Search

Microfossils, tiny fossils whose study requires the use of a microscope, have been widely applied in many fields of earth, life, and environmental sciences. The abundance and high diversity of microfossils, as well as the need for rapid identification, call for automated methods to classify microfossils. In this study, we constructed an open dataset of three-dimensional (3D) microfossils and proposed a deep learning-based approach for microfossil classification. The dataset, named `Archives of Digital Morphology' (ADMorph), currently contains more than ten thousand 3D models from five classes of 410 million-year-old fishes. The deep learning-based method includes data preprocessing, feature extraction, and 3D microfossil model classification. To assess the method performance and dataset representability, we performed extensive experiments. Compared with multiview convolutional neural networks (MVCNN) (91.54%), PointNet (64.13%), and VoxNet (78.15%), the method proposed herein had higher accuracy (97.60%) on the experimental dataset. We also verified data preprocessing (92.36%) and feature extraction (97.10%). We combined them to obtain the macroaveraging accuracy of 97.60%, the highest accuracy of 100%, and the lowest accuracy of 88.78%. We suggest that the proposed method can be applied to other 3D fossils and biomorphological research fields. The fast-accumulating 3D fossil models might become a source of information-rich datasets for deep learning.

We propose a generalized 3D shape descriptor for the efficient classification of 3D archaeological artifacts. Our descriptor is based on a multi-view approach of curvature features, consisting of the following steps: pose normalization of 3D models, local curvature descriptor calculation, construction of 3D shape descriptor using the multi-view approach and curvature maps, and dimensionality reduction by random projections. We generate two descriptors from two different paradigms: 1) handcrafted, wherein the descriptor is manually designed for object feature extraction, and directly passed on to the classifier and 2) machine learnt, in which the descriptor automatically learns the object features through a pretrained deep neural network model (VGG-16) for transfer learning and passed on to the classifier. These descriptors are applied to two different archaeological datasets: 1) non-public Mexican dataset, represented by a collection of 963 3D archaeological objects from the Templo Mayor Museum in México City that includes anthropomorphic sculptures, figurines, masks, ceramic vessels, and musical instruments; and 2) 3D pottery content-based retrieval benchmark dataset, consisting of 411 objects. Once the multi-view descriptors are obtained, we evaluate their effectiveness by using the following object classification schemes: $K$ -nearest neighbor, support vector machine, and structured support vector machine. Our object descriptors classification results are compared against five popular 3D descriptors in the literature, namely, rotation invariant spherical harmonic, histogram of spherical orientations, signature of histograms of orientations, symmetry descriptor, and reflective symmetry descriptor. Experimentally, we were able to verify that our machine learnt and handcrafted descriptors offer the best classification accuracy (20% better on average than comparative descriptors), independently of the classification methods. Our proposed descriptors are able to capture sufficient information to discern among different classes, concluding that it adequately characterizes the datasets.

In recent years, the growth of the Internet along with multimedia data delivery, have caused the misuse and even alteration of the original digital content. As a security mechanism, digital watermarking has been used to protect digital content, traditionally in copyright protection applications, and more recently digital watermarking has been used in content enhancement applications. Regardless of the application, digital watermarking schemes always cause some level of signal degradation. There are however, certain applications as in the military or medical fields, where loss of information cannot be tolerated during the reconstruction of host signals, therefore conventional watermarking schemes do not satisfy the quality requirements. In order to reconstruct signals without loss of data, reversible watermarking is employed. Recent research in reversible watermarking has focused on either, the improvement of embedding capacity or the reduction of perceptual impact of the schemes, but scarcely exploring the robustness of reversible schemes. In this survey, we provide a deep review of reversible watermarking schemes and propose a new classification outline based on the robustness point of view. Finally, some future directions for reversible watermarking research are presented and open problems in robust reversible watermarking are outlined.

We present the design and early-stage implementation of Uasisi, a system to assist the visually impaired in navigating physical spaces. In its simplest form, a single wearable device can assist the user by sensing the proximity of objects in the environment and providing vibratory feedback. More devices can be added in a modular way to form a wireless body area network, and this can, in turn, be connected to a cloud infrastructure where data related to the navigational patterns of the user can be analyzed and help in making further predictions. Our system can also adapt to form part of an augmented environment where smart objects communicate and cooperate with each other to provide a more rich and complex navigation experience for the user. We show a series of experiments where various configurations of the system are tested with users who can not see and must navigate spaces previously unknown to them.

Mutualcast is a one-to-many (peer-to-peer) scheme for content distribution that maximizes the overall throughput during a broadacast session. It is based on a fully-connected graph (full mesh topology), which introduces benefits such as robustness or simultaneous transmission from/to multiple devices. The main disadvantage of Mutualcast is scalability; it is constraint to a small P2P group for content distribution. In this paper, we make Mutualcast scalable. We propose a highly collaborative and scalable P2P tree-based architecture made of two main components: 1) Peer grouping or clustering and 2) Hierarchical tree-based content distribution. In step 1), peer nodes (content receivers) are grouped into equal-size clusters by using a proposed heuristic size-constrained algorithm based on k-means. In step 2), clusters (which become the nodes of the tree) are organized into a single hierarchical n-ary tree-based architecture, in which the root of the tree (Root Cluster) is the one closest to source peer, while intermediate and leaf clusters are positioned in the tree according to their delay-proximity to previously inserted clusters. During content distribution, the root cluster receives the blocks of content before any other cluster in the tree and directly from (and only from) the source peer; blocks are then passed on to the next hierarchical level down the tree in order (higher levels of the tree receive the content before lower levels). Inter-clusters and intra-clusters content distribution is performed concurrently and takes into account peers upload/download capacities to relay blocks of content. The evaluation of our hierarchical P2P architecture concentrates on the following metrics: scalability of the systems, overall end-to-end delay distribution, and efficient cluster size. Finally, our architecture is compared against two well-known P2P technologies in the literature, Super-Peer and Kademlia.

In this paper is described the implementation of the frequency-dependent line model (FD-Line) in a real-time digital power system simulator. The main goal with such development is to describe a general procedure to incorporate new realistic models of power system components in modern real-time simulators based on the Electromagnetic Transients Program (EMTP). In this procedure are described, firstly, the steps to obtain the time domain solution of the differential equations that models the electromagnetic behavior in multi-phase transmission lines with frequency dependent parameters. After, the algorithmic solution of the FD-Line model is implemented in Simulink environment, through an S-function programmed in C language, for running off-line simulations of electromagnetic transients. This implementation allows the free assembling of the FD Line model with any element of the Power System Blockset library and also, it can be used to build any network topology. The main advantage of having a power network built in Simulink is that can be executed in real-time by means of the commercial eMEGAsim simulator. Finally, several simulation cases are presented to validate the accuracy and the real-time performance of the FD-Line model. En este artículo se describe la implementación del modelo de línea dependiente de la frecuencia FD-Line en un simulador digital en tiempo-real de sistemas eléctricos de potencia. El objetivo principal con este desarrollo es describir un procedimiento general para incorporar nuevos modelos realistas de componentes de sistemas eléctricos de potencia en los modernos simuladores de tiempo real basados en el programa de transitorios electromagnéticos (o EMTP, por sus siglas en inglés). En dicho procedimiento se describen, primeramente, los pasos para obtener la solución en el dominio del tiempo de las ecuaciones diferenciales que modelan el comportamiento electromagnético de una línea de transmisión polifásica con parámetros dependientes de la frecuencia. Posteriormente, se implementa la solución algorítmica del modelo de línea en Simulink, por medio de una función S en lenguaje C, para ejecutar simulaciones fuera de línea de transitorios electromagnéticos. Esta implementación permite ensamblar libremente el modelo FD-Line con cualquier elemento de la librería de sistemas de potencia en Simulink y también, se puede construir cualquier topología de red eléctrica. La principal ventaja de tener una red eléctrica construida en Simulink es que puede ser ejecutada en tiempo real por medio del simulador comercial eMEGAsim. Finalmente, se presentan varios casos de simulación de sistemas de potencia para probar la precisión y el desempeño en tiempo-real del modelo FD-Line.

This work describes SenseCityVity, an approach to engage and support youth in a city in Mexico as they investigate, document, and reflect upon urban problems through mobile crowdsourcing. SenseCityVity focused on the development of a mobile crowdsourcing platform; the deployment of the Urban Data Challenge, codesigned by the authors' research team and actors to collect geolocalized images, audio, and video; and the analysis, appropriation, and creative use of the collected data for community reflection and artistic creation. The approach integrates mobile technology and community practices involving a large population of young people for urban engagement. The collective action generated a new multimedia dataset that is rich in terms of content and is enabling a number of studies aimed at better understanding the urban landscape of cities in the Global South. This article is part of a special issue on smart cities.

Resumen. De manera natural las personas vamos formando rutinas para adaptarnos al entorno social en el cual nos desempenamos. Estas rutinas se ven reflejadas en las secuencias de ubicaciones que visitamos para desplazarnos hacia los lugares donde realizamos nuestras actividades diarias, sin embargo, ?como podemos determinar la presencia de ellas? El presente trabajo consiste en un analisis exploratorio de un conjunto de datos de movilidad de usuarios utilizando modelos de topicos con el fin de determinar si existe una estructura subyacente en la movilidad de las personas de acuerdo a las rutinas que estas manifiestan en sus vidas. Se presenta la introduccion al problema explicando el modelo a utilizar aśi como su desarrollo teorico, validacion experimental de su funcionamiento a traves de conjuntos de datos sinteticos y su aplicacion a un conjunto de datos real con presentacion de los resultados obtenidos y conclusiones.

Network coding is a proposed technique for improving network capacity. This novel concept has been mainly oriented to increase throughput and reliability of communication networks. Network coding is implemented in the intermediate nodes before forwarding the encoded packets to the end nodes. The received packets are decoded in the end nodes in order to recovery the original transmitted data (video data in our case). In this work, we investigate the performance of network coding in collaborative multi-source scenarios with heterogeneous resources (video, image, audio, pdf files). Collaborative multi-source schemes are very important for critical multimedia services because multimedia content consumes an important amount of resources in the communication networks. A multi-source scheme is a useful solution when different parts of a multimedia content is generated or stored in two or more sites. Our evaluation compares the performance of a P2P networking with network coding against a client-server communications. Results show great benefits on using network coding scheme in collaborative multi-source scheme.

Recently, the United Nations Human Settlements Program (UN-HABITAT) stated that cities in Mexico are steadily expanding, sometimes outstripping the rise in population by two-fold due to urban sprawl, and that 85% of the Mexican population will live in towns and cities by 2050. Mexico also has booming young populations, who are not only the inheritors of significant social and urban problems, but also the most promising source of solutions. UrBis is a technological platform that was developed to help people in Mexico, particularly youth, document, characterize and reflect about the social and urban problems they face. The platform supports an integrative approach that combines mobile crowdsourcing, social technologies and community practices to effectively document, characterize, reflect about socio-urban issues, and to develop possible solutions through the combined effort of citizens. The purpose of this paper is to share our action-research experience with UrBis in engaging youth populations in the field, outlining our findings after analyzing collected data in an effort to determine how our technological-based approach supports the study of urban environments in five cities located across the country.

A new model-based scheme for clock offset estimationbetween two nodes is presented.The one-way delay is modeled with a gamma distribution representing the actual state of the network link.Real and modeled (with zero shift) one-way delay measurements are QQ-plot for the estimation of the clock offset.The proposed model is non-intrusive (the kernel is not modified), and can be implemented at the application layer.The model provides accuracy in the order of 3-5 milliseconds under different traffic conditions. Having in mind multimedia systems applications, we propose a novel model-based approach to estimate the clock-offset between two nodes on the Internet. Different than current clock-offset schemes in the literature, which are iterative in nature, our scheme is aimed at getting a good non-iterative clock-offset estimation in real time (in the order of milliseconds). In our clock-offset estimation approach, the One-Way Delay (OWD) measurements are modeled with a shifted gamma distribution representing the current state of the probing link. By using the QQ-probability plot technique and linear regression model, we estimate the (shift parameter or) minimum value of the gamma distribution with probability zero. This estimated value represents the clock offset plus network propagation and transmission delay (queuing delay has already been eliminated) for the corresponding receiving path. End nodes exchange their corresponding minimum estimates and get an improved final clock offset estimate considering the network path asymmetries. Based on real experiments, we show that our scheme provides an extremely fast clock-offset estimation with lower RMSE and superior stability than NTP and current NTP-like state of the art methodologies in the literature Jeske and Sampath (2003), Choi and Yoo (2005), Adhikari et al. (2003), Tsuru et al. (2002). Moreover, our proposed scheme is non-intrusive (no kernel programming needed), easy to implement, and targeted as part of more complex real-time multimedia distribution protocols requiring a fast and reliable OWD estimates.

Due to pervasive communication infrastructures, a plethora of enabling technologies is being developed over mobile and wired networks. Among these, video streaming services over IP are the most challenging in terms of quality, real-time requirements and security. In this paper, we propose a novel scheme to efficiently secure variable length coded (VLC) multimedia bit streams, such as H.264. It is based on code word error diffusion and variable size segment shuffling. The codeword diffusion and the shuffling mechanisms are based on random operations from a secure and computationally efficient chaos-based pseudo-random number generator. The proposed scheme is ubiquitous to the end users and can be deployed at any node in the network. It provides different levels of security, with encrypted data volume fluctuating between 5.5–17%. It works on the compressed bit stream without requiring any decoding. It provides excellent encryption speeds on different platforms, including mobile devices. It is 200% faster and 150% more power efficient when compared with AES software-based full encryption schemes. Regarding security, the scheme is robust to well-known attacks in the literature, such as brute force and known/chosen plain text attacks.

The aim of this paper is to describe a general procedure for incorporating new realistic models of power system components in modern EMTP-based real-time simulators. For doing this, a comprehensive description of the frequency-dependent transmission line model FD-Line is first presented. Then, it is described the implementation of this line model in Simulink environment for running off-line simulations of electromagnetic transients. This implementation allows the interaction with the Power System Blockset library and also, the real-time simulations in the PC-based real-time simulator eMEGAsim. Finally, several simulation cases of power systems are presented to test the real-time performance of this line model.

Digital generators of chaos present several limitations that affect the vulnerability of chaotic encryption systems, among the most important are degraded probability distribution and short cycle lengths. Periodic perturbations of the chaotic parameter and/or state variable have been employed to deal with these limitations, although blindfold; the periodicity of the perturbation is set up during the initialization process without reference to the cycle length of the chaotic map under consideration. For best results, the periodicity of the perturbation must be close to the actual cycle length. So far, it is analytically impossible and numerically impractical (for real-time applications) to have a priori information of the cycle length. In this work we propose an on-the-fly detection and quantification of the chaotic cycle length to eliminate short cycles (which make cryptosystems vulnerable to attacks) and maximize the strength of long cycles by perturbing the system at the right time. Our proposal consists of two algorithms: (1) Unrestricted Search Algorithm (USA), which tracks down the cycle without any assumption or restriction on the digital chaotic trajectory, and (2) Ergodic Search Algorithm (ESA), which assumes ergodic trajectories to reduce the cycle search space, without this being a necessary requirement for the analyzed trajectory. USA and ESA are intended to increase the security of chaotic encryption systems without compromising their performance. Furthermore, they can be employed for the development of new chaotic-map independent encryption systems, where a full knowledge of the map is not required.

Data interleaving schemes have proven to be an important mechanism in reducing the impact of correlated network errors on image/video transmission. Current interleaving schemes fall into two main categories: (a) schemes that interleave pixel intensity values and (b) schemes that interleave JPEG/MPEG transform blocks. The schemes in the first category suffer in terms of lower compression ratio since highly correlated information in the spatial domain is de-correlated prior to compression. The schemes in the second category interleave DCT transformed blocks. In this case, in the absence of ARQ, if a packet is lost, an entire block may be lost thus yielding poor image quality and making the error concealment task difficult. Interleaving transform coefficients is tricky and error concealment in the presence of lost coefficients is challenging. In this paper, we develop three different interleaving schemes, namely Triangular, Quadrant, and Coefficient, that interleave frequency domain transform coefficients. The transform coefficients within each block are divided into small groups and groups are interleaved with the groups from other blocks in the image, hence they are referred to as inter-block interleaving schemes. The proposed schemes differ in terms of group size. In the Triangular interleaving scheme AC coefficients in each block are divided into two triangles and interleaving is performed among triangles from different blocks. In the Quadrant interleaving scheme, coefficients in each block are divided into four quadrants and quadrants are interleaved. In the Coefficient interleaving scheme, each coefficient in a block is a group and it is interleaved with the coefficients in other blocks. The compression ratio 3 of the proposed interleaving schemes is impressive ranging from 90 to 98% of the JPEG standard compression while providing much higher robustness in the presence of correlated losses. We also propose two new variable end-of-block (VEOB) techniques, one based on the number of AC coefficients per block (VAC-EOB) and the other based on the number of bits per block (VB–EOB). Our proposed interleaving techniques combined with VEOB schemes yield significantly better compression ratios compared to JPEG (2–11%) and MPEG-2 (3–6.7%) standards while at the same time improve the resilience of the coded data in the presence of transmission errors.

In this paper we develop a new approach for reconstructing 3-D scenes from RGB-D data. We use a Markov random field to model appearance relations and geometric cues between different regions of a scene, as a means to provide robustness to noisy and incomplete data often generated by RGB-D devices. A parametric reconstruction of 3-D scenes that enable coherent physical interaction are computed, in near real time, with a standard computer that does not use specialized hardware.

The aim of this work is to provide a fast approach for monocular SLAM initialization by constructing an initial 3-D map with interest points that are susceptible to be automatically tracked. Interest points' depth is inferred by means of a linear regression model, which estimates depth on the basis of local image appearance. Our contributions are: (1) a new scheme for learning and predicting associations between depth and local image appearance using RGB-D data; and (2) the use of this scheme for the initialization of state-of-the-art visual SLAM systems from a single image frame. To the best of our knowledge, this is the first attempt to automatically initialize a SLAM system by associating depth to sensor features through machine learning techniques. We performed a series of tests by making use of the celebrated PTAM system and obtained very promising results. We show successful one-shot initialization examples accomplished by applying our proposed approach to unstructured scene environments.

We present a new LUT design consisting of a non-iterative plaintext transformation and a high dimensionally (K-map) populated dynamic Look-Up Table with random access that outperforms the security and speed of previous LUT based schemes. Experimental analysis of the proposed scheme reveals excellent statistical properties, sensitivity to plaintext/system-key changes, robustness to differential and chosen plaintext attack, and high performance for real-time multimedia communications.

Protection of private user information in computers and communication networks has been one of the major concerns during the last decade. It has become even more critical due to pervasive use of smart mobile devices, and is exacerbated due to their limited processing and battery power needed to manage complex encryption schemes, particularly for real-time multimedia applications (audio and video). Secure multimedia communication systems require processing of huge amounts of information at speeds ranging from Kilobits/sec (Kbs) to the order of Megabits/sec (Mbs). Provisioning of security for such large volumes of data in mobile devices may be simply infeasible when the complexity of related operations is beyond the processing limit of such devices. In this chapter we evaluate the performance of different encryption schemes, including AES implementations and non-conventional chaotic encryption on different architectures. Our experiments reveal that chaos-based schemes outperform the conventional AES implementation in terms of CPU usage, encryption speed, and energy consumption. Particularly they consume 300-400% less CPU power, and have over 250% faster encryption speed. However, the performance also depends on the floating point capability of the platform; a suitable scheme may be chosen depending on the CPU power of platform. The performance results reported in this chapter are based on experiments on contemporary desktops, laptops, netbooks, and cell phones (Nokia N800 and N900).

This paper presents a novel scheme to efficiently secure variable length coded (VLC) multimedia bit streams, such as H.264. The proposed scheme employs code word diffusion and content based shuffling techniques to achieve security. Specifically, it is a combination of a highly secure random number generator based on chaotic maps and a low computation complexity block shuffling procedures. The scheme is ubiquitous to the end users and can be deployed at any node in the network, without requiring any decoding. Its power efficiency and scalability is demonstrated on mobile as well as fixed platforms.

One of the main problems in Face Recognition systems is the recognition of an input face with a different expression than the available in the training database. In this work, we propose a new 3D‐face expression synthesis approach for expression independent face recognition systems (FRS). Different than current schemes in the literature, all the steps involved in our approach (face denoising, registration, and expression synthesis) are performed in the 3D domain. Our final goal is to increase the flexibility of 3D‐FRS by allowing them to artificially generate multiple face expressions from a neutral expression face. A generic 3D‐range image is modeled by the Finite Element Method with three simplified layers representing the skin, fatty tissue and the cranium. The face muscular anatomy is superimposed to the 3D model for the synthesis of expressions. Our approach can be divided into three main steps: Denoising Algorithm, which is applied to remove long peaks present in the original 3Dface samples; Automatic Control Points Detection, to detect particular facial landmarks such as eye and mouth corners, nose tip, etc., helpful in the recognition process; Face Registration of a 3D‐face model with each sample face with neutral expression in the training database in order to augment its training set (with 18 predefined expressions). Additional expressions can be learned from input faces or an unknown expression can be transformed to the closest known expression. Our results show that the 3D‐face model resembles perfectly the neutral expression faces in the training database while providing a natural change of expression. Moreover, the inclusion of our expression synthesis approach in a simple 3D‐FRS based on Fisherfaces increased significantly the recognition rate without requiring complex 3D‐face recognition chemes.

We propose a new collaborative clock offset estimation scheme between two nodes in the Internet using independent one-way offset estimations. Our proposal (different than current schemes in the literature) is intended to provide a fast and accurate clock offset estimation in approximately [Round-Trip Time (RTT)+40]ms. The scheme sends a group of 5 probes in the the forward and reverse paths, and models the One-way Transit Time (OTT) by a Gamma distribution (with parameters adapted to actual path condition) to estimate the minimum distribution value (or long-term minimum OTT value). End nodes exchange their corresponding minimum distribution values to get an improved final clock offset estimate, which takes into account the network path asymmetries. We show that our scheme provides a faster clock offset estimation with lower RMSE and superior stability than NTP and current NTP-like state of the art methodologies in the literature.

We propose a novel block-based symmetric encryption system based on an n-array of independently iterated chaotic logistic maps with global and local feedback as a diffusion process. Local feedback represents the temporal evolution of a single map, while global feedback represents the temporal evolution of the whole system (cross-map evolution). For security, the cryptosystem periodically modifies its internal configuration using a three-level random perturbation scheme, one at system-key (reset operation) and two at map array level (to increase the chaotic cycle length of the system). An analysis of the proposed scheme regarding its vulnerability to attacks, statistical properties and implementation performance is presented. To the best of our knowledge we provide a simple and secure scheme with the fastest software implementation reported in the literature.

We developed an IP mapping tool called VisRutas for discovering connection and geographic location of routers (hosts) in the public Internet. Unlike previous schemes that use large data bases and intensive route probe systems (as the Rocketfuel method) or less reliable IP-to-location mapping list (GeoTrack and GeoCluster methods), our approach is simple, fast and effective. It is based on small database tables, scalable traceroute and tracert based route information, BorderGateway protocol (BGP) tables and a reliable IP to location mapping list. VisRutas tool was applied to estimate the router-level topology (connection) map of the public Internet in Mexico, and the complete analysis took minutes to hours, whereas previous schemes take days to weeks. It is fair to mention that the other schemes provide better detail mapping inside the Internet service provider's domain, (which is not our main objective). Another advantage of our approach (VisRutas) is that it can be successfully applied anywhere in the world (as long as we input the correct end-nodes probing list), as well as to make early path diversity discovery to improve media delivery architecture design.

This paper presents a geometric error concealment scheme for DCT-based image/video data based on the Bezier polynomials (BP). The proposed scheme makes use of the average edge direction and local curvature (extracted from healthy blocks around the damaged block) as boundary conditions to create an n-degree BP. The resulting BP is then used to interpolate the edges of the lost block, as well as to directionally reconstruct its low frequency data. When multiple edges are identified crossing the lost area, a cost function defined in terms of the local and global edge curvatures is used to find the best match for each missing edge. Experimental results show that our approach provides near perfect reconstruction and excellent subjective quality of the restored data, outperforming current linear interpolation schemes in the literature

We developed an ad-hoc spatial error concealment (EC) scheme for two different inter-block interleaving techniques known as triangular and coefficient level interleaving (R. Hasimoto-Beltran et al., December 2004). Triangular EC scheme works in the transform domain and makes use of AC information in the neighboring blocks in order to estimate the missing AC coefficients in the partially damaged block. The EC for the coefficient level interleaving is applied in the space domain and the main purpose is to eliminate the blocking artifacts due to the loss of few AC coefficients. We show that Intra-block interleaving along with our developed EC schemes increase significantly the quality of the received image/video information when transmitted over bursty channels with high error rate (50%)

Signal-to-noise ratio (SNR) scalable coding over priority networks has been used as a robust mechanism to transmission errors, since the most important information (base layer) can be sent over an error free channel as in the case of ATM networks or over protected channel as in the case of IP networks. SNR scalability ensures a minimum video quality at the receiver end by sacrificing the least important information (enhancement layer) when losses occur. However, one disadvantage of scalable coding is its lower compression performance with respect to the one-layer encoding (/spl sim/10-15%, (D. Wilson and M. Ghanbari, 1999)). In this work, we propose an error resilient scheme for MPEG-2 SNR scalable video transmission that is compression efficient, in addition to providing scalable error resynchronization mechanism to enhancement layer that impedes the propagation of errors beyond the physically affected area in the bitstream. In order to provide robustness to errors during transmission, the proposed scheme uses triangular coding in the transform domain (R. Hasimoto-Beltran et al., 2000) and a modified variable end of block based on the number of bits per unit of block. Our scheme yields significantly better compression ratios compared to the MPEG-2 standard (3% to 6.7%) while at the same time improves considerably the quality of the received data (2-11 db) under different conditions of data losses.

Interleaving schemes have proven to be an important mechanism in reducing the effect of network errors on image transmission. Current interleaving schemes spatially decorrelate neighboring image blocks by putting them into packets that are a long way apart from each other in the transmission sequence. Most of the existing schemes, while achieving good performance on random packet losses, do not work well in the case of bursty packet losses. In this paper, we propose two interleaving schemes where the decorrelation process is applied not only at the block level but also at the pixel or coefficient level (i.e. on the information within the blocks) in the compressed domain. The decorrelation is achieved via a k-way shuffle of pixels among the blocks and by using the spatial properties of different space-filling curves, where k depends on the total number of blocks in the image. Our results show that in spite of the decorrelation of coefficients in the triangular interleaving scheme, the compression ratio is within 92% to -98% of JPEG standard compression. We also show that, when compared with existing interleaving schemes, our techniques provide improved image quality and a low mean square error in the transmitted images in the presence of random as well as bursty packet losses in networks.

We address the problem of transmitting JPEG compressed image over a lossy network. We introduce a combination of block level triangular interleaving scheme (TRII), and variable and selective forward error correction scheme (VS-FEC) to make the transmitted data more resilient while using the network bandwidth more efficiently in the presence of bursty losses. Our results show that TRII barely affects the final compression ratio of the transmitted image (on the average just 4% below JPEG), while considerably increases its resilience to errors. Also, in the presence of losses, a new scheme is introduced to dynamically add a variable FEC to avoid retransmission of the lost data. A small amount of FEC overhead was needed (4%, versus 10%-15% in constant FEC) to be at least 90% sure on the average that the protected information will not be damaged during transmission.

In this work, we present a new chaos-based cryptosystem scheme consisting of a noniterative plaintext transformation and a high dimensionally (K-map) populated dynamic Look-Up Table (LUT) with random access that outperforms the security and speed of previous LUT-based chaotic encryption schemes. Experimental analysis of the proposed scheme reveals excellent statistical properties, naturally extended permanent cycle, and high performance for real-time multimedia communications. Our scheme is one order of magnitude faster than the fastest LUT-based approach in the literature and robust to differential and chosen plaintext attacks.

Current chaotic encryption systems in the literature do not fulfill security and performance demands for real-time multimedia communications. To satisfy these demands, we propose a generalized symmetric cryptosystem based on N independently iterated chaotic maps (N-map array) periodically perturbed with a three-level perturbation scheme and a double feedback (global and local) to increase the system's robustness to attacks. The first- and second-level perturbations make cryptosystem extremely sensitive to changes in the plaintext data since the system's output itself (ciphertext global feedback) is used in the perturbation process. Third-level perturbation is a system reset, in which the system-key and chaotic maps are replaced for totally new values. An analysis of the proposed scheme regarding its vulnerability to attacks, statistical properties, and implementation performance is presented. To the best of our knowledge we provide a secure cryptosystem with one of the highest levels of performance for real-time multimedia communications.