Iterative global and local methods of image restoration

Most of the reconstruction algorithms smooth out the edges of the reconstructed images but iterative total variation regularization and local adaptive methods preserve the edge information without any prior knowledge about the image geometric details. The restoration performance of the proposed algorithms for image deblurring is compared in terms of the mean squared error. Computer simulation results are provided and discussed.     

Reinforcement learning to adjust parametrized motor primitives to new situations

Humans manage to adapt learned movements very quickly to new situations by generalizing learned behaviors from similar situations. In contrast, robots currently often need to re-learn the complete movement. In this paper, we propose a method that learns to generalize parametrized motor plans by adapting a small set of global parameters, called meta-parameters. We employ reinforcement learning to learn the required meta-parameters to deal with the current situation, described by states. We introduce an appropriate reinforcement learning algorithm based on a kernelized version of the reward-weighted regression. To show its feasibility, we evaluate this algorithm on a toy example and compare it to several previous approaches. Subsequently, we apply the approach to three robot tasks, i.e., the generalization of throwing movements in darts, of hitting movements in table tennis, and of throwing balls where the tasks are learned on several different real physical robots, i.e., a Barrett WAM, a BioRob, the JST-ICORP/SARCOS CBi and a Kuka KR?6.     

Die Selbsterregung von Schwingungen

Zusammenfassung Das Problem der Selbsterregung von Schwingungen wird in vollster Allgemeinheit von der rechnerischen Seite in Angriff genommen. An Hand der Differentialgleichungen für die Rückkopplung und für den Induktionsgenerator wird eine bestimmte Klasse von Differentialgleichungen — die Hillsche — als für die Selbsterregung in Frage kommend abgegrenzt. Es wird an Hand der Arbeiten von A. Erdelyi ein Bericht über den Stand der mathematischen Theorie gegeben.Die anschauliche Ausdeutung der mathematischen Ergebnisse ergibt: Wesentlich für eine Selbsterregung ist eine Kompensation der DÄmpfung (Rückkopplung, Selbststeuerung, fallende Kennlinie). Die Kompensation erfolgt durch die Einwirkung der Schwankungen von charakteristischen Parametern des Kreises (L, R, C); die van der Polsche Theorie wird erweitert und modifiziert. Ist der Einflu\ der Schwankungen schwach, so erfolgt eine Modulation (Amplituden, Phasen-Frequenzmodulation). Ist der Einflu\ stark, so kann unter genau bestimmten UmstÄnden eine Mitnahme eintreten. Die Mitnahme wird von einer zusÄtzlichen Frequenzmodulation verzerrt. Der Resonanzfall der Mitnahme ist schlie\lich die Selbsterregung, für die die zusÄtzliche Frequenzmodulation verschwindet. Hat die Parameterschwankung starke Oberschwingungen, so kann eine Mitnahme oder Selbsterregung in diesen Oberwellen stattfinden.Der Kern der quantitativen Formulierung bildet ein StabilitÄtskriterium, welches Frequenzen und Amplitude der entstehenden Schwingung in ein festes VerhÄltnis setzt. Die StabilitÄt einer Schwingung kann genau errechnet werden.     

Comparison of different failure criteria for fiber-reinforced plastics in terms of fracture curves for arbitrary stress combinations

For fiber-reinforced plastics exists a big number of different criteria for the failure prediction. The intention of this paper is to compare the TSAI-HIL-, the LaRC04- and PUCK’s criterion in terms of their fracture curves for a unidirectional glass-fiber reinforced composite layer. Therefore after the implementation of these three criteria, the two-dimensional fracture curves for all possible stress combinations, which can be derived from a general spatial stress tensor, are computed. In this way, the characteristics of the criteria, similarities and differences and possible weak points become obvious.     

Comparison of 2D simultaneous multi-slice and 3D GRASE readout schemes for pseudo-continuous arterial spin labeling of cerebral perfusion at 3 T

Objective

In this perfusion magnetic resonance imaging study, the performances of different pseudo-continuous arterial spin labeling (PCASL) sequences were compared: two-dimensional (2D) single-shot readout with simultaneous multislice (SMS), 2D single-shot echo-planar imaging (EPI) and multishot three-dimensional (3D) gradient and spin echo (GRASE) sequences combined with a background-suppression (BS) module.

Materials and methods

Whole-brain PCASL images were acquired from seven healthy volunteers. The performance of each protocol was evaluated by extracting regional cerebral blood flow (rCBF) measures using an inline morphometric segmentation prototype. Image data postprocessing and subsequent statistical analyses enabled comparisons at the regional and sub-regional levels.

Results

The main findings were as follows: (i) Mean global CBF obtained across methods was were highly correlated, and these correlations were significantly higher among the same readout sequences. (ii) Temporal signal-to-noise ratio and gray-matter-to-white-matter CBF ratio were found to be equivalent for all 2D variants but lower than those of 3D-GRASE.

Discussion

Our study demonstrates that the accelerated SMS readout can provide increased acquisition efficiency and/or a higher temporal resolution than conventional 2D and 3D readout sequences. Among all of the methods, 3D-GRASE showed the lowest variability in CBF measurements and thus highest robustness against noise.

     

Copolymerization of ethylene with vinyl acetate. II. Effect of solvent on polymer yield and molecular weight

In the free-radical solution copolymerization of ethylene and vinyl acetate, polymer yield and molecular weight were found to vary considerably with the solvent used. Solvents were toluene, benzene, hexane, heptane, and cyclohexane. Molecular weights were highest in benzene and lowest in toluene and were readily explained by different solvent chain-transfer constants. Polymer yields were highest in aliphatic solvents and lowest in aromatic solvents including benzene. This was attributed to different types of interactions of the solvents with radical species. The saturated aliphatic solvents undergo hydrogen abstraction reactions, but these give reactive alkyl radicals which reinitiate polymer growth. Toluene also undergoes hydrogen abstraction reactions, but the resultant benzyl radical is resonance stabilized and does not readily reinitiate polymerization. Benzene does not undergo hydrogen abstraction reactions. The low yields are attributed to complex formation. A consideration of kinetic theory indicates that complex formation with both initiator and growing polymer radicals is involved. Differences in viscosity, ethylene solubility, and initiator half-life in the different solvents, as well as induced decomposition of the initiator, were not determining factors.     

Image restoration with sliding sinusoidal transforms

A space-variant restoration technique with sliding sinusoidal transforms is presented. Minimum mean-square error estimators in the domains of sliding discrete sine and cosine transforms for image restoration are derived. To achieve image processing in real time, fast recursive algorithms for computing the sliding sinusoidal transforms are utilized. Computer simulation results using a real image are provided and compared with that of common restoration techniques. The text was submitted by the authors in English. Vitaly Kober obtained his MS degree in Applied Mathematics from the Air-Space University of Samara (Russia) in 1984, his PhD degree in 1992, and Doctor of Sciences degree in 2004 in Image Processing from the Institute of Information Transmission Problems, Russian Academy of Sciences. Now he is a titular researcher at CICESE, México. His research interests include signal and image processing, pattern recognition. Iosif A. Ovseyevich graduated from the Moscow Electrotechnical Institute of Telecommunications. Received Candidate’s degree in 1953 and Doctor’s degree in information theory in 1972. At present he is Emeritus Professor at the Institute of Information Transmission Problems of the Russian Academy of Sciences. His research interests include information theory, signal processing, and expert systems. He is a Member of IEEE, Popov Radio Society.     

Pulmonary thromboembolism

Pulmonary embolism (PE) is a serious complication of chronic obstructive pulmonary disease (COPD). Retrospective studies on patients with COPD treated in the intensive care unit (ICU) were performed to determine: 1) the frequency of PE; 2) the clinical course of PE in cases of COPD in the ICU; and 3) the frequency of PE as a cause of death in the studied group. The frequency of PE was 10.9% in COPD patients. In the group analysed, clinical presentation of PE was characterized by acute severe, life-threatening complications leading to death in 86.7% of cases. PE was the most frequent cause of death (40.6%) in COPD patients in the ICU. The results of treatment of pulmonary embolism in chronic obstructive pulmonary disease are poor and mortality in this group of patients is very high. We believe that improvement of management can be achieved by antithromboembolic prophylaxis, which should be instituted as soon as possible in all patients with chronic obstructive pulmonary disease in the intensive care unit.