965 research outputs found
Sort by
Advanced Restoration Techniques for Images and Disparity Maps
Full text linkWith increasing popularity of digital cameras, the field of Computa-
tional Photography emerges as one of the most demanding areas of
research. In this thesis we study and develop novel priors and op-
timization techniques to solve inverse problems, including disparity
estimation and image restoration.
The disparity map estimation method proposed in this thesis incor-
porates multiple frames of a stereo video sequence to ensure temporal
coherency. To enforce smoothness, we use spatio-temporal connec-
tions between the pixels of the disparity map to constrain our solution.
Apart from smoothness, we enforce a consistency constraint for the
disparity assignments by using connections between the left and right
views. These constraints are then formulated in a graphical model,
which we solve using mean-field approximation. We use a filter-based
mean-field optimization that perform efficiently by updating the dis-
parity variables in parallel. The parallel updates scheme, however, is
not guaranteed to converge to a stationary point. To compare and
demonstrate the effectiveness of our approach, we developed a new
optimization technique that uses sequential updates, which runs ef-
ficiently and guarantees convergence. Our empirical results indicate
that with proper initialization, we can employ the parallel update
scheme and efficiently optimize our disparity maps without loss of
quality. Our method ranks amongst the state of the art in common
benchmarks, and significantly reduces the temporal flickering artifacts
in the disparity maps.
In the second part of this thesis, we address several image restora-
tion problems such as image deblurring, demosaicing and super-
resolution. We propose to use denoising autoencoders to learn an
approximation of the true natural image distribution. We parametrize
our denoisers using deep neural networks and show that they learn
the gradient of the smoothed density of natural images. Based on
this analysis, we propose a restoration technique that moves the so-
lution towards the local extrema of this distribution by minimizing
the difference between the input and output of our denoiser. Weii
demonstrate the effectiveness of our approach using a single trained
neural network in several restoration tasks such as deblurring and
super-resolution. In a more general framework, we define a new
Bayes formulation for the restoration problem, which leads to a more
efficient and robust estimator. The proposed framework achieves state
of the art performance in various restoration tasks such as deblurring
and demosaicing, and also for more challenging tasks such as noise-
and kernel-blind image deblurring.
Keywords. disparity map estimation, stereo matching, mean-field
optimization, graphical models, image processing, linear inverse prob-
lems, image restoration, image deblurring, image denoising, single
image super-resolution, image demosaicing, deep neural networks,
denoising autoencoder
Cinemeducation: Entwicklung eines E-Learnings zur Anamneseerhebung
Full text linkFür die Medizinstudierenden des ersten Jahreskurses wurde ein ILIAS-Lernmodul entwickelt, welches sich mit der Einführung in die Anamneseerhebung befasst. Hierbei wurde mit Mitteln der Cinemeducation, also der Verwendung von Filmen und Serien mit medizinischen Inhalten, gearbeitet.
Das Lernmodul beinhaltet Videoclips, welche durch Theorieabschnitte ergänzt werden. Die vermittelten Inhalte werden regelmässig mit Multiple Choice Fragen abgefragt. Es entstand ein E-Learning, welches die Studierenden auf eine neue Weise an die Grundlagen des medizinischen Interviews heranführt.
Das Lernmodul steht ab dem Herbstsemester 2017 den Medizinstudieren zur Verfügung und wird ab dem Frühjahrssemester 2018 testatpflichtig in das Curriculum integriert. Die weiterführende Betreuung des Lernmoduls übernimmt das Berner Institut für Hausarztmedizin BIHAM
Technology Diffusion News Shocks. An Empirical and Theoretical Analysis of Anticipated and Unanticipated Productivity Shocks
Full text linkAugmenting Type Inference with Lightweight Heuristics
Full text linkStatic type information facilitates program comprehension and analysis. Yet, such information is absent in dynamically-typed languages, and that increases the time needed for software maintenance. Type inference algorithms may provide type information to developers, but in order to be fast and assist in development phase, they sacrifice the precision for speed. One of the biggest obstacles for their precision is polymorphism presence.
In this thesis we first analyse the prevalence of polymorphism in object-oriented software, to assess the criticality it imposes on simple type inference. We find that polymorphism is omnipresent in object-oriented code, and that static analysis in dynamically-typed languages is also hampered by the usage of cross-hierarchy polymorphism, i.e., duck typing. As this big obstacle for static code analysis cannot be bypassed, we propose the need for lightweight heuristics to tackle the problem of imprecision of simple type inference algorithms.
Four lightweight heuristics are employed to improve the performance of two simple and fast type inference approaches. These heuristics are founded on the source code and run-time information that are easy to collect without interrupting the workflow of a developer.
The heuristics are evaluated and compared with the underlying algorithms based on their inference time and precision. All of them show a significant improvement when compared to the basic algorithm. They introduce a negligible overhead on the inference time, thus we deem them usable during regular coding tasks
On the spectra of mapping classes and the 4-genera of positive knots
Full text linkRoughly, this thesis can be divided into three parts.
In the first part, we study the Galois conjugates of the dilatation of pseudo-Anosov mapping classes. In particular, for a product of two multitwists, we show that all Galois conjugates are either real and positive or contained in the unit circle and the positive real axis, depending on whether the products are of opposite or of the same sign. Furthermore, for each closed orientable surface, we determine the minimal dilatation among pseudo-Anosov mapping classes arising from Penner's construction.
In the second part, we consider the Alexander polynomial and the signature function of links. For a Murasugi sum of two Seifert surfaces with symmetric, definite Seifert form, we show that all zeroes of the Alexander polynomial are either real and positive or contained in unit circle and the negative real axis, depending on whether the Seifert forms are definite of opposite or the same sign. Furthermore, we prove that the signature function of a Murasugi sum of two Seifert surfaces with symmetric, definite Seifert form is monotonic. We also show that the signature of a positive arborescent Hopf plumbing is greater than or equal to two thirds of the first Betti number.
In the third part, we study the topological four-genus of positive braid knots. We show that the difference of the ordinary Seifert genus and the topological four-genus grows at least linearly with the positive braid index. In particular, we show that the positive braid knots for which the topological four-genus equals the ordinary Seifert genus are exactly the positive braid knots with maximal signature invariant
Improvement of Space Debris Orbits
Full text linkThe increasing amount of space debris requires huge efforts for the tracking networks in order to maintain the orbits of all the objects. The precise knowledge of the positions of space debris objects is fundamental for collision avoidance maneuvers performed by satellite operators and for future active debris removal missions. It is very well known that the accuracy of an orbit determination process depends on the kind of observables used, their accuracy, the length of the observed arc, the number of observations, the observer orbit, and the observer-target geometry of the observations. In this thesis an in-depth study is carried out to understand how the mentioned parameters influence the orbit determination accuracy and how we can improve the quality of the estimated orbits. After a brief introduction on the least squares adjustment algorithm and on the way of propagating and manipulating the resulting covariance matrix, we will focus essentially on the influence of the object-observer relative geometry and of the use of different observables for the orbit determination of space debris. The object-observer relative geometry is approached as an information gain problem and it is studied using simulations and covariance analysis. The main aim of the covariance analysis is to identify the optimal follow-up strategy as a function of the object-observer geometry, the interval between follow-up observations and the shape of the orbit. This analysis is applied to every orbital regime but particular attention is dedicated to the highly-populated space debris orbits. Furthermore, several particular cases are analyzed with this method: more than two
follow-ups observations, the influence of a second observing station, and the introduction of the distance as second observable. The study on the combination of different types of observables is carried out investigating the influence of addition of the laser range measurements to the classical optical astrometric observations in terms of improved accuracy of the determined orbit. In particular, after some validation tests to prove the effectiveness of the implemented algorithm, it will be shown how different kinds of observables influence the accuracies of the estimated orbital parameters. Then, the influence of the observation geometry is analyzed and finally the improvements achieved on the orbit prediction, for different orbital regimes, will be shown. All the mentioned tests are performed using real ranges from the International Laser Ranging Service (ILRS) stations and real angular/laser measurements provided by sensors of the Swiss Optical Ground Station and Geodynamics Observatory Zimmerwald owned by the Astronomical Institute of the University of Bern (AIUB)
Performing Hamlet in modern Iran (1900-2012)
Full text linkThe interest in the reception of Shakespeare beyond the borders of Britain has always been great, and scholarly writings on the issue have been very extensive. However, there are very few research projects focusing on the aspects of this reception in a country with a totally different cultural, political and social setting. It is well known to scholars in performance studies that local context could strongly influence a play's staging and interpretation. The socio-political situation and the influence of the dominant political powers on art are among the most decisive determinants of the context. When in 1932 the Shah invited a Russian-Armenian Hamlet to perform on stage in Iran, intellectuals and reformists attached great expectations to a “Hamlet” performance as a vehicle for fostering progress of modern theatre and facilitating modernisation. In the meantime, the state, as mobiliser of this phenomenon into the country, had its own political intentions. Since that date any production of “Hamlet” deals with a dynamic cultural and social exchange. This research aims at investigating this cultural mobility and its effect in the history of modern Iran. Iran is a country with a century-long history of performing "Hamlet" under three different authoritarian political regimes. The research tries to find out why Western theatre had always been an important and critical subject for Iran’s political systems, and what happened to “Hamlet” while passing cultural borders and dealing with impediments of the destination country.
The evolution of Western drama from the cycles of mystery and miracle plays is well known. Less well understood is the parallel development in Iran. By the late 19th century, the mystery play, “Taziya” was on the brink of giving birth to a secular Iranian drama. However, due to the turbulent history of the Constitutional Revolution at the beginning of the 20th century and the fundamental social and political changes in the big towns of Iran, “Taziya” lost royal and upper-class patronage. From the middle of the 19th century onward, the production of Western dramas was encouraged. Iranians had their first glimpse at Shakespeare through a translation of “The Taming of the Shrew” in 1900, and since then Shakespeare absorbed significant attention of Iranian elites who presumed theatre as the best instrument for importing modern culture to Iranian society. Shakespeare's importance in view of the Constitutional Revolution is, to some extent, that this constitutional period can be called Shakespeare period. Among all of Shakespeare's translated works, “Hamlet” received the widest attention in modern Iranian theatre.
The victory of the Islamic Revolution was followed by enthusiastic efforts aimed at transforming this very Western art of drama, into a fully local form of art based on the new revolutionary culture and values. There is no doubt that every major social event, particularly cultural and political revolutions are followed by their own specific culture, literature and art. After the initial onset of the Islamic Revolution, more Farsi translations and adaptions of “Hamlet” have appeared than of any other Shakespeare's works. Hamlet's nature, as persona, is of such fluidity that it enables him to conform to diverse circumstances. With significant growth in the use of symbolism and signs in theatrical performances, “Hamlet” turned out to perform as the best metaphor of the current situation. With the help of a descriptive research method my research tries to clarify the circulation of “Hamlet” from text to performance on Iranian stages and the role of agencies in this transportation. Based on qualitative data collection, interviews and analysis of the theory of cultural mobility and semiotics, four effective elements are being analysed: Religion, Power, Gender and Agency. The research will be narrowed by Case Study of nine highly relevant “Hamlet” productions in the historical epoch of 1900 to 2012
Interwoven Lovers: Safavid Narrative Silks Depicting Characters from the Khamsa
Full text linkIn the twelfth century, Persian language poet Niẓāmī Ganjavī (1141-1209) authored a quintet of epic poems on five subjects, known as the Khamsa or “Five Treasures.” Two of Nizami’s works in the collection are romances based on legendary figures whose tales were grounded in historical events: the love story of Sasanian king Khusrau Parvīz and Armenian princess Shīrīn, and that of Bedouin beauty Laylā and her admirer Qays, aka “Majnūn” [Ar. “madman”]. In the centuries following Nizami’s codification of the characters, several illustrated manuscripts of the Khamsa were produced by workshops for the ruling classes throughout greater Iran, some by later poets who composed their own Khamsa manuscripts. By the turn of the sixteenth century, scenes representing a few pivotal events in the respective narratives had become part of the cycle of illustration, and well known among the educated elite.
Between 1550 and 1650, silk luxury textiles depicting these scenes were produced, possibly representing royal as well as independent textile workshop manufacture. A group of eleven different signed and unsigned textile designs depicting scenes from Khusrau and Shīrīn and Laylā and Majnūn are the basis of this study. The textiles will be examined alongside contemporary Khamsa manuscript illustrations, evaluated as a group and individually, and analyzed for their iconological properties based on patronage and consumerism
Data sources:
BORIS Portal Bern Open Repository and I...BAF-Online: Proceedings of the Berner Al...HASBonline – Hefte zur Archäologie des M...Informationswissenschaft: Theorie, Metho...Jahrbuch Diakonie SchweizJournal of Eye Movement ResearchJournal of Radiation Oncology Informatic...BOP SerialsVersants. Revista suiza de literaturas r...
BORIS Theses
BORIS ThesesAccess Repository Dashboard
Do you manage BORIS Theses? Access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard!