A comparison of pixel and object-based land cover classification. A case study of the Asmara region, Eritrea

In this paper we compare the performance of two image classification paradigms (object- and pixel-based) for creating a land cover map of Asmara, the capital of Eritrea and its surrounding areas using a Landsat ETM+ imagery acquired in January 2000. The image classification methods used were maximum likelihood for the pixel-based approach and Bhattacharyya distance for the object-oriented approach available in, respectively, ArcGIS and SPRING software packages. Advantages and limitations of both approaches are presented and discussed. Classifications outputs were assessed using overall accuracy and Kappa indices. Pixel- and object-based classification methods result in an overall accuracy of 78% and 85%, respectively. The Kappa coefficient for pixel- and object-based approaches was 0.74 and 0.82, respectively. Although pixel-based approach is the most commonly used method, assessment and visual interpretation of the results clearly reveal that the object-oriented approach has advantages for this specific case-study.

DensiProbe Spine: a novel instrument for intraoperative measurement of bone density in transpedicular screw fixation

STUDY DESIGN.: Cadaver study. OBJECTIVE.: To determine bone strength in vertebrae by measuring peak breakaway torque or indentation force using custom-made pedicle probes. SUMMARY OF BACKGROUND DATA.: Screw performance in dorsal spinal instrumentation is dependent on bone quality of the vertebral body. To date no intraoperative measuring device to validate bone strength is available. Destructive testing may predict bone strength in transpedicular instrumentations in osteoporotic vertebrae. Insertional torque measurements showed varying results. METHODS.: Ten human cadaveric vertebrae were evaluated for bone mineral density (BMD) measurements by quantitative computed tomography. Peak torque and indentation force of custom-made probes as a measure for mechanical bone strength were assessed via a transpedicular approach. The results were correlated to regional BMD and to biomechanical load testing after pedicle screw implementation. RESULTS.: Both methods generated a positive correlation to failure load of the respective vertebrae. The correlation of peak breakaway torque to failure load was r = 0.959 (P = 0.003), therewith distinctly higher than the correlation of indentation force to failure load, which was r = 0.690 (P = 0.040). In predicting regional BMD, measurement of peak torque also performed better than that of indentation force (r = 0.897 [P = 0.002] vs. r = 0.777 [P = 0.017]). CONCLUSION.: Transpedicular measurement of peak breakaway torque is technically feasible and predicts reliable local bone strength and implant failure for dorsal spinal instrumentations in this experimental setting.

High-density electrode array for imaging in vitro electrophysiological activity

The development of a high-density active microelectrode array for in vitro electrophysiology is reported. Based on the Active Pixel Sensor (APS) concept, the array integrates 4096 gold microelectrodes (electrode separation 20 microm) on a surface of 2.5 mmx2.5 mm as well as a high-speed random addressing logic allowing the sequential selection of the measuring pixels. Following the electrical characterization in a phosphate solution, the functional evaluation has been carried out by recording the spontaneous electrical activity of neonatal rat cardiomyocytes. Signals with amplitudes from 130 microVp-p to 300 microVp-p could be recorded from different pixels. The results demonstrate the suitability of the APS concept for developing a new generation of high-resolution extracellular recording devices for in vitro electrophysiology.

Improved Space Object Orbit Determination Using CMOS Detectors

CMOS-sensors, or in general Active Pixel Sensors (APS), are rapidly replacing CCDs in the consumer camera market. Due to significant technological advances during the past years these devices start to compete with CCDs also for demanding scientific imaging applications, in particular in the astronomy community. CMOS detectors offer a series of inherent advantages compared to CCDs, due to the structure of their basic pixel cells, which each contains their own amplifier and readout electronics. The most prominent advantages for space object observations are the extremely fast and flexible readout capabilities, feasibility for electronic shuttering and precise epoch registration,and the potential to perform image processing operations on-chip and in real-time. Here, the major challenges and design drivers for ground-based and space-based optical observation strategies for objects in Earth orbit have been analyzed. CMOS detector characteristics were critically evaluated and compared with the established CCD technology, especially with respect to the above mentioned observations. Finally, we simulated several observation scenarios for ground- and space-based sensor by assuming different observation and sensor properties. We will introduce the analyzed end-to-end simulations of the ground- and spacebased strategies in order to investigate the orbit determination accuracy and its sensitivity which may result from different values for the frame-rate, pixel scale, astrometric and epoch registration accuracies. Two cases were simulated, a survey assuming a ground-based sensor to observe objects in LEO for surveillance applications, and a statistical survey with a space-based sensor orbiting in LEO observing small-size debris in LEO. The ground-based LEO survey uses a dynamical fence close to the Earth shadow a few hours after sunset. For the space-based scenario a sensor in a sun-synchronous LEO orbit, always pointing in the anti-sun direction to achieve optimum illumination conditions for small LEO debris was simulated.

Mapping tree density at a global scale

The global extent and distribution of forest trees is central to our understanding of the terrestrial biosphere. We provide the first spatially continuous map of forest tree density at a global scale. This map reveals that the global number of trees is approximately 3.04 trillion, an order of magnitude higher than the previous estimate. Of these trees, approximately 1.39 trillion exist in tropical and subtropical forests, with 0.74 trillion in boreal regions and 0.61 trillion in temperate regions. Biome-level trends in tree density demonstrate the importance of climate and topography in controlling local tree densities at finer scales, as well as the overwhelming effect of humans across most of the world. Based on our projected tree densities, we estimate that over 15 billion trees are cut down each year, and the global number of trees has fallen by approximately 46% since the start of human civilization.

The mammalian central nervous synaptic cleft contains a high density of periodically organized complexes.

Cryo-electron microscopy of vitreous section makes it possible to observe cells and tissues at high resolution in a close-to-native state. The specimen remains hydrated; chemical fixation and staining are fully avoided. There is minimal molecular aggregation and the density observed in the image corresponds to the density in the object. Accordingly, organotypic hippocampal rat slices were vitrified under high pressure and controlled cryoprotection conditions, cryosectioned at a final thickness of approximately 70 nm and observed below -170 degrees C in a transmission electron microscope. The general aspect of the tissue compares with previous electron microscopy observations. The detailed analysis of the synapse reveals that the density of material in the synaptic cleft is high, even higher than in the cytoplasm, and that it is organized in 8.2-nm periodic transcleft complexes. Previously undescribed structures of presynaptic and postsynaptic elements are also described.

Full correction for spatially distributed speed-of-sound in echo ultrasound based on measuring aberration delays via transmit beam steering

Aberrations of the acoustic wave front, caused by spatial variations of the speed-of-sound, are a main limiting factor to the diagnostic power of medical ultrasound imaging. If not accounted for, aberrations result in low resolution and increased side lobe level, over all reducing contrast in deep tissue imaging. Various techniques have been proposed for quantifying aberrations by analysing the arrival time of coherent echoes from so-called guide stars or beacons. In situations where a guide star is missing, aperture-based techniques may give ambiguous results. Moreover, they are conceptually focused on aberrators that can be approximated as a phase screen in front of the probe. We propose a novel technique, where the effect of aberration is detected in the reconstructed image as opposed to the aperture data. The varying local echo phase when changing the transmit beam steering angle directly reflects the varying arrival time of the transmit wave front. This allows sensing the angle-dependent aberration delay in a spatially resolved way, and thus aberration correction for a spatially distributed volume aberrator. In phantoms containing a cylindrical aberrator, we achieved location-independent diffraction-limited resolution as well as accurate display of echo location based on reconstructing the speed-of-sound spatially resolved. First successful volunteer results confirm the clinical potential of the proposed technique.

The Effects of Density, Spatial Pattern, and Competitive Symmetry on Size Variation in Simulated Plant Populations

Patterns of size inequality in crowded plant populations are often taken to be indicative of the degree of size asymmetry of competition, but recent research suggests that some of the patterns attributed to size‐asymmetric competition could be due to spatial structure. To investigate the theoretical relationships between plant density, spatial pattern, and competitive size asymmetry in determining size variation in crowded plant populations, we developed a spatially explicit, individual‐based plant competition model based on overlapping zones of influence. The zone of influence of each plant is modeled as a circle, growing in two dimensions, and is allometrically related to plant biomass. The area of the circle represents resources potentially available to the plant, and plants compete for resources in areas in which they overlap. The size asymmetry of competition is reflected in the rules for dividing up the overlapping areas. Theoretical plant populations were grown in random and in perfectly uniform spatial patterns at four densities under size‐asymmetric and size‐symmetric competition. Both spatial pattern and size asymmetry contributed to size variation, but their relative importance varied greatly over density and over time. Early in stand development, spatial pattern was more important than the symmetry of competition in determining the degree of size variation within the population, but after plants grew and competition intensified, the size asymmetry of competition became a much more important source of size variation. Size variability was slightly higher at higher densities when competition was symmetric and plants were distributed nonuniformly in space. In a uniform spatial pattern, size variation increased with density only when competition was size asymmetric. Our results suggest that when competition is size asymmetric and intense, it will be more important in generating size variation than is local variation in density. Our results and the available data are consistent with the hypothesis that high levels of size inequality commonly observed within crowded plant populations are largely due to size‐asymmetric competition, not to variation in local density.

Neighbourhood abundance and small-tree survival in a lowland Bornean rainforest

Conspecific effects of neighbours on small-tree survival may have a role in tree population dynamics and community composition of tropical forests. This notion was tested with data from two 4-ha plots in lowland forest at Danum, Sabah (Borneo), for a 21-year interval (censuses at 1986, 1996, 2001, 2007). Species with ≥45 focal trees 10 to <100 cm stem girth per plot in 1986 were selected. Logistic regressions fitted mean focal tree size and mean inverse-distance-weighted basal area abundance of neighbours (within 20 m), for the periods over which each focus tree was alive. Coefficients of variation of neighbourhood basal area abundance, both spatially and temporally, quantified the changing environment of each focus tree. Fits were critically and individually evaluated, with corrections for spatial autocorrelation. Conspecific effects at Danum was generally very weak or non-existent: species’ mortality rates varied also across plots. The main reasons appear to be that (1) species were not dense enough to interact despite frequent although weak spatial aggregation, and their neighbourhoods were highly differing in species composition; and (2) these neighbourhoods were highly variable temporally, meaning that focus trees experienced stochastically fluctuating neighbourhood environments. Only one species, Dimorphocalyx muricatus, showed strong conspecific effects (varying between plots) which can be explained by its distinct ecology. This understorey species is highly aggregated on ridges and is drought-tolerant. That this functionally and habitat-specialized species, has implied intraspecific density-dependent feedback in its dynamics is a remarkable indication of the overall processes maintaining stability of the Danum forest.

GTInspector: A Moldable Domain-Aware Object Inspector

Understanding the run-time behaviour of object-oriented applications entails the comprehension of run-time objects. Traditional object inspectors favor generic views that focus on the low-level details of the state of single objects. While universally applicable, this generic approach does not take into account the varying needs of developers that could benefit from tailored views and exploration possibilities. GTInspector is a novel moldable object inspector that provides different high-level ways to visualize and explore objects, adapted to both the object and the current developer need.