Underwater Gigamosaicing

El procés de fusió de dues o més imatges de la mateixa escena en una d'única i més gran és conegut com a Image Mosaicing. Un cop finalitzat el procés de construcció d'un mosaic, els límits entre les imatges són habitualment visibles, degut a imprecisions en els registres fotomètric i geomètric. L'Image Blending és l'etapa del procediment de mosaicing a la que aquests artefactes són minimitzats o suprimits. Existeixen diverses metodologies a la literatura que tracten aquests problemes, però la majoria es troben orientades a la creació de panorames terrestres, imatges artístiques d'alta resolució o altres aplicacions a les quals el posicionament de la càmera o l'adquisició de les imatges no són etapes rellevants. El treball amb imatges subaquàtiques presenta desafiaments importants, degut a la presència d'scattering (reflexions de partícules en suspensió) i atenuació de la llum i a condicions físiques extremes a milers de metres de profunditat, amb control limitat dels sistemes d'adquisició i la utilització de tecnologia d'alt cost. Imatges amb il·luminació artificial similar, sense llum global com la oferta pel sol, han de ser unides sense mostrar una unió perceptible. Les imatges adquirides a gran profunditat presenten una qualitat altament depenent de la profunditat, i la seva degradació amb aquest factor és molt rellevant. El principal objectiu del treball és presentar dels principals problemes de la imatge subaquàtica, seleccionar les estratègies més adequades i tractar tota la seqüència adquisició-procesament-visualització del procés. Els resultats obtinguts demostren que la solució desenvolupada, basada en una Estratègia de Selecció de Límit Òptim, Fusió en el Domini del Gradient a les regions comunes i Emfatització Adaptativa d'Imatges amb baix nivell de detall permet obtenir uns resultats amb una alta qualitat. També s'ha proposat una estratègia, amb possibilitat d'implementació paral·lela, que permet processar mosaics de kilòmetres d'extensió amb resolució de centímetres per píxel.

Paral·lelització especulativa de programes sobre un clúster

Estudi sobre la millora de rendiment (en temps d’execució) al'algorisme de gràfics Fast Multipath Radiosity Using Hierarchical Subscenes gràcies a l’execució paral•lela especulada que ens permet obtenir el motor d'especulació per a clústers desenvolupat en el grup de recerca BCDS de la Universitat de Girona

Ventral and dorsal pathways of speech perception: An intracerebral ERP study.

Recent theory of physiology of language suggests a dual stream dorsal/ventral organization of speech perception. Using intra-cerebral Event-related potentials (ERPs) during pre-surgical assessment of twelve drug-resistant epileptic patients, we aimed to single out electrophysiological patterns during both lexical-semantic and phonological monitoring tasks involving ventral and dorsal regions respectively. Phonological information processing predominantly occurred in the left supra-marginal gyrus (dorsal stream) and lexico-semantic information occurred in anterior/middle temporal and fusiform gyri (ventral stream). Similar latencies were identified in response to phonological and lexico-semantic tasks, suggesting parallel processing. Typical ERP components were strongly left lateralized since no evoked responses were recorded in homologous right structures. Finally, ERP patterns suggested the inferior frontal gyrus as the likely final common pathway of both dorsal and ventral streams. These results brought out detailed evidence of the spatial-temporal information processing in the dual pathways involved in speech perception.

De la perception à l'apprentissage [From perception to learning]

Dans le domaine de la perception, l'apprentissage est contraint par la présence d'une architecture fonctionnelle constituée d'aires corticales distribuées et très spécialisées. Dans le domaine des troubles visuels d'origine cérébrale, l'apprentissage d'un patient hémi-anopsique ou agnosique sera limité par ses capacités perceptives résiduelles, mais un déficit de reconnaissance visuelle de nature apparemment perceptive, peut également être associé à une altération des représentations en mémoire à long terme. Des réseaux neuronaux distincts pour la reconnaissance - cortex temporal - et pour la localisation des sons - cortex pariétal - ont été décrits chez l'homme. L'étude de patients cérébro-lésés confirme le rôle des indices spatiaux dans un traitement auditif explicite du « where » et dans la discrimination implicite du « what ». Cette organisation, similaire à ce qui a été décrit dans la modalité visuelle, faciliterait les apprentissages perceptifs. Plus généralement, l'apprentissage implicite fonde une grande partie de nos connaissances sur le monde en nous rendant sensible, à notre insu, aux règles et régularités de notre environnement. Il serait impliqué dans le développement cognitif, la formation des réactions émotionnelles ou encore l'apprentissage par le jeune enfant de sa langue maternelle. Le caractère inconscient de cet apprentissage est confirmé par l'étude des temps de réaction sériels de patients amnésiques dans l'acquisition d'une grammaire artificielle. Son évaluation pourrait être déterminante dans la prise en charge ré-adaptative. [In the field of perception, learning is formed by a distributed functional architecture of very specialized cortical areas. For example, capacities of learning in patients with visual deficits - hemianopia or visual agnosia - from cerebral lesions are limited by perceptual abilities. Moreover a visual deficit in link with abnormal perception may be associated with an alteration of representations in long term (semantic) memory. Furthermore, perception and memory traces rely on parallel processing. This has been recently demonstrated for human audition. Activation studies in normal subjects and psychophysical investigations in patients with focal hemispheric lesions have shown that auditory information relevant to sound recognition and that relevant to sound localisation are processed in parallel, anatomically distinct cortical networks, often referred to as the "What" and "Where" processing streams. Parallel processing may appear counterintuitive from the point of view of a unified perception of the auditory world, but there are advantages, such as rapidity of processing within a single stream, its adaptability in perceptual learning or facility of multisensory interactions. More generally, implicit learning mechanisms are responsible for the non-conscious acquisition of a great part of our knowledge about the world, using our sensitivity to the rules and regularities structuring our environment. Implicit learning is involved in cognitive development, in the generation of emotional processing and in the acquisition of natural language. Preserved implicit learning abilities have been shown in amnesic patients with paradigms like serial reaction time and artificial grammar learning tasks, confirming that implicit learning mechanisms are not sustained by the cognitive processes and the brain structures that are damaged in amnesia. In a clinical perspective, the assessment of implicit learning abilities in amnesic patients could be critical for building adapted neuropsychological rehabilitation programs.]

Local and Global Feature Extraction for Invariant Object Recognition

Perceiving the world visually is a basic act for humans, but for computers it is still an unsolved problem. The variability present innatural environments is an obstacle for effective computer vision. The goal of invariant object recognition is to recognise objects in a digital image despite variations in, for example, pose, lighting or occlusion. In this study, invariant object recognition is considered from the viewpoint of feature extraction. Thedifferences between local and global features are studied with emphasis on Hough transform and Gabor filtering based feature extraction. The methods are examined with respect to four capabilities: generality, invariance, stability, and efficiency. Invariant features are presented using both Hough transform and Gabor filtering. A modified Hough transform technique is also presented where the distortion tolerance is increased by incorporating local information. In addition, methods for decreasing the computational costs of the Hough transform employing parallel processing and local information are introduced.

Exploration and Design of Power-Efficient Networked Many-Core Systems

Multiprocessing is a promising solution to meet the requirements of near future applications. To get full benefit from parallel processing, a manycore system needs efficient, on-chip communication architecture. Networkon- Chip (NoC) is a general purpose communication concept that offers highthroughput, reduced power consumption, and keeps complexity in check by a regular composition of basic building blocks. This thesis presents power efficient communication approaches for networked many-core systems. We address a range of issues being important for designing power-efficient manycore systems at two different levels: the network-level and the router-level. From the network-level point of view, exploiting state-of-the-art concepts such as Globally Asynchronous Locally Synchronous (GALS), Voltage/ Frequency Island (VFI), and 3D Networks-on-Chip approaches may be a solution to the excessive power consumption demanded by today’s and future many-core systems. To this end, a low-cost 3D NoC architecture, based on high-speed GALS-based vertical channels, is proposed to mitigate high peak temperatures, power densities, and area footprints of vertical interconnects in 3D ICs. To further exploit the beneficial feature of a negligible inter-layer distance of 3D ICs, we propose a novel hybridization scheme for inter-layer communication. In addition, an efficient adaptive routing algorithm is presented which enables congestion-aware and reliable communication for the hybridized NoC architecture. An integrated monitoring and management platform on top of this architecture is also developed in order to implement more scalable power optimization techniques. From the router-level perspective, four design styles for implementing power-efficient reconfigurable interfaces in VFI-based NoC systems are proposed. To enhance the utilization of virtual channel buffers and to manage their power consumption, a partial virtual channel sharing method for NoC routers is devised and implemented. Extensive experiments with synthetic and real benchmarks show significant power savings and mitigated hotspots with similar performance compared to latest NoC architectures. The thesis concludes that careful codesigned elements from different network levels enable considerable power savings for many-core systems.

Turbulent shallow-water model for orographic subgrid-scale perturbations

A parallel pseudo-spectral method for the simulation in distributed memory computers of the shallow-water equations in primitive form was developed and used on the study of turbulent shallow-waters LES models for orographic subgrid-scale perturbations. The main characteristics of the code are: momentum equations integrated in time using an accurate pseudo-spectral technique; Eulerian treatment of advective terms; and parallelization of the code based on a domain decomposition technique. The parallel pseudo-spectral code is efficient on various architectures. It gives high performance onvector computers and good speedup on distributed memory systems. The code is being used for the study of the interaction mechanisms in shallow-water ows with regular as well as random orography with a prescribed spectrum of elevations. Simulations show the evolution of small scale vortical motions from the interaction of the large scale flow and the small-scale orographic perturbations. These interactions transfer energy from the large-scale motions to the small (usually unresolved) scales. The possibility of including the parametrization of this effects in turbulent LES subgrid-stress models for the shallow-water equations is addressed.

Video transcoding in a distributed cloud computing environment

Video transcoding refers to the process of converting a digital video from one format into another format. It is a compute-intensive operation. Therefore, transcoding of a large number of simultaneous video streams requires a large amount of computing resources. Moreover, to handle di erent load conditions in a cost-e cient manner, the video transcoding service should be dynamically scalable. Infrastructure as a Service Clouds currently offer computing resources, such as virtual machines, under the pay-per-use business model. Thus the IaaS Clouds can be leveraged to provide a coste cient, dynamically scalable video transcoding service. To use computing resources e ciently in a cloud computing environment, cost-e cient virtual machine provisioning is required to avoid overutilization and under-utilization of virtual machines. This thesis presents proactive virtual machine resource allocation and de-allocation algorithms for video transcoding in cloud computing. Since users' requests for videos may change at di erent times, a check is required to see if the current computing resources are adequate for the video requests. Therefore, the work on admission control is also provided. In addition to admission control, temporal resolution reduction is used to avoid jitters in a video. Furthermore, in a cloud computing environment such as Amazon EC2, the computing resources are more expensive as compared with the storage resources. Therefore, to avoid repetition of transcoding operations, a transcoded video needs to be stored for a certain time. To store all videos for the same amount of time is also not cost-e cient because popular transcoded videos have high access rate while unpopular transcoded videos are rarely accessed. This thesis provides a cost-e cient computation and storage trade-o strategy, which stores videos in the video repository as long as it is cost-e cient to store them. This thesis also proposes video segmentation strategies for bit rate reduction and spatial resolution reduction video transcoding. The evaluation of proposed strategies is performed using a message passing interface based video transcoder, which uses a coarse-grain parallel processing approach where video is segmented at group of pictures level.

Density, proportion, and dendritic coverage of retinal ganglion cells of the common marmoset (Callithrix jacchus jacchus)

We performed a quantitative analysis of M and P cell mosaics of the common-marmoset retina. Ganglion cells were labeled retrogradely from optic nerve deposits of Biocytin. The labeling was visualized using horseradish peroxidase (HRP) histochemistry and 3-3'diaminobenzidine as chromogen. M and P cells were morphologically similar to those found in Old- and New-World primates. Measurements were performed on well-stained cells from 4 retinas of different animals. We analyzed separate mosaics for inner and outer M and P cells at increasing distances from the fovea (2.5-9 mm of eccentricity) to estimate cell density, proportion, and dendritic coverage. M cell density decreased towards the retinal periphery in all quadrants. M cell density was higher in the nasal quadrant than in other retinal regions at similar eccentricities, reaching about 740 cells/mm² at 2.5 mm of temporal eccentricity, and representing 8-14% of all ganglion cells. P cell density increased from peripheral to more central regions, reaching about 5540 cells/mm² at 2.5 mm of temporal eccentricity. P cells represented a smaller proportion of all ganglion cells in the nasal quadrant than in other quadrants, and their numbers increased towards central retinal regions. The M cell coverage factor ranged from 5 to 12 and the P cell coverage factor ranged from 1 to 3 in the nasal quadrant and from 5 to 12 in the other quadrants. These results show that central and peripheral retinal regions differ in terms of cell class proportions and dendritic coverage, and their properties do not result from simply scaling down cell density. Therefore, differences in functional properties between central and peripheral vision should take these distinct regional retinal characteristics into account.

What ethologically based models have taught us about the neural systems underlying fear and anxiety

Classical Pavlovian fear conditioning to painful stimuli has provided the generally accepted view of a core system centered in the central amygdala to organize fear responses. Ethologically based models using other sources of threat likely to be expected in a natural environment, such as predators or aggressive dominant conspecifics, have challenged this concept of a unitary core circuit for fear processing. We discuss here what the ethologically based models have told us about the neural systems organizing fear responses. We explored the concept that parallel paths process different classes of threats, and that these different paths influence distinct regions in the periaqueductal gray - a critical element for the organization of all kinds of fear responses. Despite this parallel processing of different kinds of threats, we have discussed an interesting emerging view that common cortical-hippocampal-amygdalar paths seem to be engaged in fear conditioning to painful stimuli, to predators and, perhaps, to aggressive dominant conspecifics as well. Overall, the aim of this review is to bring into focus a more global and comprehensive view of the systems organizing fear responses.

Properties and Algorithms of the KCube Interconnection Networks

The KCube interconnection network was first introduced in 2010 in order to exploit the good characteristics of two well-known interconnection networks, the hypercube and the Kautz graph. KCube links up multiple processors in a communication network with high density for a fixed degree. Since the KCube network is newly proposed, much study is required to demonstrate its potential properties and algorithms that can be designed to solve parallel computation problems. In this thesis we introduce a new methodology to construct the KCube graph. Also, with regard to this new approach, we will prove its Hamiltonicity in the general KC(m; k). Moreover, we will find its connectivity followed by an optimal broadcasting scheme in which a source node containing a message is to communicate it with all other processors. In addition to KCube networks, we have studied a version of the routing problem in the traditional hypercube, investigating this problem: whether there exists a shortest path in a Qn between two nodes 0n and 1n, when the network is experiencing failed components. We first conditionally discuss this problem when there is a constraint on the number of faulty nodes, and subsequently introduce an algorithm to tackle the problem without restrictions on the number of nodes.

Performance analysis of FIR digital filter design: RNS versus traditional

Residue Number System (RNS) based Finite Impulse Response (FIR) digital filters and traditional FIR filters. This research is motivated by the importance of an efficient filter implementation for digital signal processing. The comparison is done in terms of speed and area requirement for various filter specifications. RNS based FIR filters operate more than three times faster and consumes only about 60% of the area than traditional filter when number of filter taps is more than 32. The area for RNS filter is increasing at a lesser rate than that for traditional resulting in lower power consumption. RNS is a nonweighted number system without carry propogation between different residue digits.This enables simultaneous parallel processing on all the digits resulting in high speed addition and multiplication in the RNS domain

Performance Analysis of Reversible Fast Decimal Adders

This paper presents a performance analysis of reversible, fault tolerant VLSI implementations of carry select and hybrid decimal adders suitable for multi-digit BCD addition. The designs enable partial parallel processing of all digits that perform high-speed addition in decimal domain. When the number of digits is more than 25 the hybrid decimal adder can operate 5 times faster than conventional decimal adder using classical logic gates. The speed up factor of hybrid adder increases above 10 when the number of decimal digits is more than 25 for reversible logic implementation. Such highspeed decimal adders find applications in real time processors and internet-based applications. The implementations use only reversible conservative Fredkin gates, which make it suitable for VLSI circuits.

Paral·lelització especulativa de programes sobre un clúster

Estudi sobre la millora de rendiment (en temps d’execució) a l'algorisme de gràfics Fast Multipath Radiosity Using Hierarchical Subscenes gràcies a l’execució paral•lela especulada que ens permet obtenir el motor d'especulació per a clústers desenvolupat en el grup de recerca BCDS de la Universitat de Girona

Information servers, present and future

Mainframes, corporate and central servers are becoming information servers. The requirement for more powerful information servers is the best opportunity to exploit the potential of parallelism. ICL recognized the opportunity of the 'knowledge spectrum' namely to convert raw data into information and then into high grade knowledge. Parallel Processing and Data Management Its response to this and to the underlying search problems was to introduce the CAFS retrieval engine. The CAFS product demonstrates that it is possible to move functionality within an established architecture, introduce a different technology mix and exploit parallelism to achieve radically new levels of performance. CAFS also demonstrates the benefit of achieving this transparently behind existing interfaces. ICL is now working with Bull and Siemens to develop the information servers of the future by exploiting new technologies as available. The objective of the joint Esprit II European Declarative System project is to develop a smoothly scalable, highly parallel computer system, EDS. EDS will in the main be an SQL server and an information server. It will support the many data-intensive applications which the companies foresee; it will also support application-intensive and logic-intensive systems.