Classificazione di oggetti in immagini attraverso il modello Bag of Visual Words

Generic object recognition is an important function of the human visual system and everybody finds it highly useful in their everyday life. For an artificial vision system it is a really hard, complex and challenging task because instances of the same object category can generate very different images, depending of different variables such as illumination conditions, the pose of an object, the viewpoint of the camera, partial occlusions, and unrelated background clutter. The purpose of this thesis is to develop a system that is able to classify objects in 2D images based on the context, and identify to which category the object belongs to. Given an image, the system can classify it and decide the correct categorie of the object. Furthermore the objective of this thesis is also to test the performance and the precision of different supervised Machine Learning algorithms in this specific task of object image categorization. Through different experiments the implemented application reveals good categorization performances despite the difficulty of the problem. However this project is open to future improvement; it is possible to implement new algorithms that has not been invented yet or using other techniques to extract features to make the system more reliable. This application can be installed inside an embedded system and after trained (performed outside the system), so it can become able to classify objects in a real-time. The information given from a 3D stereocamera, developed inside the department of Computer Engineering of the University of Bologna, can be used to improve the accuracy of the classification task. The idea is to segment a single object in a scene using the depth given from a stereocamera and in this way make the classification more accurate.

Numerical and semi-analytical models of sliding masses: application to the 1783 Scilla tsunamigenic landslide

The Scilla rock avalanche occurred on 6 February 1783 along the coast of the Calabria region (southern Italy), close to the Messina Strait. It was triggered by a mainshock of the Terremoto delle Calabrie seismic sequence, and it induced a tsunami wave responsible for more than 1500 casualties along the neighboring Marina Grande beach. The main goal of this work is the application of semi-analtycal and numerical models to simulate this event. The first one is a MATLAB code expressly created for this work that solves the equations of motion for sliding particles on a two-dimensional surface through a fourth-order Runge-Kutta method. The second one is a code developed by the Tsunami Research Team of the Department of Physics and Astronomy (DIFA) of the Bologna University that describes a slide as a chain of blocks able to interact while sliding down over a slope and adopts a Lagrangian point of view. A wide description of landslide phenomena and in particular of landslides induced by earthquakes and with tsunamigenic potential is proposed in the first part of the work. Subsequently, the physical and mathematical background is presented; in particular, a detailed study on derivatives discratization is provided. Later on, a description of the dynamics of a point-mass sliding on a surface is proposed together with several applications of numerical and analytical models over ideal topographies. In the last part, the dynamics of points sliding on a surface and interacting with each other is proposed. Similarly, different application on an ideal topography are shown. Finally, the applications on the 1783 Scilla event are shown and discussed.

Synthesis and Characterization of Double Metal Hexacyanoferrates

Manganese Hexacyanoferrate (MnHCF) and nickel doped manganese hexacyanoferrate were synthesized by simple co-precipitation method. The water content and chemical formula was obtained by TGA and MP-AES measurements, functional groups by FT-IR analysis, the crystal structure by PXRD and a local geometry by XAS. Elemental species of cycled samples were further investigated by TXM and 2D XRF. Electrochemical tests were performed in the glass cell. With addition of nickel, vacancies and water content increased in the sample. Crystal structure changed from monoclinic to cubic. Ni disturbed the local structure of Mn, site, however, almost no change was observed in Fe site. After charge/discharge cycling of MnHCF intercalation was already found in the peripheries of charged species after 20 cycle in 2D XRF analysis and randomly distributed intercalated regions after 50 cycles in TXM analysis. Cyclic voltammetry showed that peak-to-peak separation is increasing in case of the addition of Ni to MnHCF.

Cadmium Telluride Magic-Sized Clusters and Superstructures

The aim of the present work is to gain new insights into the formation mechanism of CdTe magic-sized clusters (MSCs) at low temperatures, as well as on their evolution towards 1D and 2D nanostructures and assemblies thereof, under mild reaction conditions. The reaction system included toluene as solvent, octylamine as primary alkylamine, trioctylphosphine-Te as chalcogenide precursor and Cd(oleate)2 as metal precursor. UV-Vis absorption spectroscopy and transmission electron microscopy (TEM) were used to analyze samples containing concentrations of octylamine of 0.2, 0.8 and 2 M: well-defined, sharp absorption peaks were observed, with peaks maxima at 449, 417 and 373 nm respectively, and 1D structures with a string-like appearance were displayed in the TEM images. Investigating peaks growth, step-wise peaks shift to lower energies and reverse, step-wise peak shift to higher energies allowed to propose a model to describe the system, based on interconnected [CdTe]x cluster units originating an amine-capped, 1-dimensional, polymer-like structure, in which different degrees of electronic coupling between the clusters are held responsible for the different absorption transitions. The many parameters involved in the synthesis procedure were then investigated, starting from the Cd:Te ratio, the role of the amine, the use of different phosphine-Te and Cd precursors. The results allowed to gain important information of the reaction mechanism, as well as on the different behavior of the species featuring the sharp absorption peaks in each case. Using Cd(acetate)2 as metal precursor, 2D structures were found to evolve from the MSCs solutions over time, and their tendency to self-assemble was then analyzed employing two amines of different alkyl chain length, octylamine (C-8) and oleylamine (C-18). Their co-presence led to the formation of free-floating triangular nanosheets, which tend to readily aggregate if only octylamine is present in solution.