CHARACTER RECOGNITION SYSTEMS

Author: GONZÁLEZ VELASCO HORACIO MANUEL.
Year: 2002.
University: EXTREMADURA [www.unex.es].
Place of defense: FACULTAD DE CIENCIAS.
Place of preparation: FACULTAD DE CIENCIAS.

Summary: The thesis proposes a methodology for obtaining the contours side, front and rear animals in cattle regardless of their position or rotate the image and the scale of the same. This part of a database with the contours of -- of cattle raised with whom earn around contours approximately standard. From this, and after getting an image of the contour gradients are looking for the animal through an adjustment genetic and deformable contour, which has taken into account the changes -- to make the image of the insensitive rotations --- , relocations and changes of scale.

Author: PÉREZ TUDELA JULIO DOMINGO.
Year: 2005.
University: BARCELONA [www.ub.es].
Place of defense: FACULTAT DE FÍSICA.
Place of preparation: UNVIERSITAT DE BARCELONA.

Summary: Within the name "recognition of forms" covers all those methods by which it is possible to detect the presence of a certain image or reference within another picture or scene. Within the set of optical techniques, highlighting based on the comparison between the elements of the scene and the image that we want to detect. This comparative method, called optical correlation, exploits the properties of lenses and their ability to perform Fourier transform. The investigation has as its primary objective the study of the effects of these lenses cause aberrations in the recognition process. We have developed a method of calculating the wave aberration of an optical system with symmetry of revolution and an object timely located on the optical axis. This method is based on the exact route of lightning and the relationship between lightning and aberration of the wave aberration. It uses the theory of formation of image Abbe-Rayleigh, which brings together all intents and purposes difractivos in pupils in the system, either at the entrance or the exit, and geometrical optics used to describe the propagation of light between both. The reliability of the method has been observed experimentally using techniques inteferómetrias, and then compared the results for the wave aberration with those calculated by commercial software design and evaluation of optical systems. Once secured the proper functioning of the calculation algorithms, architectures have been analyzed two classic correlation: the correlator transformed joint monobanco and correlator converged Vander Lugt. The first of them has made an independent study of the influence of aberación each of the difractómeros that form at the end correlation, and from the results obtained proposes a method for optimizing the correlator. This study has also been conducted for the correlator Vander Lugt, with similar results. In the case of this second correlator has also studied the inclusion of information about the aberration of the two stages of correlator in the design of the filters. Finally, based on the above results, it is also proposed that a new configuration improved substantially the same as the conduct reconnaissance team of ways.

Author: INGO RUDOLF KECK.
Year: 2005.
University: GRANADA [www.ugr.es].
Place of defense: ESCUELA TÉCNICA SUPERIOR DE INGENIERÍAS INFORMÁTICAS Y DE TELECOMUNICACIÓN.
Place of preparation: ESCUELA TÉCNICA SUPERIOR DE INGENIERÍAS INFORMÁTICA Y DE TELECOMUNICACIÓN.

Summary: At present communications acquire ever more importance. In many real-world situations is working with the sending and receiving of information, which usually on the way to its destination suffers various distortions that corrupt the signal, resulting in the end a mixture of information that must be processed in order to obtain correct results. There are countless areas where the signal processing serves a variety of functions essential; these fields includes among others: the communication of data, voice and images, seismology, medicine, acoustics, sonar, instrumentation, robotics, and so on. The problem of separation blind signal is the recovery of the original signals from mixtures detected by sensors, knowing only the latter. These mixtures of signals occurring in the environment in which propagate and sensors and characteristics of this method is that a priori do not have any information of the original signals or the way they were mixed. A clear example that provides an overview of this problem is the well-known effect "Cocktail Party", which consists in the ability to possess the human being to perceive separate and a voice of noise or a set of voices speaking simultaneously. Once this separation, the voices recovered or individual sources, are known as "separate components." This doctoral work aims to use the tool of independent component analysis applied to the brain signals "fMRI." To understand how it works fMRI worth mentioning that the brains of persons is divided into two hemispheres (right and left), and each is divided into lobes which accomplish various tasks related to feel, hear, see, etc. . Every action, thought, sensation is produced by the activity of cells located in a specific area of the brain. When a group of cells are altered for any reason beginning to shoot, increasing the electric current, transmission and metabolism. This also leads to needing more blood because they need more energy. The activation of an area of the brain producing the receipt of my blood and that there was a dilation of the blood veins. Therefore, any action of the brain caused by an increase in blood at the point of the cortex which governs the action of magnetic resonance images can see the anatomy of the internal organs of the body. The functional magnetic resonance imaging are the result of the values of the signal strength that occurs with increasing blood.

Author: M.I. ODEH SUHAIL.
Year: 2005.
University: GRANADA [www.ugr.es].
Place of defense: ESCUELA TÉCNICA SUPERIOR DE INGENIERÍAS INFORMÁTICA Y DE TELECOMUNICACIÓN.
Place of preparation: ESCUELA TÉCNICA SUPERIOR DE INGENIERÍAS INFORMÁTICA Y DE TELECOMUNICACIÓN.

Summary: This work focuses on the development of automatic diagnostic schemes and flexible. This will explore various alternatives capable of efficiently using information from a group of cases "tagged" for the diagnosis of other new cases. Such diagnostic tools can be used for various pathologies. This work provided the development of an automated system for diagnosing diseases of skin-based image processing. The work uses images of fluoroscopy (containing information about the deep dermis) from patients with various diseases. In this dissertation is a study of different parameters extracted by image processing and assesses the importance of different features to the problem of diagnosis of skin cancer. To do so, rather than undertake a study looking conventional statistical correlations and levels of significance between the specifics, there is a modular automatic classification system (based on K-vecinas nearest, neural network and nuera-difusa) and is optimized with schemes combinatorial search (method forward and backward method) and genetic algorithms (search multi-camino). Thus we move a step beyond the simple study of significance of the features extracted by designing a system of automatic diagnosis of skin cancer based on images of fluoroscopy. In addition, the classification system and scheme optimization are general and constitute an application biomedical diagnostics that can be applied to other pathologies.

Author: KECK INGO RUDOLF.
Year: 2005.
University: GRANADA [www.ugr.es].
Place of defense: E.T.S. de Inge. Infor. y de Teleco..
Place of preparation: ESCUELA TÉCNICA SUPERIOR DE INGENIERÍAS INFORMÁTICA Y DE TELECOMUNICACIÓN.

Summary: At present communications acquire ever more importance. In many real-world situations is working with the sending and receiving of information, which usually on the way to its destination suffers various distortions that corrupt the signal, resulting in the end a mixture of information that must be processed in order to obtain correct results. There are countless areas where the signal processing serves a variety of functions essential; these fields includes among others: the communication of data, voice and images, seismology, medicine, acoustics, sonar, instrumentation, robotics, and so on. The problem of separation blind signal is the recovery of the original signals from mixtures detected by sensors, knowing only the latter. These mixtures of signals occurring in the environment in which propagate and sensors and feature of this method is that a priori do not have any information of the original signals or the way they were mixed. A clear example that provides an overview of this problem is the well-known effect 'Cocktail Party', which consists in the ability to possess the human being to perceive separate and a voice of noise or a set of voices speaking simultaneously. Once this separation, the voices recovered or individual sources, are called 'separate components.' This doctoral work aims to use the tool of independent component analysis applied to the brain signals' fMRI. To understand how it works fMRI worth mentioning that the brains of persons is divided into two hemispheres (right and left), and each is divided into lobes which accomplish various tasks related to feel, hear, see, etc. . Every action, thought, sensation is produced by the activity of cells located in a specific area of the brain. When a group of cells are altered for any reason beginning to shoot, increasing the electric current, transmission and metabolism. This also leads to needing more blood because they need more energy. The activation of an area of the brain producing the receipt of my blood and that is causing a dilation of the blood veins. Therefore, any action of the brain caused by an increase in blood at the point of the cortex which governs the action. The magnetic resonance imaging can see the anatomy of the internal organs of the body. The functional magnetic resonance imaging are the result of the values of the signal strength that occurs with increasing blood.

Author: M. I. ODEH SUHAIL.
Year: 2005.
University: GRANADA [www.ugr.es].
Place of defense: E.T.S. DE INGEN. INFOR. Y DE TELECO..
Place of preparation: ESCUELA TÉCNICA SUPERIOR DE INGENIERÍAS INFORMÁTICA Y DE TELECOMUNICACIÓN.

Summary: This work focuses on the development of automatic diagnostic schemes and flexible. This will explore various alternatives capable of efficiently using information from a group of cases' tagged 'for the diagnosis of other new cases. Such diagnostic tools can be used for various pathologies. In this paper we explored various alternatives and finally we used in a diagnostic application specific. The final chapters of this paper presents the development of an automated system for diagnosing diseases of skin-based image processing. The work uses images of fluoroscopy (containing information about the deep dermis) from patients with various diseases. In this dissertation is a study of different parameters extracted by image processing and assesses the importance of different features to the problem of diagnosis of skin cancer. To do so, rather than undertake a study looking conventional statistical correlations and levels of significance between the specifics, there is a modular automatic classification system (based on K-nearest neighbors, neural network and nuera-difusa) and is optimized with schemes combinatorial search (method forward and backward method) and genetic algorithms (search multi-camino). Thus we move a step beyond the simple study of significance of the features extracted by designing a system of automatic diagnosis of skin cancer based on images of fluoroscopy. In addition, the classification system and scheme optimization are general and constitute an application biomedical diagnostics that can be applied to other pathologies.

Author: Ramos Terrades Oriol.
Year: 2006.
University: AUTÓNOMA DE BARCELONA [www.uab.es].
Place of defense: Centro de Visión por Computador.
Place of preparation: Centro de Visión por Computador.

Summary: In the field of Document Analysis we would like to be able to automatically process any kind of digital document. This is a challenger problem that has motivated different lines of research in the field of Document Analysis at different levels. We have focused on the shape description and also on classifier fusion, to apply them to one of the application fields in the Document Analysis: the graphics recognition. In shape recognition, many applications have to face the problem of describing a large number of complex shapes for recognition or retrieval in large databases. One of the key issues is the design of highly discriminant shape descriptors. Unfortunately, one kind of descriptor is not usually enough to achieve satisfactory results and hence, we have to combine the information from different sources to improve the global performance of the recognition system. These theoretical approaches have been evaluated through an experimental evaluation in ridgelets descriptors, classifier fusion and applying the classifier fusion methods to ridgelets descriptors, obtaining quite satisfying results.