Search

Capacitive pressure sensors are widely used in a variety of applications and are built using a variety of processes, including 3D printing technology. The use of this technology could lead us to a situation of large deflections, depending on the mechanical properties of the materials and the resolution of the machines used. This aspect is rarely reported in previous research works that focus on improving the performance in terms of linearity and sensitivity of these sensors. This paper describes the realization of relative pressure sensors designed as two different structures; the first one is the classical design composed of a single capacitor, while the second one is composed of two capacitors, designed in such a way that they both vary according to the applied pressure but in opposite senses to each other. The purpose is to study in particular the performance of the second structure in the case of large deflections for the context of educational use. Polylactic acid (PLA) is used as the manufacturing material to print the sensors by means of a printer based on fused deposing modeling, while conductive materials are used to provide the electrical conductivity required for the printed sensors. The manufactured sensors were tested under pressure in the range of [0; 9] KPa. Compared to the performance obtained with the first structure, simulation and experimental results show that the second structure improves linearity and allows the sensitivity to be increased from a minimum of 9.98×10-2 pF/hPa to a minimum of 3.4×10-1 pF/hPa.

Background The Brain Computer Interfaces (BCI) are devices allowing direct communication between the brain of a user and a machine. This technology can be used by disabled people in order to improve their independence and maximize their capabilities such as finding an object in the environment. Such devices can be realized by the non-invasive measurement of information from the cortex by electroencephalography (EEG). Methods Our work proposes a novel BCI system that consists of controlling a robot arm based on the user's thought. Four subjects (1 female and 3 males) aged between 20 and 29 years have participated to our experiment. They have been instructed to imagine the execution of movements of the right hand, the left hand, both right and left hands or the movement of the feet depending on the protocol established. EMOTIV EPOC headset was used to record neuronal electrical activities from the subject's scalp, these activities were then sent to the computer for analysis. Feature extraction was performed using the Principal Component Analysis (PCA) method combined with the Fast Fourier transform (FFT) spectrum within the frequency band responsible for sensorimotor rhythms (8 Hz–22 Hz). These features were then fed into a Support Vector Machine (SVM) classifier based on a Radial Base Function (RBF) whose outputs were translated into commands to control the robot arm. Results The proposed BCI enabled the control of the robot arm in the four directions: right, left, up and down, achieving an averaged accuracy of 85.45% across all the subjects. Conclusion The results obtained would encourage, with further developments, the use of the proposed BCI to perform more complex tasks such as execution of successive movements or stopping the execution once a searched object is detected. This would provide a useful assistance means for people with motor impairment.

Brain–machine interfaces are systems that allow the control of a device such as a robot arm through a person’s brain activity; such devices can be used by disabled persons to enhance their life and improve their independence. This paper is an extended version of a work that aims at discriminating between left and right imagined hand movements using a support vector machine (SVM) classifier to control a robot arm in order to help a person to find an object in the environment. The main focus here is to search for the best features that describe efficiently the electroencephalogram data during such imagined gestures by comparing two feature extraction methods, namely the continuous wavelet transform (CWT) and the empirical modal decomposition (EMD), combined with the principal component analysis (PCA) that were fed through a linear and radial basis function (RBF) kernel SVM classifier. The experimental results showed high performance achieving an average accuracy across all the subjects of 92.75% with an RBF kernel SVM classifier using CWT and PCA compared to 80.25% accuracy obtained with EMD and PCA. The proposed system has been implemented and tested using data collected from five male subjects and it enabled the control of the robot arm in the right and the left direction.

In this article, the continuous wavelet transform, based on the Mexican Hat function, is used to achieve an automatic processing and classification of visual evoked potentials. Indeed, the representations of the modulus of the wavelet coefficients in the time-scale plan, allowed us to define qualitative criterion for the discrimination of normal and pathological cases. In order to enhance the classification rate, we developed a new method that considers the visual evoked potential's scalogram as an image which is segmented, according to the maximal level of energy density, into many regions of significant interest. The analysis of the resulting segmented image permits to extract a vector of most significant features that will be used to classify normal and pathological signals through SVM classifier. The obtained results show that this technique not only, allows a more reliable distinction between normal and pathological cases, with a very high classification rate (93%) in comparison with the one of the conventional latency measurement (67%), but also, suggests that indications on the progression of the pathology can be provided.

International audience; Gesture recognition is one of the important tasks for human System Interaction (HRI). This paper describes a novel approach intended to recognize 3D dynamic composed gestures by combining Dynamic Time Warping (DTW) with an Adaptive Sliding Window which the name Adaptive Dynamic Time Warping (ADTW). We use the skeleton algorithm provided by the Kinect SDK to track the upper part of body and extract joints angles based on depth information. Each gesture is represented by the combination of angles variations and stored described as a vector. A composed gesture is a sequence of two simple gestures or more performed successively in time. We chose five simple gestures : come, recede, point to the right, point to the left and stop. For each simple gesture, we chose a reference sequence that perfectly represents it. In order to recognize all gesture of the composed gesture in the right order, we combine (DTW) with an Adaptive Sliding Window. In one hand, we use an adaptive window to browse through the sequence of the composed gesture by feeding it to each time with new data. In other hand, we use DTW to compare between the reference gestures and the the sequences defined by the adaptive window. In fact, by comparing each two sequences, DTW computes the euclidean distance between them. Finally, the reference gesture which gives the lower distance is considered as the source class of the tested gesture.

Sleep spindles are among the hallmarks observed in the electroencephalogram (EEG) that occur during non-rapid eye movement sleep precisely in stage 2. They are transient waveforms of biological and clinical interest. In this paper, we present a method to detect spindles in raw EEG recordings during sleep. The method consists of processing the signal through an adaptive autoregressive model whose features are represented by the zeroes of the model polynomial rather than the prediction coefficients. Tracing in time of the zeros’ modulus shows sharp transitions indicating statistical changes due to the occurrence of spindles. The results obtained were compared to those reported by other techniques based on traditional EEG visual reading used by neurophysiologists.

Resume L’etude des dimensions des dents permet d’evaluer l’encombrement dentaire qui est un des principaux parametres d’une dysharmonie dentomaxillaire ; calcule systematiquement pour chaque patient consultant en orthopedie dentofaciale. La presente etude a pour objectif de comparer entre quatre methodes de mesure de la taille des dents. Elle a ete realisee sur 30 moulages de patients marocains ayant consulte au service d’orthopedie dentofaciale de la faculte de medecine dentaire de Rabat. Elle consiste en la mesure du plus grand diametre mesiodistal des dents mandibulaires en avant des premieres molaires par les quatre methodes de mesure. L’analyse statistique a demontre que les methodes utilisant le pied a coulisse numerique, le pied a coulisse analogique avec vernier et l’outil informatique sont fiables, precises et peuvent etre interchangeables. Alors que la methode a l’aide du pied a coulisse analogique sans tenir compte du vernier est moins precise, d’ou l’interet de l’utilisation de ce dernier.

Resume La dysharmonie dento-maxillaire est la cause de consultation la plus frequente en orthopedie dento-faciale. Lorsque le patient se presente en denture mixte, le praticien doit diagnostiquer un eventuel probleme de manque de place en predisant la place necessaire a l’etablissement d’une denture permanente complete a partir de la taille des dents deja evoluees. Dans cette etude, nous avons etudie la fiabilite, dans une population marocaine, de deux methodes de prevision de la taille des dents intermediaires a partir de la taille des incisives inferieures, celle de Moyers et celle de Tanaka et Johnson. Au terme de l’etude, la methode de Tanaka et Johnson nous semble plus appropriee.

International audience; Gesture recognition is one of the important tasks for human Robot Interaction (HRI). This paper describes a novel system intended to recognize 3D dynamic gestures based on depth information provided by Kinect sensor. The proposed system utilizes tracking for the upper body part and combines the hidden Markov models (HMM) and dynamic time warping (DTW) to avoid gestures misclassification. By using the skeleton algorithm provided by the Kinect SDK, body is tracked and joints information are extracted. Each gesture is characterized by one of the angles which remains active when executing it. The variations of the angles throughout the gesture are used as inputs of Hidden Markov Models (HMM) in order to recognize the dynamic gestures. By feeding the output of (HMM) back to (DTW), we achieved good classification performances without any misallocation. Besides that, using depth information only makes our method robust against environmental conditions such as illumination changes and scene complexity.

International audience; In this paper we propose a recognition technique of 3D dynamic gesture for human robot interaction (HRI)based on depth information provided by Kinect sensor. The body is tracked using the skeleton algorithm providedby the Kinect SDK. The main idea of this work is to compute the angles of the upper body joints whichare active when executing gesture. The variation of these angles are used as inputs of Hidden Markov Models(HMM) in order to recognize the dynamic gestures. Results demonstrate the robustness of our method againstenvironmental conditions such as illumination changes and scene complexity due to using depth informationonly

P300 is one of the most studied and used event related potentials (ERP) in brain computer interfaces (BCI). The classical oddball paradigm is usually used to evoke the P300 from Electroencephalogram (EEG) signals. However, EEG raw data are noisy which make the P300 detection very difficult. In this paper, we aim to detect the P300 wave as accurate as possible using appropriate feature extraction method combined with powerful classifier. We compared four methods: Kernel principal component analysis (KPCA), Principal component analysis (PCA), Independent component analysis (ICA) and Linear discriminant analysis (LDA). Each method is used with a linear support vector machine (SVM) classifier and tested on EEG signals from three channels (FZ, CZ and PZ) of four healthy subjects. The results show that the P300 wave is efficiently detected in PZ channel using KPCA-SVM.

International audience; We have previously shown that mental rehearsal can replace up to 75% of physical practice for learning a visuomotor task (Allami, Paulignan, Brovelli, & Boussaoud, (2008). Experimental Brain Research,184,105-113). Presumably, mental rehearsal must induce brain changes that facilitate motor learning. We tested this hypothesis by recording scalp electroencephalographic activity (EEG) in two groups of subjects. In one group, subjects executed a reach to grasp task for 240 trials. In the second group, subjects learned the task through a combination of mental rehearsal for the initial 180 trials followed by the execution of 60 trials. Thus, one group physically executed the task for 240 trials, the other only for 60 trials. Amplitudes and latencies of event-related potentials (ERPs) were compared across groups at different stages during learning. We found that ERP activity increases dramatically with training and reaches the same amplitude over the premotor regions in the two groups, despite large differences in physically executed trials. These findings suggest that during. mental rehearsal, neuronal changes occur in the motor networks that make physical practice after mental rehearsal more effective in configuring functional networks for skilful behaviour. (C) 2013 Elsevier Ltd. All rights reserved.

Breast cancer remains one of the deadliest diseases among women. Microcalcifications can be an early indicator of a breast cancer. Thus when they are present their detection during a screening test is very crucial. But due to their small size and low contrast in mammographies their detection is difficult. Therefore many computer aided diagnosis mathods have been developped to help and assist rediologist during their screening tests. This paper presents a novel approach to detect microcalcifications on digitized mammaographies using fuzzy logic and support vector machines. Our method was tested on 16 mammograms from Mias database including both positive and negative cases. We have obtained very satisfactory results with a sensitivity of 99,60% and a specificity of 99,11% during the learning phase.

Functional magnetic resonance imaging (fMRI) exploits blood-oxygen-level-dependent (BOLD) contrasts to map neural activity associated with a variety of brain functions including sensory processing, motor control, and cognitive and emotional functions. The general linear model (GLM) approach is used to reveal task-related brain areas by searching for linear correlations between the fMRI time course and a reference model. One of the limitations of the GLM approach is the assumption that the covariance across neighbouring voxels is not informative about the cognitive function under examination. Multivoxel pattern analysis (MVPA) represents a promising technique that is currently exploited to investigate the information contained in distributed patterns of neural activity to infer the functional role of brain areas and networks. MVPA is considered as a supervised classification problem where a classifier attempts to capture the relationships between spatial pattern of fMRI activity and experimental conditions. In this paper , we review MVPA and describe the mathematical basis of the classification algorithms used for decoding fMRI signals, such as support vector machines (SVMs). In addition, we describe the workflow of processing steps required for MVPA such as feature selection, dimensionality reduction, cross-validation, and classifier performance estimation based on receiver operating characteristic (ROC) curves.

This paper proposes a novel approach that allows region-based active contour energy to be re-expressed combining local and global information. The basic idea of this technique consists in extracting image statistics locally from the heterogeneous region (foreground or background) and globally from the other region at each point along the curve. By exploiting benefits of both local-based and global-based statistics, this technique proves to be robust against heterogeneity and noise and shows low sensitivity to curve initialization. Experimental results for synthetic and real images reveal significant improvement compared to conventional methods.

Tooth-size assessment is a useful indicator of dental crowding which constitutes one of the main parameters in tooth-size arch-length discrepancy (TSALD). Systematic TSALD evaluation is imperative in all patients consulting for dentofacial orthopedic treatment. The present study aims to compare four measurement techniques for calibrating tooth size. It was performed on casts taken from 30 Moroccan patients consulting at the Dentofacial Orthopedics Department of the Rabat School of Dentistry. Four measurement methods were used to calculate the largest mesiodistal diameter of mandibular teeth mesial to the first molars. Statistical analysis showed that the methods using a digital caliper, the analog vernier caliper and the computerized technique are reliable, accurate, and interchangeable. In contrast, the analog caliper with no vernier is less accurate.

Encoding video in scalable formats supports fast and efficient adaptation. Fully scalable video bit-stream requires layered structure of most of its components. However, layered representation requires new entropy coding strategies able to efficiently protecting different layers. In this work, we apply multiplexed codes on a scalable video coding (SVC) scheme with a spatial scalability as well as a temporal scalability. Experimental results reveal a very high error resilience at almost no cost in compression efficiency. Tests indicate that with multiplexed codes, it is possible to ensure auto_synchronisation decoder in the case of corrupt bit-stream, without increasing the bit rate or the complexity of treatment.

X-ray radiography is one of the most used techniques in the Non Destructive Testing (NDT). It allows the detection of weld defects the most dangerous for the weld's integrity. Because X-ray images of welds are noisy and low contrasted, it is difficult to detect weld defects inside. The goal of this paper is to segment the defects in X-ray images. However, the segmentation remains among the most difficult tasks in image processing, especially in the case of noisy or low contrasted images. Many researchers used neural networks, fuzzy logic methods or SVM-based methods to segment this type of images. The results are impressive; however they require a complex implementation and are time consuming because of learning stage. In this work, we present a new method of segmentation of digitized radiographic images of welds which is based on thresholding techniques and compare it with a multiple thresholding and Support Vector Machines based method. We obtained the same results in terms of visual segmentation quality, but our algorithm is faster.

Detection of defects in metallic pieces is an important application in the field of Non-Destructive Testing (NDT), particularly in an industrial setting. These defects are mainly due to manufacturing errors or welding processes. In this article we will focus on this second category of defects using segmentation techniques applied to thewelded joints. Segmentation is one of the most difficulttasks in image processing, particularly in the case of noisy or low contrast images such as radiographic images of welds. In segmenting this type of image, many researchers have used neural networks and fuzzy logic methods. The results are impressive, however the methods require a complex implementation and are time consuming. In this work, we propose a new method for segmenting digitized radiographic images which is based on histogram analysis, contrast enhancement and image thresholding.Computing time is optimized by using integral images to calculate the local thresholds. Although the method gives comparable results to those obtained by previous methods in terms of visual segmentation quality, it is significantly simpler to implement.

In most image processing applications, the user is only interested in part of the information conveyed through the image. In this paper, we present an adaptive method of compression of still images. It consists first on a semi-automatic extraction of the region of interest (ROI). The background (BG) and the ROI are then processed separately so as to preserve the information in the ROI and degrade in a controlled way the BG. using reversible and irreversible techniques such as JPEG, JPEG 2000 and Run Length Encoding (RLE). Performances of these techniques in terms of compression rate are compared based on their applications to biomedical and satellite images.

The goal of Human System Interaction(HSI) research is to increase the performance of humansystem interaction in order to make it similarto human-human interaction. As for communicationbetween humans, gestural communication isalso widely used in human system interaction. Thisthesis is about gesture recognition for natural humansystem interaction based on gestures. The goalof works conducted in this thesis is to proposerecognition approaches for different kind of dynamicgesture: simple gesture and composed gesture. Fistof all, we proposed a novel 3D gesture descriptorcomputed by human body joints angles provided byKinect sensor. Second, we proposed two dynamicgestures recognition approaches: a) simple gesturesrecognition approach, b) composed gestures recognitionapproach. The first approach is based onHidden Markov Models (HMM). One HMM wascreated for each gesture. The joints angles variationshave been used as input for HMMs. Then,this method has been combined with the DynamicTime Warping algorithm in order to eliminate badclassification. The second approach treats the caseof composed and successive gestures in the samesequence. This approach combines Dynamic TimeWarping method with an adaptative window, hencethe name: Adaptive Dynamic Time Warping. Twoversions have been proposed: Forward and Backwardversion for online and offline recognition., Le but des applications visées par l’interaction homme-système est de parvenir à une interaction naturelle qui simule l’interaction homme-homme.Comme dans la communication homme-homme,les gestes sont aussi très utilisés dans la communication homme-système. Cette thèse porte sur la reconnaissance de gestes pour l’interaction naturelle homme-système basée sur les gestes. L’objectif des travaux menés durant cette thèse est de proposer des approches de reconnaissance de différents types de geste dynamiques : gestes simples et gestes composés. Tous d'abord, nous avons proposé un nouveau descripteur 3D de gestes calculé par les angles des articulations du corps humain à partir d’un flux de profondeur fourni par le capteur Kinect. Ensuite, nous avons proposé deux approches pour la reconnaissance de gestes dynamiques : a)une approche de reconnaissance des gestes simples,b) une approche de reconnaissance des gestes composés.La première approche est basée sur les Modèles de Markov Cachés. Un modèle MMC pour chaque geste a été réalisé. La variation des angles entre les articulations est utilisée comme entrée des Modèles de Markov Cachés. Cette méthode a été combinée avec la méthode de la Déformation Temporelle Dynamique (Dynamic Time Warping) pour éliminer les mauvaises classifications. La deuxième approche traite le cas des gestes composés et successifs dans une même séquence. Cette approche combine la méthode de la Déformation Temporelle Dynamique avec une fenêtre glissante adaptative d’où le nom de l’approche: Adaptive Dynamic TimeWarping. Deux versions de cette approche ont été proposées : version Forward et version Backward pour la reconnaissance en ligne et hors ligne.