transform (ART). The ART, recommended by the MPEG-7 standard, is only limited to binary images and is not robust to perspective deformations. We propose two generalizations of the ART allowing to apply it to color images and to make it robust to all possible rotations and to perspective deformations.

Because of the semantic gap between low-level descriptors and semantic concepts, image understanding systems need to use domain knowledge in order to map regions of a segmented image into semantic objects. However, such systems are often strongly dependent on an application since domain knowledge is not separated from procedures. We therefore argue for a generic grouping mechanism, based on the vision procedures of Gestalt theory, able to evaluate the relevance of its groupings with domain knowledge. In order to be used in a flexible way, such domain knowledge is stored independently in ontologies. We emphasize the need for domain knowledge to be thought as scene knowledge on a multilevel of composition, in order to be efficiently used. We also present the whole framework of our system and an application example within a specific domain: image understanding of thessalian graves' images.

3D discretized objects become more and more present in many applications for computer graphics. The need to store and transmit these data grows permanently. This paper introduces a new scheme of 3D object progressive transmission. It is based on an union of cone sections and balls selected from the medial axis and optimized for a better scalability in terms of bit rates and quality. This representation of a 3D object is a lossless approach and is very efficient for compression and progressive transmission, as shown in the results presented in this paper.

This study addresses the problem of the objective performance evaluation of image retrieval systems. Considering the color feature, a tool for synthetic image databases generation is proposed. This allows to control the number and location of the dominant colors in the Lab space and provides also their spatial coherency. It is then possible to sort exactly the images for the color feature. We also propose a new similarity measure to compare images described by the color feature. Based on the assumption of a Gaussian distribution for each dominant color, each image is modeled by the sum of these Gaussian distributions. The similarity measure performs a Kullman's distance between two modeled distributions. The objective performance evaluation based on the synthetic database is done for comparing our image retrieval system (IMALBUM) which uses the new similarity measure and MPEG7 approach. Experiments on MPEG7 database are also presented as a subjective evaluation and discussed.

A method for the design and reconstruction of rough surfaces is introduced. A fractal model based on projected IFS attractors allows the definition of free form fractal shapes controlled with a set of points. Ibis flexible model has good fitting properties for recovering surfaces. The approximation is formulated as a non-linear fitting problem and resolved using a modified LEVENBERG-MARQUARDT minimisation method. The main applications are shape design, shape reconstruction and geometric data compression.

We present a geometric coding method for 3D real and virtual objects transmission and visualization. We describe object surfaces with a set of superellipsoids which are structured with a constructive solid geometry (CSG) tree. An unstructured cloud of 3D points lying on the original object surface is considered as initial data. The tree is obtained using a split and merge algorithm. At each splitting step, the set of points is partitioned into two superellipsoids. The parameters of each ellipsoid is computed with a nonlinear regression technique (Levenberg-Marquardt). This decomposition is stopped when the estimation error reaches a given threshold (the accepted distortion). The objective of the merging algorithm is to reduce the number of superellipsoids used to describe the complex 3D object without increasing the global distortion. This algorithm allows one to merge both connected and non-connected elements of the superellipsoids set obtained from the decomposition step. The whole scheme is applied to synthetic objects.

This paper addresses the problem of approximation of natural complex shapes. Using MPEG-7 terminology, this problem can be considered as the search of a descriptor for a shape feature. This shape can be defined either as a frontier between image regions or a natural curve. For this purpose, an original descriptor which combines iterated function system (IFS) model and the notion of free form curves is proposed. A set of control points allows one to define the IFS model in a barycentric space. This generalization adds a real flexibility to fractal approximation techniques enriching the set of contractive operators which are candidate to model the self-similarity. This new descriptor named projected IFS model allows the reconstruction of a shape using a projection via the control points. It is adapted to the representation of both smooth shapes (man-made objects, body,..) and fractal shapes (mountain, cloud, tree,..). Results on synthetic shapes and a real mountain shape are presented.

This paper presents a new binary shape sampling and coding method. Within the framework of the MPEG-4 standard, the binary shapes in video object plans have to be coded with a near to lossless approach in order to obtain a perfect spatial and temporal localization. For this purpose, a new sampling method based on the object skeleton is developed. The "r-sampling" is defined function of the local feature size (LFS) which is the distance between a shape point and the closest point on the medial axis of this shape. A weighting convex function /spl phi/ depending on the local curvature is also introduced to increase the local adaptiveness of the sampling ("the r&c-sampling"). This new approach is tested on contours presenting high level details. The first results obtained are satisfactory in terms of both the reduction of samples and the reconstruction error.

We present a new gradient model for colour video images. Video sources have the characteristic, either for transmission or for storage, to present a reduced bandwidth of colour components compared to that of brightness. Traditional multispectral gradient operators amplify the noise from the colour components and they are not very sensitive to colour edges. We propose an adapted vector gradient model to solve this problem and introduce the computation of the partial derivatives at different scales according to the resolution of each component. We show that a weighting function is necessary to control the relative influence of the colour and brightness components to obtain the multiscale colour gradient (MCG). The application of the MCG to colour edge detection illustrates the advantages of the proposed method. Finally, the MCG image is used for the segmentation by active contours of crystals in microscopic images of cement clinker (industrial application).

In this paper, we consider the texture coding of quadrangular cells which compose active meshes. Active mesh algorithms which have been developed for region based video coding mainly use interpolation techniques to quantize the cell texture. We propose a new quantization scheme which associates image warping and an adaptive JPEG scheme. Our approach leads to an interesting trade-off between mesh complexity and image quality.

In a previous work we have presented an adaptive segmentation algorithm of dirty images in a Bayesian framework. The segmentation problem is formulated as a maximum a posteriori (MAP) estimation problem. The optimization is achieved using Besag's iterated conditional modes algorithm. A multiresolution implementation of the segmentation algorithm, using the discrete wavelet transform, has been used. This work focuses on the adaptive character of the algorithm and discusses how global and local statistics can be taken into account in the segmentation process. We propose an improvement on the adaptivity by introducing an enhancement to control the adaptive properties of the segmentation process. A weighting function taking into account both local and global statistics is introduced in the minimization. The new formulation of the segmentation problem allows us to control the effective contribution of each statistic. Results of segmentation carried out on synthetic images are presented.

Two enhanced subband coding schemes using a regularized image restoration technique are proposed: the first controls the global regularity of the decompressed image; the second extends the first approach at each decomposition level. The quantization scheme incorporates scalar quantization (SQ) and pyramidal lattice vector quantization (VQ) with both optimal bit and quantizer allocation. Experimental results show that both the block effect due to VQ and the quantization noise are significantly reduced.

We consider the coding of a quadrangular mesh structure generated by a multiresolution deformable quadtree algorithm. Such a structure has been found efficient for region based video coding, but few works deal with the coding of the mesh structure. We propose to use the rules of the mesh generation to significantly reduce the number of bits required for the coding of the nodes position defining the mesh structure. This reduction induces an improvement of either the image quality or either the global bit rate.

The influence of the phase of conjugate quadrature filters (CQFs) on the performances of a subband coding scheme is analyzed. When the filter length is short, the phase characteristic has virtually no influence on the peak signal-to-noise ratio (PSNR)/bit rate results and on the performance of postprocessing algorithms such as edge detection.

We propose a new image compression scheme based on fractal coding of wavelet transform coefficients using a fast non-iterative algorithm for the codebook generation. The original image is first decomposed into subbands containing information in different spatial directions and different scales, using an orthogonal wavelet filter bank. Subbands are encoded using local iterated function systems (LIFS) with range and domain blocks presenting horizontal or vertical directionalities. Their sizes are estimated according to the correlation lengths and resolution of each subband. The computational complexity is greatly decreased by using subband decomposition. In addition a fast non-iterative algorithm is implemented for block classification. This algorithm progressively creates the codebook during only one scanning of the training set. We prove the efficiency of the proposed approach both in terms of PSNR/bit rate and computation time.

In this study, we consider the DCT based subband decomposition and show the efficiency of this approach for low contrast images in comparison to 9-7 bi-orthogonal FIR filters generally used in such a scheme. First, we consider the theoretical coding gain for the two approaches. Afterwards, the distortion versus rate performances are analyzed and the perceptual qualities of the reconstructed images are discussed. Finally, an edge detection operator is applied on the reconstructed images in order to show the robustness of DCT based subband decomposition to post-processing.

An optimized dequantization array is proposed for the Joint Photographers Expert Group (JPEG) scheme. The reconstruction error is minimized subjected to a regularity constraint. Both the peak-to-peak signal-to-noise ratio (PPSNR) and the visual quality are improved. In addition, the method reduces discontinuities in postprocessing such as edge detection. This approach is in full compliance with the JPEG standard.

A transform image coding method is presented. A discrete cosine transform (DCT) is performed on blocks of 16*16 size. A threshold mask is applied by using a fixed threshold determined by the statistical properties of the DCT coefficients. Selected coefficients are coded with a Huffman code. Three methods for coding the threshold mask are compared. zigzag, orthogonal, and quadtree scan. Three Huffman code tables are presented for coding the runlengths of zeros. A high-quality decoded image is shown at 1 bit/pixel.>