A stable graph-based representation for object recognition through high-order matching

Many Object recognition techniques perform some flavour of point pattern matching between a model and a scene. Such points are usually selected through a feature detection algorithm that is robust to a class of image transformations and a suitable descriptor is computed over them in order to get a reliable matching. Moreover, some approaches take an additional step by casting the correspondence problem into a matching between graphs defined over feature points. The motivation is that the relational model would add more discriminative power, however the overall effectiveness strongly depends on the ability to build a graph that is stable with respect to both changes in the object appearance and spatial distribution of interest points. In fact, widely used graph-based representations, have shown to suffer some limitations, especially with respect to changes in the Euclidean organization of the feature points. In this paper we introduce a technique to build relational structures over corner points that does not depend on the spatial distribution of the features. © 2012 ICPR Org Committee.

Reply: 'Myopic foveoschisis: An ectatic retinopathy, not a schisis'

Full text: We thank Tsilimbaris et al1 for their comments on the appropriateness of the term ‘myopic foveoschisis’ to describe the condition that is characterized by the separation of neural retina layers associated with high myopia and posterior staphyloma. They have proposed the term ‘myopic ectatic retinopathy’ as a more literal and functionally more accurate descriptor of the condition to avoid the use of the word ‘schisis’, which may be misleading because it is also used to describe other conditions where there is separation of neural retina layers without the presence of staphyloma.2 Using the word ‘ectatic’ for this condition would imply that we are fairly certain about the pathogenesis and mechanistic factors that underlie its development and progression. However, this is not the case, unfortunately, as our review of the literature has shown. There are several theories ranging from vitreous traction to sclerosing changes of retinal vessels to progression of staphylomas as possible etiological factors. Therefore, it is likely to be multifactorial in nature—hence the success reported with different procedures that address either the vitreous traction factor using vitrectomy, peel plus tamponade or the scleral ectasia factor using posterior buckling techniques. In the absence of a good understanding of underlying pathogenesis, it is probably best to use purely descriptive names rather than mechanistic terms. The use of descriptive terms, even though similar, do not necessarily cause confusion as long as they are widely accepted as differentiating terminology, for example, postoperative pseudophakic cystoid macular edema (Irvine–Gass syndrome) vs cystoid macular edema associated with posterior uveitis in a phakic patient. The introduction of too many mechanistic or pathogenetic terms in the absence of clear understating of etiology can in fact cause more confusion, for example, serous chorioretinopathy vs central serous retinopathy vs serous choroidopathy. The confinement to broad descriptive terms can enhance communication and reduce confusion without committing to any presumption about etiology until it is better understood. This approach is probably best illustrated by the recent advances in the understanding of mactel21, a condition initially described and classified, using descriptive nomenclature, by Don Gass as bilateral, idiopathic acquired juxtafoveolar telangiectasis (Group2A) and as distinctly different from unilateral, congenital parafoveolar telangiectasis (Group 1A; Gass,3 pp 504–506 vs 127–128). Finally, it is worthy to note that for myopic foveoschisis associated with a staphyloma that is associated with outer layer macular detachment, Don Gass also descriptively included the additional observation (before the advent of OCT) that the retinal profile was concave rather than convex in shape, thereby differentiating it from rhegmatogenous detachments with recruitment of subretinal fluid that is associated with posteriorly located breaks and macular holes in myopic eyes. References 1.Tsilimbaris MK, Vavvas DG, Bechrakis NE. Myopic foveoschisis: an ectatic retinopathy, not aschisis. Eye 2016; 30: 328–329. 2.Powner MB, Gillies MC, Tretiach M, Scott A, Guymer RH, Hageman GS et al. Perifoveal müller cell depletion in a case of macular telangiectasia type 2. Ophthalmology 2010; 117(12): 2407–2416. 3.Gass DM. Stereoscopic Atlas of Macular Diseases: Diagnosis and Treatment, 4th edn. Mosby-Yearbook: St. Louis, 1997.

A research platform for artificial neural networks with applications in pediatric epilepsy

This dissertation established a state-of-the-art programming tool for designing and training artificial neural networks (ANNs) and showed its applicability to brain research. The developed tool, called NeuralStudio, allows users without programming skills to conduct studies based on ANNs in a powerful and very user friendly interface. A series of unique features has been implemented in NeuralStudio, such as ROC analysis, cross-validation, network averaging, topology optimization, and optimization of the activation function’s slopes. It also included a Support Vector Machines module for comparison purposes. Once the tool was fully developed, it was applied to two studies in brain research. In the first study, the goal was to create and train an ANN to detect epileptic seizures from subdural EEG. This analysis involved extracting features from the spectral power in the gamma frequencies. In the second application, a unique method was devised to link EEG recordings to epileptic and nonepileptic subjects. The contribution of this method consisted of developing a descriptor matrix that can be used to represent any EEG file regarding its duration and the number of electrodes. The first study showed that the inter-electrode mean of the spectral power in the gamma frequencies and its duration above a specific threshold performs better than the other frequencies in seizure detection, exhibiting an accuracy of 95.90%, a sensitivity of 92.59%, and a specificity of 96.84%. The second study yielded that Hjorth’s parameter activity is sufficient to accurately relate EEG to epileptic and non-epileptic subjects. After testing, accuracy, sensitivity and specificity of the classifier were all above 0.9667. Statistical tests measured the superiority of activity at over 99.99 % certainty. It was demonstrated that (1) the spectral power in the gamma frequencies is highly effective in locating seizures from EEG and (2) activity can be used to link EEG recordings to epileptic and non-epileptic subjects. These two studies required high computational load and could be addressed thanks to NeuralStudio. From a medical perspective, both methods proved the merits of NeuralStudio in brain research applications. For its outstanding features, NeuralStudio has been recently awarded a patent (US patent No. 7502763).

Rigid and non-rigid point-based medical image registration

The primary goal of this dissertation is to develop point-based rigid and non-rigid image registration methods that have better accuracy than existing methods. We first present point-based PoIRe, which provides the framework for point-based global rigid registrations. It allows a choice of different search strategies including (a) branch-and-bound, (b) probabilistic hill-climbing, and (c) a novel hybrid method that takes advantage of the best characteristics of the other two methods. We use a robust similarity measure that is insensitive to noise, which is often introduced during feature extraction. We show the robustness of PoIRe using it to register images obtained with an electronic portal imaging device (EPID), which have large amounts of scatter and low contrast. To evaluate PoIRe we used (a) simulated images and (b) images with fiducial markers; PoIRe was extensively tested with 2D EPID images and images generated by 3D Computer Tomography (CT) and Magnetic Resonance (MR) images. PoIRe was also evaluated using benchmark data sets from the blind retrospective evaluation project (RIRE). We show that PoIRe is better than existing methods such as Iterative Closest Point (ICP) and methods based on mutual information. We also present a novel point-based local non-rigid shape registration algorithm. We extend the robust similarity measure used in PoIRe to non-rigid registrations adapting it to a free form deformation (FFD) model and making it robust to local minima, which is a drawback common to existing non-rigid point-based methods. For non-rigid registrations we show that it performs better than existing methods and that is less sensitive to starting conditions. We test our non-rigid registration method using available benchmark data sets for shape registration. Finally, we also explore the extraction of features invariant to changes in perspective and illumination, and explore how they can help improve the accuracy of multi-modal registration. For multimodal registration of EPID-DRR images we present a method based on a local descriptor defined by a vector of complex responses to a circular Gabor filter.

Automated detection of hematological abnormalities through classification of flow cytometric data patterns

Flow Cytometry analyzers have become trusted companions due to their ability to perform fast and accurate analyses of human blood. The aim of these analyses is to determine the possible existence of abnormalities in the blood that have been correlated with serious disease states, such as infectious mononucleosis, leukemia, and various cancers. Though these analyzers provide important feedback, it is always desired to improve the accuracy of the results. This is evidenced by the occurrences of misclassifications reported by some users of these devices. It is advantageous to provide a pattern interpretation framework that is able to provide better classification ability than is currently available. Toward this end, the purpose of this dissertation was to establish a feature extraction and pattern classification framework capable of providing improved accuracy for detecting specific hematological abnormalities in flow cytometric blood data. ^ This involved extracting a unique and powerful set of shift-invariant statistical features from the multi-dimensional flow cytometry data and then using these features as inputs to a pattern classification engine composed of an artificial neural network (ANN). The contribution of this method consisted of developing a descriptor matrix that can be used to reliably assess if a donor’s blood pattern exhibits a clinically abnormal level of variant lymphocytes, which are blood cells that are potentially indicative of disorders such as leukemia and infectious mononucleosis. ^ This study showed that the set of shift-and-rotation-invariant statistical features extracted from the eigensystem of the flow cytometric data pattern performs better than other commonly-used features in this type of disease detection, exhibiting an accuracy of 80.7%, a sensitivity of 72.3%, and a specificity of 89.2%. This performance represents a major improvement for this type of hematological classifier, which has historically been plagued by poor performance, with accuracies as low as 60% in some cases. This research ultimately shows that an improved feature space was developed that can deliver improved performance for the detection of variant lymphocytes in human blood, thus providing significant utility in the realm of suspect flagging algorithms for the detection of blood-related diseases.^

A Research Platform for Artificial Neural Networks with Applications in Pediatric Epilepsy

This dissertation established a state-of-the-art programming tool for designing and training artificial neural networks (ANNs) and showed its applicability to brain research. The developed tool, called NeuralStudio, allows users without programming skills to conduct studies based on ANNs in a powerful and very user friendly interface. A series of unique features has been implemented in NeuralStudio, such as ROC analysis, cross-validation, network averaging, topology optimization, and optimization of the activation function’s slopes. It also included a Support Vector Machines module for comparison purposes. Once the tool was fully developed, it was applied to two studies in brain research. In the first study, the goal was to create and train an ANN to detect epileptic seizures from subdural EEG. This analysis involved extracting features from the spectral power in the gamma frequencies. In the second application, a unique method was devised to link EEG recordings to epileptic and non-epileptic subjects. The contribution of this method consisted of developing a descriptor matrix that can be used to represent any EEG file regarding its duration and the number of electrodes. The first study showed that the inter-electrode mean of the spectral power in the gamma frequencies and its duration above a specific threshold performs better than the other frequencies in seizure detection, exhibiting an accuracy of 95.90%, a sensitivity of 92.59%, and a specificity of 96.84%. The second study yielded that Hjorth’s parameter activity is sufficient to accurately relate EEG to epileptic and non-epileptic subjects. After testing, accuracy, sensitivity and specificity of the classifier were all above 0.9667. Statistical tests measured the superiority of activity at over 99.99 % certainty. It was demonstrated that 1) the spectral power in the gamma frequencies is highly effective in locating seizures from EEG and 2) activity can be used to link EEG recordings to epileptic and non-epileptic subjects. These two studies required high computational load and could be addressed thanks to NeuralStudio. From a medical perspective, both methods proved the merits of NeuralStudio in brain research applications. For its outstanding features, NeuralStudio has been recently awarded a patent (US patent No. 7502763).

Rigid and Non-rigid Point-based Medical Image Registration

The primary goal of this dissertation is to develop point-based rigid and non-rigid image registration methods that have better accuracy than existing methods. We first present point-based PoIRe, which provides the framework for point-based global rigid registrations. It allows a choice of different search strategies including (a) branch-and-bound, (b) probabilistic hill-climbing, and (c) a novel hybrid method that takes advantage of the best characteristics of the other two methods. We use a robust similarity measure that is insensitive to noise, which is often introduced during feature extraction. We show the robustness of PoIRe using it to register images obtained with an electronic portal imaging device (EPID), which have large amounts of scatter and low contrast. To evaluate PoIRe we used (a) simulated images and (b) images with fiducial markers; PoIRe was extensively tested with 2D EPID images and images generated by 3D Computer Tomography (CT) and Magnetic Resonance (MR) images. PoIRe was also evaluated using benchmark data sets from the blind retrospective evaluation project (RIRE). We show that PoIRe is better than existing methods such as Iterative Closest Point (ICP) and methods based on mutual information. We also present a novel point-based local non-rigid shape registration algorithm. We extend the robust similarity measure used in PoIRe to non-rigid registrations adapting it to a free form deformation (FFD) model and making it robust to local minima, which is a drawback common to existing non-rigid point-based methods. For non-rigid registrations we show that it performs better than existing methods and that is less sensitive to starting conditions. We test our non-rigid registration method using available benchmark data sets for shape registration. Finally, we also explore the extraction of features invariant to changes in perspective and illumination, and explore how they can help improve the accuracy of multi-modal registration. For multimodal registration of EPID-DRR images we present a method based on a local descriptor defined by a vector of complex responses to a circular Gabor filter.

Geochemistry on manganese nodules from the central Pacific

During Cruise VA 04/2 of the R. V. Valdivia in 1972, numerous samples were collected of manganese nodules, sodiments, and the pore water of the Sediments in the area betwoen 6° N and 11° N and between 149° W and 154° W in the Central Pacific. This paper reports on the geochemical study of 70 manganese nodules from 12 stations (box-core and dredge samples). The nodules Irom a single Station vary considerably in diemical composition. These variations are of the same magnitude as the variations in nodule composition over the entire survey area. The ratios of manganese to nickel, copper, and zinc show good positive correlation as do the ratios of iron to lead and, to a lesser extent, of iron to cobalt. No correlation was found between the environment of the manganese nodules and their metal content. Three internal zones, the outer. intermediate, and core zones, of some nodules were studied in detail. Trends of differences of Chemical composition from surface to core were found for numerous elements and elemental ratios.

Ground-based electromagnetic (EM) and drill-hole ice and snow thickness and melt pond depth measurements during Polarstern cruise ARK-IX/4, ARK-XI/1, and ARK-XII

Drillhole-determined sea-ice thickness was compared with values derived remotely using a portable small-offset loop-loop steady state electromagnetic (EM) induction device during expeditions to Fram Strait and the Siberian Arctic, under typical winter and summer conditions. Simple empirical transformation equations are derived to convert measured apparent conductivity into ice thickness. Despite the extreme seasonal differences in sea-ice properties as revealed by ice core analysis, the transformation equations vary little for winter and summer. Thus, the EM induction technique operated on the ice surface in the horizontal dipole mode yields accurate results within 5 to 10% of the drillhole determined thickness over level ice in both seasons. The robustness of the induction method with respect to seasonal extremes is attributed to the low salinity of brine or meltwater filling the extensive pore space in summer. Thus, the average bulk ice conductivity for summer multiyear sea ice derived according to Archie's law amounts to 23 mS/m compared to 3 mS/m for winter conditions. These mean conductivities cause only minor differences in the EM response, as is shown by means of 1-D modeling. However, under summer conditions the range of ice conductivities is wider. Along with the widespread occurrence of surface melt ponds and freshwater lenses underneath the ice, this causes greater scatter in the apparent conductivity/ice thickness relation. This can result in higher deviations between EM-derived and drillhole determined thicknesses in summer than in winter.

An augmented reality pipeline to create scenes with coherent illumination using textured cuboids

Shadows and illumination play an important role when generating a realistic scene in computer graphics. Most of the Augmented Reality (AR) systems track markers placed in a real scene and retrieve their position and orientation to serve as a frame of reference for added computer generated content, thereby producing an augmented scene. Realistic depiction of augmented content with coherent visual cues is a desired goal in many AR applications. However, rendering an augmented scene with realistic illumination is a complex task. Many existent approaches rely on a non automated pre-processing phase to retrieve illumination parameters from the scene. Other techniques rely on specific markers that contain light probes to perform environment lighting estimation. This study aims at designing a method to create AR applications with coherent illumination and shadows, using a textured cuboid marker, that does not require a training phase to provide lighting information. Such marker may be easily found in common environments: most of product packaging satisfies such characteristics. Thus, we propose a way to estimate a directional light configuration using multiple texture tracking to render AR scenes in a realistic fashion. We also propose a novel feature descriptor that is used to perform multiple texture tracking. Our descriptor is an extension of the binary descriptor, named discrete descriptor, and outperforms current state-of-the-art methods in speed, while maintaining their accuracy.

Modelo para reconstrução 3D de cenas baseado em imagens

3D Reconstruction is the process used to obtain a detailed graphical model in three dimensions that represents some real objectified scene. This process uses sequences of images taken from the scene, so it can automatically extract the information about the depth of feature points. These points are then highlighted using some computational technique on the images that compose the used dataset. Using SURF feature points this work propose a model for obtaining depth information of feature points detected by the system. At the ending, the proposed system extract three important information from the images dataset: the 3D position for feature points; relative rotation and translation matrices between images; the realtion between the baseline for adjacent images and the 3D point accuracy error found.

Meating the social: sharing atiku-euiash in Sheshatshiu, Labrador

This work examines atiku-euiash (caribou meat) sharing practices in Sheshatshiu, Newfoundland and Labrador, and aims to elucidate an overarching question: how do sharing practices participate in the co-constitution of the Innu ‘social’? The ‘social’ is understood in this work as a descriptor that refers to the emergent properties of the Innu collective. The thesis is that sharing practices participate in the co-constitution of the Innu social and enact its boundaries. Inside these boundaries, atiku-euiash is more than simply a food resource: by realizing Innu values of generosity, respect and autonomy, sharing implicates the associations of human, animal, and animal masters that constitute the Innu world. Sharing is connected with the enskilment of the younger generations by their el-ders, and thus with the reproduction of Innu values through time. The ways of sharing are relevant because changes in such practices affect the constitution of the Innu social. Giv-en Euro-Canadian colonization, the Innu are in a fraught social space in which sharing is interrupted by colonization practices and values. Understanding sharing is necessary to develop policies that do not interrupt the reproduction of the Innu world This work uses several research methods: participant observation, sharing surveys, and interviews. It also uses network analysis as sharing practices leave traces of giving and receiving actions and these traces can be represented as a network of givers, receivers and circulating caribou meat. There are two main ways in which caribou is hunted and shared: household-based hunts and community-based hunts. The household-based hunts are organized by the hunters themselves, who are able and willing to hunt. Community-based hunts are completely organized and funded by the SIFN or the Innu Nation. In or-der to understand the differences in the distribution of the two hunt types, the categories of centrality and clustering are used to show how the flow of atiku-eiuash and its associ-ated realization of values and enskilment correlate with different degrees of centralization inside the sharing clusters.

Progettazione e implementazione di una applicazione didattica interattiva per il riconoscimento di oggetti basata sull'algoritmo SIFT

Nell'elaborato viene introdotto l'ambito della Computer Vision e come l'algoritmo SIFT si inserisce nel suo panorama. Viene inoltre descritto SIFT stesso, le varie fasi di cui si compone e un'applicazione al problema dell'object recognition. Infine viene presentata un'implementazione di SIFT in linguaggio Python creata per ottenere un'applicazione didattica interattiva e vengono mostrati esempi di questa applicazione.

Physical oceanography, sea-bed photographs and videos of benthos from the Weddell Sea taken with remote operated vehicle CHEROKEE during POLARSTERN cruise ANT-XXIII/8

The marine ecosystem on the eastern shelf of the Antarctic Peninsula was surveyed 5 and 12 years after the climate-induced collapse of the Larsen A and B ice shelves. An impoverished benthic fauna was discovered, that included deep-sea species presumed to be remnants from ice-covered conditions. The current structure of various ecosystem components appears to result from extremely different response rates to the change from an oligotrophic sub-ice-shelf ecosystem to a productive shelf ecosystem. Meiobenthic communities remained impoverished only inside the embayments. On local scales, macro- and mega-epibenthic diversity was generally low, with pioneer species and typical Antarctic megabenthic shelf species interspersed. Antarctic Minke whales and seals utilised the Larsen A/B area to feed on presumably newly established krill and pelagic fish biomass. Ecosystem impacts also extended well beyond the zone of ice-shelf collapse, with areas of high benthic disturbance resulting from scour by icebergs discharged from the Larsen embayments.

Surface characteristics and metazoa occurrence at ice station PS67/006-1 during POLARSTERN cruise ANT-XXII/2 (Nov 2004-Jan 2005)

The surface and sub-ice layer habitats and their metazoan fauna were studied on a drifting pack-ice floe in the western Weddell Sea from 29 November 2004 to 1 January 2005 during the "Ice Station POLarstern" (ISPOL). Flooding of the floe occurred at some places, and the establishment of surface layers with a brownish colour due to growing algae was observed at several sampling sites. The average surface-layer temperature, brine salinity and brine volume were -1.4 °C, 25.3 and 54%, respectively. The temperature-salinity relationship in the surface layer was seldom at equilibrium conditions. Chlorophyll a (Chl a) concentrations in the brine varied between 1.0 and 53.5 µg /L. Surface-layer thickness, salinity, Chl a concentration and copepod abundances were generally higher at the edge of the floe than in the inner part. The sympagic copepod species Drescheriella glacialis/racovitzai and Stephos longipes, with abundances ranging between 0 and 3830 ind/L (median: 2 ind/L) and 0 and 1293 ind/L (median: 4 ind/L), respectively, were the dominant members of the surface-layer meiofauna. Their populations consisted mainly of adults and early naupliar stages, which points to an active reproduction of these species within the surface layer. Other taxa found in the surface layer were undetermined turbellarians, the gastropod Tergipes antarcticus, and, for the first time, the ctenophore Callianira antarctica, and the amphipods Eusirus antarcticus and Eusirus tridentatus. During the course of our study, slight melting at the ice underside took place, releasing sympagic organisms to the water column. Chl a concentrations in the sub-ice water layer were very low (0.1-0.5 µg /L), except for 25 December when the Chl a concentration at 0 m depth increased to 2.3 µg /L. The most dominant sympagic copepod species found in the sub-ice layer was Ectinosoma sp., with abundances ranging between 1 and 599 ind/m**3 (median: 25 ind/m**3). Other sympagic copepod species occurring regularly in this habitat were D. glacialis/racovitzai, Diarthrodes cf. lilacinus, Idomene antarctica and S. longipes. All of these sympagic species were generally found in higher abundances at 0 m depth underneath the ice than at 5 m depth, in contrast to pelagic copepod species that occurred more frequently at 5 m depth. Niche separation and probable life-cycle strategies of dominant sympagic metazoans are discussed.