The main objective of this paper is to develop a method of removing noise from signal. This method is based on the radial-basis function networks and the principle of cross-validation in statistics. In this method, we detect noise by estimating the magnitude of validation error after training the network. Besides, this paper applies the concept of predictive coding to select data set from image when the proposed method used to deal with the noise removal problem of two-dimensional image signals. Finally, the proposed method has been employed to deal with noise removal problems of one-dimensional and two-dimensional signals. From the result of simulation, the proposed method could remove noise from signals effectively.

Η διπλωματική εργασία μου ασχολείται με την αφαίρεση θορύβου από Ηλεκτροκαρδιογράφημα(ΗΚΓ), με σκοπό τη καθαρότερη λήψη του σήματος για διαγνωστικούς σκοπούς. Ο θόρυβος στο ΗΚΓ προέρχεται από διάφορες πηγές, είτε τεχνικές, είτε φυσιολογικές. Η αφαίρεσή του είναι ουσιαστικής σημασίας. Στόχος της εργασίας αυτής είναι η παράθεση των κυρίων μεθόδων αφαίρεσης θορύβου από ΗΚΓ και του τρόπου λειτουργίας τους, καθώς και η παράθεση παραδειγμάτων από την πρόσφατη βιβλιογραφία. / --

Στην παρούσα διπλωματική εργασία ασχοληθήκαμε με τον μη επεμβατικό προσδιορισμό του κορεσμού οξυγόνου στο αίμα με τη χρήση παλμικού οξύμετρου (ή πιο απλά σφυγμοξύμετρου) και ειδικότερα με την αφαίρεση θορύβου από το πληθυσμιογραφικό σήμα του. Αρχικά, γίνεται μια εισαγωγή στη βιοϊατρική τεχνολογία, στις εφαρμογές και τις υπηρεσίες που αυτή παρέχει στον τομέα της υγείας στον 21ο αιώνα έτσι ώστε να καταστεί σαφής η χρησιμότητα της βιοϊατρικής τεχνολογίας σε ευρύτατους τομείς της ζωής μας. Στη συνέχεια γίνεται μια πρώτη γνωριμία με το σφυγμοξύμετρο εισάγοντας έννοιες και μαθηματικούς νόμους που περιγράφουν τις γενικές αρχές σχεδίασής του. Έπειτα, παραθέτουμε αναλυτικά όλα τα είδη θορύβου τα οποία μπορούν να θέσουν την ακρίβεια των μετρήσεων του σφυγμοξύμετρου υπό αίρεση καθώς και μια πληθώρα απλών και σύνθετων μεθόδων για την αφαίρεση θορύβου και τον ακριβή προσδιορισμό του κορεσμού οξυγόνου. Τέλος, παρουσιάζεται αναλυτικά ο προσαρμοστικός ακυρωτής θορύβου που προτείνεται από την εταιρεία Masimo Corporation αλλά και η υπολογιστική μέθοδος καθορισμού της τιμής του κορεσμού οξυγόνου που προτείνεται από την εταιρεία Texas Instruments. Τα ανωτέρω υλοποιούνται σε ένα ενιαίο υπολογιστικό πρόγραμμα σε γλώσσα προγραμματισμού C που εκτελεί προσαρμοστική ακύρωση θορύβου και υπολογίζει τον κορεσμό οξυγόνου και το οποίο εφαρμόζεται σε πέντε διαφορετικά σετ μετρούμενων σημάτων σφυγμοξύμετρου. / -

Baigiamojo magistro darbo tikslas – aprašyti ir ištirti šiuolaikinę atvirojo kodo garso apdorojimo programinę įrangą, jos privalumus ir trūkumus. Darbe apžvelgiami pagrindiniai garso apdorojimo uždaviniai. Trumpai aprašius keletą populiariausių garso apdorojimo programų, darbe išsamiai nagrinėjama atvirojo kodo programa Audacity. Analizuojamos šios sistemos garso redagavimo galimybės, pateikiama daug iliustracijų, įvairių garso redagavimo ir analizės pavyzdžių. Aprašomi su Audacity atlikti įvairūs garso redagavimo eksperimentai. Taip pat aprašomas atliktas triukšmo įtakos žodžių atpažinimo tikslumui tyrimas. Pateikiamos darbo išvados, literatūros ir interneto šaltinių sąrašas. Darbo apimtis – 82 psl. be priedų. / The basic task of the final master paper was to describe and to explore the modern open source sound processing software, to investigate its advantages and disadvantages. Basic audio processing tasks are reviewed in the paper. After the short description of the several popular audio processing programs, one of them – audio editor Audacity – is investigated thorougly. The main audio editing facilities of this program are described with a lot of illustrations, many examples of audio editting and analysis are presented in the paper. Various experiments of sound processing with Audacity were performed and are described here. The analysis of noise influence to word recognition accuracy was done. Conclusions, references to literature and other sources on internet are also presented The paper consists of 82 pages without enclosures.

Um dos problemas da extração de dados na web é a remoção de ruído existente nas páginas. Esta tarefa busca identificar todos os elementos não informativos em meio ao conteúdo, como por exemplo cabeçalhos, menus ou propagandas. A presença de ruído pode prejudicar seriamente o desempenho de motores de busca e tarefas de mineração de dados na web. Este trabalho aborda o problema da descoberta de ruído em páginas da web oculta, a parte da web que é acessível apenas através do preenchimento de formulários. No processamento da web oculta, a extração de dados geralmente é precedida por uma etapa de inserção de dados, na qual os formulários que dão acesso às páginas ocultas são automaticamente ou semi-automaticamente preenchidos. Durante esta fase, são coleta- dos dados do domínio em questão, como os rótulos e valores dos campos. A proposta deste trabalho é agregar este tipo de dados com informações sintáticas dos elementos que compõem a página. É mostrado empiricamente que esta combinação atinge resultados melhores que uma abordagem baseada apenas em informações sintáticas. / One of the problems of data extraction from web pages is the identification of noise in pages. This task aims at identifying non-informative elements in pages, such as headers, menus, or advertisement. The presence of noise may hinder the performance of search engines and web mining tasks. In this paper we tackle the problem of discovering noise in web pages found in the hidden web, i.e., that part of the web that is only accessible by filling web forms. In hidden web processing, data extraction is usually preceeded by a form filling step, in which the query forms that give access to the hidden web pages are automatically or semi-automatically filled. During form filling relevant data about the queried domain are collected, as field names and field values. Our proposal combines this type of data with syntactic information about the nodes that compose the page. We show empirically that this combination achieves better results than an approach that is based solely on syntactic information. Keywords:

Um dos problemas da extração de dados na web é a remoção de ruído existente nas páginas. Esta tarefa busca identificar todos os elementos não informativos em meio ao conteúdo, como por exemplo cabeçalhos, menus ou propagandas. A presença de ruído pode prejudicar seriamente o desempenho de motores de busca e tarefas de mineração de dados na web. Este trabalho aborda o problema da descoberta de ruído em páginas da web oculta, a parte da web que é acessível apenas através do preenchimento de formulários. No processamento da web oculta, a extração de dados geralmente é precedida por uma etapa de inserção de dados, na qual os formulários que dão acesso às páginas ocultas são automaticamente ou semi-automaticamente preenchidos. Durante esta fase, são coleta- dos dados do domínio em questão, como os rótulos e valores dos campos. A proposta deste trabalho é agregar este tipo de dados com informações sintáticas dos elementos que compõem a página. É mostrado empiricamente que esta combinação atinge resultados melhores que uma abordagem baseada apenas em informações sintáticas. / One of the problems of data extraction from web pages is the identification of noise in pages. This task aims at identifying non-informative elements in pages, such as headers, menus, or advertisement. The presence of noise may hinder the performance of search engines and web mining tasks. In this paper we tackle the problem of discovering noise in web pages found in the hidden web, i.e., that part of the web that is only accessible by filling web forms. In hidden web processing, data extraction is usually preceeded by a form filling step, in which the query forms that give access to the hidden web pages are automatically or semi-automatically filled. During form filling relevant data about the queried domain are collected, as field names and field values. Our proposal combines this type of data with syntactic information about the nodes that compose the page. We show empirically that this combination achieves better results than an approach that is based solely on syntactic information. Keywords:

Um dos problemas da extração de dados na web é a remoção de ruído existente nas páginas. Esta tarefa busca identificar todos os elementos não informativos em meio ao conteúdo, como por exemplo cabeçalhos, menus ou propagandas. A presença de ruído pode prejudicar seriamente o desempenho de motores de busca e tarefas de mineração de dados na web. Este trabalho aborda o problema da descoberta de ruído em páginas da web oculta, a parte da web que é acessível apenas através do preenchimento de formulários. No processamento da web oculta, a extração de dados geralmente é precedida por uma etapa de inserção de dados, na qual os formulários que dão acesso às páginas ocultas são automaticamente ou semi-automaticamente preenchidos. Durante esta fase, são coleta- dos dados do domínio em questão, como os rótulos e valores dos campos. A proposta deste trabalho é agregar este tipo de dados com informações sintáticas dos elementos que compõem a página. É mostrado empiricamente que esta combinação atinge resultados melhores que uma abordagem baseada apenas em informações sintáticas. / One of the problems of data extraction from web pages is the identification of noise in pages. This task aims at identifying non-informative elements in pages, such as headers, menus, or advertisement. The presence of noise may hinder the performance of search engines and web mining tasks. In this paper we tackle the problem of discovering noise in web pages found in the hidden web, i.e., that part of the web that is only accessible by filling web forms. In hidden web processing, data extraction is usually preceeded by a form filling step, in which the query forms that give access to the hidden web pages are automatically or semi-automatically filled. During form filling relevant data about the queried domain are collected, as field names and field values. Our proposal combines this type of data with syntactic information about the nodes that compose the page. We show empirically that this combination achieves better results than an approach that is based solely on syntactic information. Keywords:

京都大学 / 0048 / 新制・課程博士 / 博士(工学) / 甲第18280号 / 工博第3872号 / 新制||工||1594(附属図書館) / 31138 / 京都大学大学院工学研究科電気工学専攻 / (主査)教授 小林 哲生, 教授 中村 裕一, 准教授 古谷 栄光 / 学位規則第4条第1項該当 / Doctor of Philosophy (Engineering) / Kyoto University / DFAM

This thesis presents online 3D reconstruction and ground segmentation using unmanned aerial vehicle (UAV) based stereo vision. For this purpose, a long baseline stereo vision system has been designed and built. Application of this system is to work as part of an air and ground based multi-robot autonomous terrain surveying project at Unmanned Systems Lab (USL), Virginia Tech, to act as a first responder robotic system in disaster situations. Areas covered by this thesis are design of long baseline stereo vision system, study of stereo vision raw output, techniques to filter out outliers from raw stereo vision output, a 3D reconstruction method and a study to improve running time by controlling the density of point clouds. Presented work makes use of filtering methods and implementations in Point Cloud Library (PCL) and feature matching on graphics processing unit (GPU) using OpenCV with CUDA. Besides 3D reconstruction, the challenge in the project was speed and several steps and ideas are presented to achieve it. Presented 3D reconstruction algorithm uses feature matching in 2D images, converts keypoints to 3D using disparity images, estimates rigid body transformation between matched 3D keypoints and fits point clouds. To correct and control orientation and localization errors, it fits re-projected UAV positions on GPS recorded UAV positions using iterative closest point (ICP) algorithm as the correction step. A new but computationally intensive process of use of superpixel clustering and plane fitting to increase resolution of disparity images to sub-pixel resolution is also presented. Results section provides accuracy of 3D reconstruction results. The presented process is able to generate application acceptable semi-dense 3D reconstruction and ground segmentation at 8-12 frames per second (fps). In 3D reconstruction of an area of size 25 x 40 m2, with UAV flight altitude of 23 m, average obstacle localization error and average obstacle size/dimension error is found to be of 17 cm and 3 cm, respectively. / MS / This thesis presents near real-time, called online, visual reconstruction in 3-dimensions (3D) using ground facing camera system on an unmanned aerial vehicle. Another result of this thesis is separating ground from obstacles on the ground. To do this the camera system using two cameras, called stereo vision system, with the cameras being positioned comparatively far away from each other at 60 cm was designed as well as an algorithm and software to do the visual 3D reconstruction was developed. Application of this system is to work as part of an air and ground based multi-robot autonomous terrain surveying project at Unmanned Systems Lab, Virginia Tech, to act as a first responder robotic system in disaster situations. Presented work makes use of Point Cloud Library and library functions on graphics processing unit using OpenCV with CUDA, which are popular Computer Vision libraries. Besides 3D reconstruction, the challenge in the project was speed and several steps and ideas are presented to achieve it. Presented 3D reconstruction algorithm is based on feature matching, which is a popular way to mathematically identify unique pixels in an image. Besides using image features in 3D reconstruction, the algorithm also presents a correction step to correct and control orientation and localization errors using iterative closest point algorithm. A new but computationally intensive process to improve resolution of disparity images, which is an output of the developed stereo vision system, from single pixel accuracy to sub-pixel accuracy is also presented. Results section provides accuracy of 3D reconstruction results. The presented process is able to generate application acceptable 3D reconstruction and ground segmentation at 8-12 frames per second. In 3D reconstruction of an area of size 25 x 40 m2 , with UAV flight altitude of 23 m, average obstacle localization error and average obstacle size/dimension error is found to be of 17 cm and 3 cm, respectively.

During the acquisition and transmission of images, it is important that the information is retained with the highest quality. Occasionally noise from various sources can corrupt an image. In this case, various image processing filtering techniques that are effective in removing noise can be used. Noise removal filters are designed to remove specific types of noise. However, they also degrade the image and detail is lost. This is particularly troublesome in cases where there is very little noise. A number of detail preserving filters have been proposed in the past, but most of the work was focused on monochrome images. With today's technological advancements, most of the images used are in full color, and detail preserving filters that have been designed for monochrome, cannot be directly applied for color. In this work, several detail preserving filters that have been designed to remove impulsive noise from color images are first surveyed and compared. We then consider the extension of detail preserving median filters to color images. At the end, we propose new filters that are capable of removing colored impulsive noise with minimum image degradation. Two user adjustable parameters can vary the strength of the filter for best results. Several comparisons with other previously proposed filters will be presented. Finally, in order to make the filter user friendly, an Adobe Photoshop plug-in is to be developed.