The notion of watershed, used in morphological segmentation, has only a digital definition. In this paper, we propose to extend this definition to the continuous plane. Using this continuous definition, we present the watershed differences with classical edge detectors. We then present a metric in the plane for which the watershed is a skeleton by influence zones and show the lower semicontinuous behaviour of the associated skeleton. This theoretical approach suggests an algorithm for solving the eikonal equation: ∇f = g. Finally, we end with some new watershed algorithms, which present the advantage of allowing the use of markers and/or anchor points, thus opening the way towards grey-tone skeletons.
RésuméLa notion de ligne de partage des eaux, utilisée en segmentation morphologique dispose uniquement d'une définition digitale. Dans cet article, nous proposons d'étendre la définition de la ligne de partage des eaux au plan
In the data set collected by the authors in freshly excised breast tissue, the admittance loci generally differed from circular arcs, rendering the calculation of the usual set of parameters impossible. Alternative parameters were used for the analysis of these data. The present study consists of the definition and evaluation of a set of such parameters aimed at the characterization and differentiation of breast tissues. These parameters were defined so that their calculation does not require the fit of circular arcs to the experimental points and is independent of any equivalent circuit model. The results of the statistical analysis showed significant differences between most of the tissue groups, especially between cancerous tissue and all the other groups, which confirmed that impedance spectroscopy can be considered as potentially suitable for breast cancer detection.
Six experiments were carried out to investigate the issue of cross-modality between exogenous auditory and visual spatial attention employing Posner's cueing paradigm in detection, localization, and discrimination tasks. Results indicated cueing in detection tasks with visual or auditory cues and visual targets but not with auditory targets (Experiment 1). In the localization tasks, cueing was found with both visual and auditory targets. Inhibition of return was apparent only in the within-modality conditions (Experiment 2). This suggests that it is important whether the attention system is activated directly (within a modality) or indirectly (between modalities). Increasing the cue validity from 50% to 80% influenced performance only in the localization task (Experiment 4). These findings are interpreted as being indicative for modality-specific but interacting attention mechanisms. The results of Experiments 5 and 6 (up/down discrimination tasks) also show cross-modal cueing but not with visual cues and auditory targets. Furthermore, there was no inhibition of return in any condition. This suggests that some cueing effects might be task dependent.
Despite agreement among many attentional theories that processing resources are limited and allocated according to task demands, controversy continues about the locus of selectivity. Studies of spatial orientation of attention suggest an early effect. These results, however, can be explained instead by effects of decision processes. The present study avoids this difficulty by directly manipulating attention in a dual-task paradigm and by using SDT to dissociate sensory tuning from criterion shifts. Ten subjects judged whether two lines to the left of fixation were the same or different in length; they also judged two lines presented simultaneously to the right. In a given block of 64 trials, the subject was to allocate 8O%, 50%, or 20% of attention to one pair of lines and the rest to the other. On every trial, the subject judged both pairs. Results showed that d increased from 0.77 with 20% allocation to 1.69 with 80%, indicating that sensitivity is modulated by attentional instructions. These results are predicted quantitatively by Luce's sample-size model.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.