A tutorial on modelling and inference in undirected graphical models for hyperspectral image analysis

ABSTRACT

Undirected graphical models have been successfully used to jointly model the spatial and the spectral dependencies in earth observing hyperspectral images. They produce less noisy, smooth, and spatially coherent land-cover maps and give top accuracies on many datasets. Moreover, they can easily be combined with other state-of-the-art approaches, such as deep learning. This has made them an essential tool for remote-sensing researchers and practitioners. However, graphical models have not been easily accessible to the larger remote-sensing community as they are not discussed in standard remote-sensing textbooks and not included in the popular remote-sensing software and toolboxes. In this tutorial, we provide a theoretical introduction to Markov random fields and conditional random fields-based spatial–spectral classification for land-cover mapping along with a detailed step-by-step practical guide on applying these methods using freely available software. Furthermore, the discussed methods are benchmarked on four public hyperspectral datasets for a fair comparison among themselves and easy comparison with the vast number of methods in literature which use the same datasets. The source code necessary to reproduce all the results in the paper is published on-line to make it easier for the readers to apply these techniques to different remote-sensing problems.     

Parameter selection for region‐growing image segmentation algorithms using spatial autocorrelation

Region‐growing segmentation algorithms are useful for remote sensing image segmentation. These algorithms need the user to supply control parameters, which control the quality of the resulting segmentation. An objective function is proposed for selecting suitable parameters for region‐growing algorithms to ensure best quality results. It considers that a segmentation has two desirable properties: each of the resulting segments should be internally homogeneous and should be distinguishable from its neighbourhood. The measure combines a spatial autocorrelation indicator that detects separability between regions and a variance indicator that expresses the overall homogeneity of the regions.     

Virtualized Web server cluster self-configuration to optimize resource and power use

This work proposes a reusable architecture that enables the self-configuration of a supporting infrastructure for Web server clusters using virtual machines. The goal of the architecture is to ensure service quality, evaluating how broadly it complies with the application's operating restrictions and proportionally acting on the configuration of physical servers (hosts) or virtual machines. In addition, through the rational use of resources, the proposal aims at saving energy. A prototype of the architecture was developed and a performance evaluation carried out with two different resource management approaches. This evaluation shows how fully functional and advantageous the proposal is in terms of using resources, avoiding waste, yet maintaining the application's quality of service within acceptable levels. The architecture also shows to be flexible enough to accept, with a reasonable amount of effort, different resource self-configuration policies.     

All-pole phase-locked loops: calculating lock-in range by using Evan's root-locus

The dynamics of a phase-locked loop (PLL), an essential component for the electronic synchronization processes, is described by an ordinary nonlinear differential equation with an order equal to 1 plus the order of its internal linear low-pass filter. Literature contains several results mainly concerned to the second-order loops giving expressions for lock-in range and transient response specifying a linear equivalent second-order system. However, in some applications, more accurate transient responses are necessary and the PLL performance can be improved by considering the higher-order filters resulting in the nonlinear loops with order greater than 2. Such systems, due to high order and nonlinear terms, depending on the parameters combination can present some undesirable behaviors, resulting from bifurcations, as error oscillation and chaos, decreasing the synchronization ranges. Implication to engineering design is that some regions of the parameter space become forbidden limiting the circuit options. This work is a contribution on establishing the lock-in range for a PLL of generic order n?+?1, considering that the filter is all-pole linear stable low-pass of order n. Analysis is performed by detecting a Hopf bifurcation on the synchronous state by using the root-locus method combined with the dynamical system theory. The lock-in range is calculated by applying the classical control tools defining an equivalent feedback control system.     

NbO/Nb2O5 core–shells by thermal oxidation

Niobium oxides have been reported as an alternative to tantalum, and its oxide, for the production of capacitors, but these materials still require a further and exhaustive study. Understanding the dynamics of the oxidation process from NbO to Nb₂O₅ is essential to explain and control the properties of a niobium oxide based capacitors. In situ XRD analysis of NbO powders was performed from 100 to 1100 °C, which showed a temperature range where only amorphous phase is present. The capacitor anodes, formed from the same type of NbO powders, were then heat-treated in air from room-temperature (RT) to 550 °C and recorded with a video camera to see the colour shift. The presence of core–shell structures and the increase of the shell thickness/amount with temperature are visible in SEM. Finally, in situ spectroscopy studies were performed, in order to observe the interference spectra which are responsible for the observed colours.     

Phase transformation and microstructural evolution after heat treatment of a terbium-doped lithium–aluminum phosphate glass

The crystallization kinetics and phase transformation of a transparent Tb³⁺-doped lithium–aluminum phosphate glass, prepared by melt quenching, were investigated. The energy associated to the glass transition and the crystallization parameters (activation energy for crystallization and Avrami exponent) were evaluated by different methods using the experimental data obtained by differential thermal analysis performed at different heating rates. Using an isoconversional method to determine the change of the activation energy for crystallization with the fraction of crystallization, it was verified that with the increase in the fraction of crystallization from 0.1 to 0.9, the value of the activation energy decreased slightly from ~370 to ~310 kJ mol^?1 and that the Avrami exponent varied from 0.8 to 1, suggesting a surface crystal growth mechanism. Observation of the microstructural evolution of heat-treated glass samples confirmed a surface crystallization process revealing spherulitic crystals constituted mainly by aluminum metaphosphate.     

Correlates of physical activity in a national sample of girls and boys in Grades 4 through 12.

Psychological, biological, social, and physical environmental variables were examined for their association with physical activity of young people. A national sample of 1,504 parents and children in Grades 4–12 were interviewed by telephone. Twenty-two potential determinants were assessed along with an 11-item child physical activity index (α?=?.76). Hierarchical multiple regressions were conducted separately for 6 age–sex subgroups. Percentage of variance explained ranged from 18% for boys in Grades 4–6 to 59% for girls in Grades 10–12. Three variables had strong and consistent associations with the child physical activity index that generalized across subgroups: use of afternoon time for sports and physical activity, enjoyment of physical education, and family support for physical activity. These 3 variables should be targeted for change to promote physical activity in all groups of young people. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

A note on the use of developer charges in Australian local government

A pronounced degree of vertical imbalance in Australian fiscal federalism has placed increasing financial pressures on lower levels of government, including local government. This has led local government to seek additional ways of financing infrastructure development through developer charges. Due to the lower priority the Australian Bureau of Statistics has accorded local government statistics in recent years, it has been difficult to gauge the extent of use of developer charges as a revenue source. This paper attempts to review the evolution of developer contributions policy and describe the increasing importance of these charges in Australian local government, with special emphasis on New South Wales.     

Oxidation and reduction effects of propane–oxygen on Pd–chlorine/alumina catalysts

Pd–chloride precursor salt was used to prepare Pd/Al₂O₃ catalysts. TPSR measurements showed three distinct reactions for the oxidation of propane on palladium surface under excess of hydrocarbon: complete oxidation, steam reforming and propane hydrogenolysis. Propane oxidation on palladium catalysts was related to the Pd²⁺ sites observed on Pd/Al₂O₃ through infrared of adsorbed carbon monoxide. In fresh catalysts reduced by H₂, the IR spectra showed the linear and bridge adsorbed CO species on the Pd⁰ surface. After propane reaction, a new band at 2130 cm^-1 related to CO adsorption on Pd²⁺ species was noted. Carbon monoxide species adsorbed on Pd⁰ were also observed in all samples after reaction. Our results suggest surface ratios of Pd⁰/PdO during the propane oxidation. On the other hand, time on stream conversions of the complete oxidation of propane were affected by either the water generated during the reaction or added as a reactant at 10 vol%. The water generated by the reaction helped to eliminate chlorine residues in the form of oxychloride species leading to an increasing of the activity. However, the presence of water into the reaction mixture caused a strong decreasing of the activity. The inhibition mechanism of propane oxidation in the presence of water consisted in the dissociative adsorption of water on palladium sites with the possible formation of palladium hydroxide (Pd–OH) at the surface, diminishing the number of active surface sites. Dynamic fluctuations into the reaction conditions supported the idea that a pseudo‐equilibrium adsorption–desorption of water was reached. After water removal or increasing in the reaction temperature the equilibrium was shifted to the direction of OH–Pd decomposition. This behavior suggests that the inhibitory effect of water is a reversible phenomenon, being a function of the amount of water and the reaction temperature. This revised version was published online in July 2006 with corrections to the Cover Date.