Property:Abstract
From ISLAB/CAISR
Jump to navigationJump to searchThis is a property of type Text.
C
Combining neural networks, fuzzy sets, and the evidence theory based techniques for detecting colour specks +
<p>An approach to detecting colour specks in an image taken from a pulp sample of recycled paper is presented. The task is solved through pixel-wise colour classification by an artificial neural network and post-processing based on the evidence theory. The network is trained using possibilistic target values, which are determined through a self-organising process in a 2D and 1D map of chromaticity and lightness, respectively. The problem of post-processing of a pixelwise-classified image is addressed from the point of view of the Dempster-Shafer theory of evidence. Each neighbour of a pixel being analysed is considered as an item of evidence supporting particular hypotheses regarding the class label of that pixel. The strength of support is defined as a function of the degree of uncertainty in class label of the neighbour, and the distance between the neighbour and the pixel being considered. The experiments performed have shown that the colour classification results correspond well with the human perception of colours of the specks.</p> +
Combining traditional and neural-based techniques for ink feed control in a newspaper printing press +
<p>A SOM based model combination strategy, allowing to create adaptive – data dependent – committees, is proposed. Both, models included into a committee and aggregation weights are specific for each input data point analyzed. The possibility to detect outliers is one more characteristic feature of the strategy.</p> +
<p>In the above paper, an example is given, showing that the LQ controller gives an arbitrary small gain margin with respect to variations of the open-loop plant. As a remedy, a dynamic-state feedback is proposed which is claimed to give an arbitrary large gain margin. This is incorrect. In fact, the proposed dynamic state feedback controller does not even stabilize the nominal system.</p> +
<p>In this paper, we present a new approach for periocular recognition based on the Symmetry Assessment by Feature Expansion (SAFE) descriptor, which encodes the presence of various symmetric curve families around image key points. We use the sclera center as single key point for feature extraction, highlighting the object-like identity properties that concentrates to this unique point of the eye. As it is demonstrated, such discriminative properties can be encoded with a reduced set of symmetric curves. Experiments are done with a database of periocular images captured with a digital camera. We test our system against reference periocular features, achieving top performance with a considerably smaller feature vector (given by the use of a single key point). All the systems tested also show a nearly steady correlation between acquisition distance and performance, and they are also able to cope well when enrolment and test images are not captured at the same distance. Fusion experiments among the available systems are also provided. © 2016 IEEE.</p> +
Compact orientation and frequency estimation with applications in biometrics : Biometrics on the orientation express +
<p>Automatic feature extraction still remains a relevant image and signal processing problem even tough both the field and technologies are developing rapidly. Images of low quality, where it is extremely difficult to reliably process image information automatically, are of special interest. To such images we can refer forensic fingerprints, which are left unintentionally on different surfaces andare contaminated by several of the most difficult noise types. For this reason, identification of fingerprints is mainly based on the visual skills of forensic examiners. We address the problem caused by low quality in fingerprints by connecting different sources of information together, yielding dense frequency and orientation maps in an iterative scheme. This scheme comprises smoothing ofthe original, but only along, ideally never across, the ridges. Reliable estimation of dense maps allows to introduce a continuous fingerprint ridge counting technique. In fingerprint scenario the collection of irrefutable tiny details, e.g. bifurcation of ridges, called minutiae, is used to tie the pattern of such points and their tangential directions to the finger producing the pattern. This limited feature set, location and direction of minutiae, is used in current AFIS systems, while fingerprint examiners use the extended set of features, including the image information between the points. With reasonably accurate estimationsof dense frequency and orientation maps at hand, we have been able to propose a novel compact feature descriptor of arbitrary points. We have used these descriptors to show that the image information between minutiae can be extracted automatically and be valuable for identity establishment of forensic images even if the underlying images are noisy. We collect and compress the image information in the neighborhoods of the fine details, such as minutiae, to vectors, one per minutia, and use the vectors to "color" the minutiae. When matching two patterns (of minutiae) even the color of the minutia must match to conclude that they come from the same identity. This feature development has been concentrated and tested on forensic fingerprint images. However, we have also studied an extension of its application area to other biometrics, periocular regions of faces. This allowed us to test the persistence of automatically extracted features across different types of imagesand image qualities, supporting its generalizability.</p>
<p>The aim of this study was to compare acoustic and throat microphones in the voice pathology detection task. Recordings of sustained phonation /a/ were used in the study. Each recording was characterized by a rather large set of diverse features, 1051 features in total. Classification into two classes, namely normal and pathological, was performed using random forest committees. Models trained using data obtained from the throat microphone provided lower classification accuracy. This is probably due to a narrower frequency range of the throat microphone leading to loss of important information.</p> +
Comparison and Fusion of Multiple Iris and Periocular Matchers Using Near-Infrared and Visible Images +
<p>Periocular refers to the facial region in the eye vicinity. It can be easily obtained with existing face and iris setups, and it appears in iris images, so its fusion with the iris texture has a potential to improve the overall recognition. It is also suggested that iris is more suited to near-infrared (NIR) illu- mination, whereas the periocular modality is best for visible (VW) illumination. Here, we evaluate three periocular and three iris matchers based on different features. As experimen- tal data, we use five databases, three acquired with a close-up NIR camera, and two in VW light with a webcam and a dig- ital camera. We observe that the iris matchers perform better than the periocular matchers with NIR data, and the opposite with VW data. However, in both cases, their fusion can pro- vide additional performance improvements. This is specially relevant with VW data, where the iris matchers perform sig- nificantly worse (due to low resolution), but they are still able to complement the periocular modality. © 2015 IEEE.</p> +
Comparison of Machine Learning Approaches for Soil Embedding Detection of Planetary Exploration Rovers +
<p>This paper analyzes the advantages and limitations of known machine learning approaches to cope with the problem of incipient rover embedding detection based on propioceptive signals. In particular, two supervised learning approaches (Support Vector Machines and Feed-forward Neural Networks) are compared to two unsupervised learning approaches (K-means and Self-Organizing Maps) in order to identify various degrees of slip (e.g. low slip, moderate slip, high slip). A real dataset collected by a single-wheel testbed available at MIT has been used to validate each strategy. The SVM algorithm achieves the best performance (accuracy >95 %). However, the SOM algorithm represents a better solution in terms of accuracy and the need of hand-labeled data for training the classifier (accuracy >84 %).</p> +
<p>The paper presents results of the face verification contest that was organized in conjunction with International Conference on Pattern Recognition 2000 (14). Participants had to use identical data sets from a large, publicly available multimodal database XM2VTSDB. Training and evaluation was carried out according to an a priori known protocol ((7)). Verification results of all tested algorithms have been collected and made public on the XM2VTSDB website (15), facilitating large scale experiments on classifier combination and fusion. Tested methods included, among others, representatives of the most common approaches to face verification - elastic graph matching, Fisher's linear discriminant and Support vector machines.</p> +
<p>Activity recognition in smart environments is essential for ensuring the wellbeing of older residents. By tracking activities of daily living (ADLs), a person’s health status can be monitored over time. Nonetheless, accurate activity classification must overcome the fact that each person performs ADLs in different ways and in homes with different layouts. One possible solution is to obtain large amounts of data to train a supervised classifier. Data collection in real environments, however, is very expensive and cannot contain every possible variation of how different ADLs are performed. A more cost-effective solution is to generate a variety of simulated scenarios and synthesize large amounts of data. Nonetheless, simulated data can be considerably different from real data. Therefore, this paper proposes the use of regression models to better approximate real observations based on simulated data. To achieve this, ADL data from a smart home were first compared with equivalent ADLs performed in a simulator. Such comparison was undertaken considering the number of events per activity, number of events per type of sensor per activity, and activity duration. Then, different regression models were assessed for calculating real data based on simulated data. The results evidenced that simulated data can be transformed with a prediction accuracy of R2 = 97.03%.</p><p>© Springer Science+Business Media, LLC, part of Springer Nature 2020</p> +
Complex Filters Applied to Fingerprint Images Detecting Prominent Symmetry Points Used for Alignment +
<p>For the alignment of two fingerprints position of certain landmarks are needed. These should be automatically extracted with low misidentification rate. As landmarks we suggest the prominent symmetry points (core-points) in the fingerprint. They are extracted from the complex orientation field estimated from the global structure of the fingerprint, i.e. the overall pattern of the ridges and valleys. Complex filters, applied to the orientation field in multiple resolution scales, are used to detect the symmetry and the type of symmetry. Experimental results are reported.</p> +
Comprehensive Bioinformatic Analysis of the Specificity of Human Immunodeficiency Virus Type 1 Protease +
<p>Rapidly developing viral resistance to licensed human immunodeficiency virus type 1 (HIV-1) protease inhibitors is an increasing problem in the treatment of HIV-infected individuals and AIDS patients. A rational design of more effective protease inhibitors and discovery of potential biological substrates for the HIV-1 protease require accurate models for protease cleavage specificity. In this study, several popular bioinformatic machine learning methods, including support vector machines and artificial neural networks, were used to analyze the specificity of the HIV-1 protease. A new, extensive data set (746 peptides that have been experimentally tested for cleavage by the HIV-1 protease) was compiled, and the data were used to construct different classifiers that predicted whether the protease would cleave a given peptide substrate or not. The best predictor was a nonlinear predictor using two physicochemical parameters (hydrophobicity, or alternatively polarity, and size) for the amino acids, indicating that these properties are the key features recognized by the HIV-1 protease. The present in silico study provides new and important insights into the workings of the HIV-1 protease at the molecular level, supporting the recent hypothesis that the protease primarily recognizes a conformation rather than a specific amino acid sequence. Furthermore, we demonstrate that the presence of 1 to 2 lysine residues near the cleavage site of octameric peptide substrates seems to prevent cleavage efficiently, suggesting that this positively charged amino acid plays an important role in hindering the activity of the HIV-1 protease.</p> +
<p>In the cochlea of the inner ear, outer hair cells (OHC) together with the local passive structures of the tectorial and basilar membranes comprise non-linear resonance circuits with the local and central (afferent–efferent) feedback. The characteristics of these circuits and their control possibilities depend on the mechanomotility of the OHC. The main element of our functional model of the OHC is the mechanomotility circuit with the general transfer characteristic y = k tanh(x − a). The parameter k of this characteristic reflects the axial stiffness of the OHC, and the parameter a working position of the hair bundle. The efferent synaptic signals act on the parameter k directly and on the parameter a indirectly through changes in the membrane potential. The dependences of the sensitivity and selectivity on changes in the parameters a and k are obtained by the computer simulation. Functioning of the model at low-level input signals is linear. Due to the non-linearity of the transfer characteristic of the mechanomotility circuit the high-level signals are compressed. For the adaptation and efferent control, however, the transfer characteristic with respect to the initial operating point should be asymmetrical (a > 0). The asymmetry relies on the deflection of the hair bundle from the axis of the OHC.</p> +
<p>In most real-time applications, deadlines are artifices that need to be enforced to meet different performance requirements. For example, in periodic task sets, jitter requirements can be met by assigning suitable relative deadlines and guaranteeing the feasibility of the schedule. This paper presents a method (called minD) for calculating the minimum EDF-feasible deadline of a real-time task. More precisely, given a set of periodic tasks with hard real-time requirements, which is feasible under EDF, the proposed algorithm allows computing the shortest deadline that can be assigned to an arbitrary task in the set, or to a new incoming task (periodic or aperiodic), still preserving the EDF feasibility of the new task set. The algorithm has a pseudo polynomial complexity and handles arbitrary relative deadlines, which can be less than, equal to, or greater than periods.</p> +
<p>Methods for equipment monitoring are traditionally constructed from specific sensors and/or knowledge collected prior to implementation on the equipment. A different approach is presented here that builds up knowledge over time by exploratory search among the signals available on the internal field-bus system and comparing the observed signal relationships among a group of equipment that perform similar tasks. The approach is developed for the purpose of increasing vehicle uptime, and is therefore demonstrated in the case of a city bus and a heavy duty truck. However, it also works fine for smaller mechatronic systems like computer hard-drives. The approach builds on an onboard self-organized search for models that capture relations among signal values on the vehicles’ data buses, combined with a limited bandwidth telematics gateway and an off-line server application where the parameters of the self-organized models are compared. The presented approach represents a new look at error detection in commercial mechatronic systems, where the normal behavior of a system is actually found under real operating conditions, rather than the behavior observed in a number of laboratory tests or test-drives prior to production of the system. The approach has potential to be the basis for a self-discovering system for general purpose fault detection and diagnostics.</p> +
<p>This work deals with the development of an interface to control a smart conference room using passive BCI (Brain Computer Interface). It compares a genetic algorithm developed in a previous project to control the smart conference room with a random control algorithm. The system controls features of the conference room such as air conditioner, lightning systems, electric shutters, entertainment devices, etc. The parameters of the algorithm are extracted from users biosignal using Emotiv Epoc Headset while the user performs an attention test. The tests indicate that the decisions made by the genetic algorithm lead to better results, but in a single execution cannot be considered an effective optimization algorithm. © Springer-Verlag Berlin Heidelberg 2016.</p> +
<p>Periocular recognition has gained attention in the last years thanks to its high discrimination capabilities in less constraint scenarios than other ocular modalities. In this paper we propose a method for periocular verification under different light spectra using CNN features with the particularity that the network has not been trained for this purpose. We use a ResNet-101 pretrained model for the ImageNet Large Scale Visual Recognition Challenge to extract features from the IIITD Multispectral Periocular Database. At each layer the features are compared using χ <sup>2</sup> distance and cosine similitude to carry on verification between images, achieving an improvement in the EER and accuracy at 1% FAR of up to 63.13% and 24.79% in comparison to previous works that employ the same database. In addition to this, we train a neural network to match the best CNN feature layer vector from each spectrum. With this procedure, we achieve improvements of up to 65% (EER) and 87% (accuracy at 1% FAR) in cross-spectral verification with respect to previous studies.</p> +
<p>This work presents the 2nd Cross-Spectrum Iris/Periocular Recognition Competition (Cross-Eyed2017). The main goal of the competition is to promote and evaluate advances in cross-spectrum iris and periocular recognition. This second edition registered an increase in the participation numbers ranging from academia to industry: five teams submitted twelve methods for the periocular task and five for the iris task. The benchmark dataset is an enlarged version of the dual-spectrum database containing both iris and periocular images synchronously captured from a distance and within a realistic indoor environment. The evaluation was performed on an undisclosed test-set. Methodology, tested algorithms, and obtained results are reported in this paper identifying the remaining challenges in path forward. © 2017 IEEE</p> +
<p>Thisworkpresentsthe2ndCross-SpectrumIris/PeriocularRecognitionCompetition(Cross-Eyed2017).The main goal of the competition is topromote and evaluate advances in cross-spectrum iris andperiocular recognition.This second edition registeredan increase in the participation numbers ranging fromacademia to industry: five teams submitted twelve methodsfor the periocular task and five for the iris task. The bench-mark dataset is an enlarged version of the dual-spectrumdatabase containing both iris and periocular images syn-chronously captured from a distance and within a realisticindoor environment. The evaluation was performed on anundisclosed test-set. Methodology, tested algorithms, andobtained results are reported in this paper identifying theremaining challenges in path forward.</p> +
<p>Periocular recognition has gained attention in the last years thanks to its high discrimination capabilities in less constraint scenarios than face or iris. In this paper we propose a method for periocular verification under different light spectra using CNN features with the particularity that the network has not been trained for this purpose. We use a ResNet-101 pretrained model for the ImageNet Large Scale Visual Recognition Challenge to extract features from the IIITD Multispectral Periocular Database. At each layer the features are compared using χ 2 distance and cosine similitude to carry on verification between images, achieving an improvement in the EER and accuracy at 1% FAR of up to 63.13% and 24.79% in comparison to previous works that employ the same database. In addition to this, we train a neural network to match the best CNN feature layer vector from each spectrum. With this procedure, we achieve improvements of up to 65% (EER) and 87% (accuracy at 1% FAR) in cross-spectral verification with respect to previous studies.</p> +
