Search

This paper will present the project RAPIDMAP. The project is part of CONCERT-Japan, an ERA-NET initiative funded through the FP7 INCO project frame for enhancing research cooperation between European countries and Japan on two topics, one of which is Resilience Against Disasters. The project started in June 2013 and has a duration of 2 years. In the paper, we will outline the aims of the project, methodologies and techniques to be developed and some test data. Remote Sensing (RS) and Geographic Information System (GIS) are powerful technologies for collecting useful information on the damages of disasters in short time. However, since many different types of RS data are available (satellite, aerial, UAV, terrestrial), data co-registration, information integration and feature extraction need reliable and advanced methodologies. In the RAPIDMAP project, we will develop practical ways to integrate RS data processing tools in near-real-time and allow users to use this data soon after the disasters by means of WebGIS tools. This will help not only decision makers but also end-users in the disaster area. The key components of this project are: (1) Near-real-time monitoring: the procedure of near-real-time monitoring with satellites as well as Unmanned Airborne Vehicles (UAV) will be set up and demonstrated. (2) Data co-registration: in disasters, various images as well as maps come from different sources. The co-registration of multiple images is a key technology for information integration. In this project, a system to co-register multiple images in near-real-time will be developed. (3) Data fusion and change detection: one of the advantages of RS is to collect information with multiple sensors. Various methods for fusing optical with active microwave (SAR) sensor data for information extraction and change detection will be developed. (4) Decision Support System (DSS) based on WebGIS technologies: the collected and integrated information has to be easily accessible and visible by decision makers and end-users in near-real-time and worldwide. By using WebGIS technologies, wireless networks and portable terminals, a DSS will allow easy access, retrieval and visualization of all information (fused data, images, maps, etc.) in very short time after data collection and processing. The project will be practically tested and demonstrated at the Tohoku area in Japan and another test site, which were recently affected by large disasters.

Artificial neural networks using pattern recognition methodology for security assessment of electric power systems is presented. Conventional numerical methods are either too complex or time consuming. An alternative method using neural networks to address the security assessment problem and its effectiveness against conventional methods is discussed. Neural networks using pattern recognition techniques is a promising methodology for different types of security assessment. Feature selection and extraction are used for selecting best features having highest discriminating capabilities. An important feature of the approach is that it can be generalized for steady state, transient and dynamic security assessment, which is a desirable feature for on-line security analysis. The proposed approach has been tested on the WSCC 9-bus 3-generator system. Steady state, transient and dynamic security assessment classification and contingency ranking results are provided to highlight the overall classification accuracy and suitability of the approach. � 2007 Elsevier B.V. All rights reserved.

Trabajo presentado al 20th ICPR celebrado en Estambul (Turquía) del 23 al 26 de agosto de 2010., We propose a new algorithm for detecting multiple object categories that exploits the fact that different categories may share common features but with different geometric distributions. This yields an efficient detector which, in contrast to existing approaches, considerably reduces the computation cost at runtime, where the feature computation step is traditionally the most expensive. More specifically, at the learning stage we compute common features by applying the same Random Ferns over the Histograms of Oriented Gradients on the training images. We then apply a boosting step to build discriminative weak classifiers, and learn the specific geometric distribution of the Random Ferns for each class. At runtime, only a few Random Ferns have to be densely computed over each input image, and their geometric distribution allows performing the detection. The proposed method has been validated in public datasets achieving competitive detection results, which are comparable with state-of-the-art methods that use specific features per class., This work was supported by projects: 'CONSOLIDER-INGENIO 2010 Multimodal interaction in pattern recognition and computer vision' (V-00069), 'Robotica ubicua para entornos urbanos' (J-01225), 'Percepción y acción ante incertidumbre' (4803), 'Gardening with a cognitive system' (4920). This work has been partially funded by the Spanish Ministry of Science and Innovation under projects UbROB DPI2007-61452, PAU DPI2008-06022, and MIPRCV Consolider Ingenio 2010 CSD2007-00018; and by the EU project GARNICS FP7-247947. The first au- thor is funded by the Technical University of Catalonia (UPC).

Trabajo presentado a la ICPR 2010 celebrada en Estambul del 23 al 26 de agosto., Cast shadows add additional difficulties on detecting objects because they locally modify image intensity and color. Shadows may appear or disappear in an image when the object, the camera, or both are free to move through a scene. This work evaluates the performance of an object detection method based on boosted HOG paired with three different image representations in outdoor video sequences. We follow and extend on the taxonomy from van de Sande with considerations on the constraints assumed by each descriptor on the spatial variation of the illumination. We show that the intrinsic image representation consistently gives the best results. This proves the usefulness of this representation for object detection in varying illumination conditions, and supports the idea that in practice local assumptions in the descriptors can be violated., This work has been partially funded by the Spanish Ministry of Science and Innovation under projects UbROB DPI2007-61452, and MIPRCV Consolider Ingenio 2010 CSD2007-00018. The first author is funded by the Technical University of Catalonia (UPC).

The objective of this work was to develop a computer aided system which will facilitate the automated characterization of carotid plaques recorded from high resolution ultrasound images for the identification of individuals with asymptomatic carotid stenosis at risk of stroke. The plaques were classified into (i) symptomatic or (ii) asymptomatic. Ten different texture feature sets were extracted from the segmented plaque image using the following algorithms: fast order statistics, spatial gray level dependence matrices, gray level difference statistics, neighborhood gray tone difference matrix, statistical feature matrix, Laws texture energy measures, fractal dimension texture analysis, Fourier power spectrum and shape parameters. Although the statistics of all features extracted for the two classes indicated a high degree of overlap, a classification of the plaques was possible using the unsupervised self-organizing feature map (SOFM) classifier and combining techniques. The classification results of the different feature sets were combined using (i) majority voting and (ii) weighted averaging based on a confidence measure derived from the SOFM. Combining the classification results of the ten different feature sets improved significantly the classification results obtained by the individual feature sets, reaching an average diagnostic yield of 75%. 5 3591 3596 Cited By :15

International Conference on Pattern Recognition (ICPR), 1996, Viena (Austria), The computation of linear moment matrices, whose elements are defined as zeroth order integration values of an image, was recently introduced as a tool to reduce the computational cost required to obtain the geometric moments of an image. The main relevance of these matrices is twofold: on one hand, they can be efficiently obtained by means of accumulation filters, which only require additions; on the other one, their relation to geometric moments, as well as to discrete Fourier spectrum coefficients, allows the exchange and interpretation of many results from different areas of image processing and pattern recognition. Taking into account the relevance of these matrices, a new recursive property that allows their efficient computation in sliding-window processes is presented here. First, a scalar recursive updating rule is formulated. It relates the value of each element of a linear moment matrix to those calculated in the previous location of the sliding-window. Then, this result is reformulated to obtain an explicit matrix formula. The obtained recursive updating rule has a straightforward application in many different fields involving sliding window processes in order to efficiently obtain local features related to geometric moments and discrete Fourier spectrum coefficients.

The shape analysis of otoliths, which are calcified structures in the inner ear of teleostean fishes, is known to be particularly relevant to address species identification and stock discrimination. Generally, scientists use classical methodologies of statistical analysis and shape recognition such as Fourier shape descriptors and Principal Component Analysis (PCA). These methods are subject to several limitations mainly to their incapacity to locate irregularities because they are based on global characterization of shape. Recently, more advanced techniques are proposed in this context in order to improve classification accuracies. The first recent method exploits the potential of shape geodesics which rely on local shape features for classification issues. The second one addresses the Best-Basis paradigm which combines the Wavelet Transform, and the potential of statistical analysis in order to fully automate the selection process of efficient features for classification. These methods have been shown to significantly outperform the standard approaches but they are not compared together yet. This study compare these two methods on a real dataset. The comparison is performed on 600 striped red mullet calcified structures collected for the NESPMAN European project. For each method, performances are reported for the classification of samples coming from three geographical zones in the Northwest European seas: the Bay of Biscay, a mixing zone composed of the Celtic Sea and the Western English Channel and a northern zone composed of the Eastern English Channel and the North Sea. Comparison shows that both methods lead to same conclusions.

Face classification can be defined as the problem of assigning a predefined label to an image or subpart of an image that contains one ore more faces. This definition comprises many sub disciplines in the visual pattern recognition field: (i) face detection, where the goal is to detect the presence of a face on an image, (ii) face recognition, where we assign an identifier label to the detected face, (iii) face verification, where the identity of the subject is given, and we should assure its truthfulness and (iv) gender recognition where the label male or female is assigned to each face image. The information source of a facial image can divided in two sets, depending on the zone of the face. The internal information is composed by the eyes, nose and mouth, while the external features are the regions of the hair, forehead, both laterals, ears, jaw line and chin. Traditionally, face recognition algorithms have used only the internal information of face images for classification purposes since these features can be easily extracted. In fact, most of these algorithms use the aligned thumbnails as an input for some feature extraction process that yields a final feature set used to train the classifier. Classic examples of this approach are the eigenfaces technique (Turk & Pentland, 1991), or the use of Fisher Linear Discriminant Analysis (Hespanha Belhumeur & Kriegman, 1997). Moreover, in the face classification field, there are a lot of security related applications where the reliability obtained by the internal features is essential: notice that the external information is more variable and easier to imitate. For this reason, the use of external features for these securityrelated tasks has often been ignored, given their changing nature. However, with the advances of technology in chip integration, small embedded computers are more integrated in our everyday life, favouring the appearance of new applications not directly related to security dealing with face classification, where the users do not make specific efforts to mislead the classifier. Typical examples are embedded camera-devices for human userfriendly interfaces, user profiling, or reactive marketing. In these cases we consider the external features as an extra source of information for improving the accuracies obtained using only internal features. Furthermore, notice that this consideration can be specially beneficial in natural and uncontrolled environments, where usually artefacts such as strong local illumination changes or partial occlusions difficult the classification task.

[ANGLÈS] Pedestrian detection has been an active area of research in recent years; its interest relies on the potential positive impact on quality of life of the related applications (surveillance systems, automotive safety, robotics, multimedia content analysis, assistive technology and advanced interactive interfaces, among others). The large variability of human appearances, poses and context conditions makes pedestrian detection to be one of the most challenging tasks in computer vision. Although many significant approaches have been proposed lately, pedestrian detection still offers a wide framework of improvement, mainly in terms of accuracy and efficiency. The present thesis aims to study the influence of different training parameters values on the performance of a pedestrian detector. First, a pedestrian detector, using a boosted cascade of Histograms of Oriented Gradients, is built from scratch. Afterwards, a sensitivity analysis is carried out taking into account significant variables, such as the number of training samples, the feature (HOG) or classifier (SVM) parameters, the feature selection technique, the negatives resampling treatment and the typology of the employed weak classifier. [CASTELLÀ] La detección de personas ha sido un área de investigación muy activa en los últimos años; su interés se basa en el gran impacto positivo que las aplicaciones derivadas (sistemas de vigilancia, seguridad en automóviles, robótica, análisis de contenido multimedia, tecnología de ayuda a discapacitados e interfaces interactivas, entre otras) pueden tener sobre la calidad de vida diaria. La gran variabilidad de la apariencia humana, múltiples poses y condiciones del entorno, hace que la detección de personas sea una de las tareas más difíciles de resolver en visión artificial. Aunque últimamente se han propuesto diversas soluciones significativas, la detección de personas todavía ofrece un amplio marco de mejora, sobretodo en términos de precisión y eficiencia. El presente proyecto tiene como objetivo estudiar la influencia sobre el desempeño de un detector de personas de los distintos parámetros de entrenamiento. En primer lugar, se desarrolla un detector de personas desde cero, utilizando una cascada impulsado de histogramas de Gradientes Orientados. Posteriormente, se lleva a cabo un análisis de sensibilidad teniendo en cuenta las variables de entrenamiento más importantes, como son el número de muestras, las características extraídas (HOG) o el algoritmo de clasificación (SVM), la técnica de selección de características, el tratamiento de las muestras negativas y la tipología de clasificador empleado. [CATALÀ] La detecció de persones ha estat una àrea de recerca molt activa en els últims anys; el seu interès es basa en el gran impacte positiu que les aplicacions derivades (sistemes de vigilància, seguretat en automòbils, robòtica, anàlisi de contingut multimèdia, tecnologia d'ajuda a discapacitats i interfícies interactives, entre d'altres) poden tenir sobre la qualitat de vida diària. La gran variabilitat de l'aparença humana, múltiples posicions i condicions d’entorn, fa que la detecció de persones sigui una de les tasques més difícils de resoldre en visió artificial. Encara que últimament s'han proposat diverses solucions significatives, la detecció de persones encara ofereix un ampli marc de millora, sobretot en termes de precisió i eficiència. El present projecte té com a objectiu estudiar la influència sobre el funcionament d'un detector de persones dels diferents paràmetres d'entrenament. En primer lloc, es desenvolupa un detector de persones des de zero, utilitzant una cascada impulsat d'histogrames de Gradients Orientats. Posteriorment, es duu a terme una anàlisi de sensibilitat tenint en compte les variables d'entrenament més importants, com són el nombre total de mostres, les característiques extretes (HOG) o l'algoritme de classificació (SVM), la tècnica de selecció de característiques, el tractament de les mostres negatives i la tipologia de classificador emprat.