Pierre-Yves Burgi

A 128 /spl times/ 128 pixel 120 dB dynamic range vision sensor chip for image contrast and orientation extraction

This vision sensor outputs luminance, contrast magnitude and contrast orientation of image features for surveillance and automotive applications. The sensor produces a contrast representation with a dynamic range of 120 dB and a sensitivity of 2%. The chip is fabricated in a 0.5 /spl mu/m 3M 2P process and dissipates 300 mW at 3.3 V.

Probabilistic Motion Estimation Based on Temporal Coherence

We develop a theory for the temporal integration of visual motion motivated by psychophysical experiments. The theory proposes that input data are temporally grouped and used to predict and estimate the motion flows in the image sequence. This temporal grouping can be considered a generalization of the data association techniques that engineers use to study motion sequences. Our temporal grouping theory is expressed in terms of the Bayesian generalization of standard Kalman filtering. To implement the theory, we derive a parallel network that shares some properties of cortical networks. Computer simulations of this network demonstrate that our theory qualitatively accounts for psychophysical experiments on motion occlusion and motion outliers. In deriving our theory, we assumed spatial factorizability of the probability distributions and made the approximation of updating the marginal distributions of velocity at each point. This allowed us to perform local computations and simplified our implementation. We argue that these approximations are suitable for the stimuli we are considering (for which spatial coherence effects are negligible).

Dynamics of the Mount Nyiragongo lava lake

The permanent and presently rising lava lake at Mount Nyiragongo constitutes a major potential geological hazard to the inhabitants of the Virunga volcanic region in the Democratic Republic of Congo (DRC) and Rwanda. Based on two field campaigns in June 2010 and 2011, we estimate the lava lake level from the southeastern crater rim (~400 m diameter) and lava lake area (~46,550 m2), which constrains, respectively, the lava lake volume (~9 × 106 m3) and volume flow rate needed to keep the magma in a molten state (0.6 to 3.5 m3 s−1). A bidirectional magma flow model, which includes the characterization of the conduit diameter and funnel-shaped lava lake geometry, is developed to constrain the amount of magma intruded/emplaced within the magmatic chamber and rift-related structures that extend between Mount Nyiragongos volcanic center and the city of Goma, DRC, since Mount Nyiragongos last eruption (17 January 2002). Besides matching field data of the lava lake level covering the period 1977 to 2002, numerical solutions of the model indicate that by 2022, 20 years after the January 2002 eruption, between 300 and 1700 × 106 m3 (0.3 to 1.7 km3) of magma could have intruded/emplaced underneath the edifice, and the lava lake volume could exceed 15 × 106 m3.

Asynchrony in image analysis: using the luminance-to-response-latency relationship to improve segmentation

We deal with the problem of segmenting static images, a procedure known to be difficult in the case of very noisy patterns. The proposed approach rests on the transformation of a static image into a data flow in which the first image points to be processed are the brighter ones. This solution, inspired by human perception, in which strong luminances elicit reactions from the visual system before weaker ones, has led to the notion of asynchronous processing. The asynchronous processing of image points has required the design of a specific architecture that exploits time differences in the processing of information. The results obtained when very noisy images are segmented demonstrate the strengths of this architecture; they also suggest extensions of the approach to other computer vision problems.

Toward a biophysically plausible bidirectional Hebbian rule

Although the commonly used quadratic Hebbian-;anti-Hebbian rules lead to successful models of plasticity and learning, they are inconsistent with neurophysiology. Other rules, more physiologically plausible, fail to specify the biological mechanism of bidirectionality and the biological mechanism that prevents synapses from changing from excitatory to inhibitory, and vice versa. We developed a synaptic bidirectional Hebbian rule that does not suffer from these problems. This rule was compared with physiological homosynaptic conditions in the hippocampus, with the results indicating the consistency of this rule with long-term potentiation (LTP) and long-term depression (LTD) phenomenologies. The phenomenologies considered included the reversible dynamics of LTP and LTD and the effects of N-methyl-D-aspartate blockers and phosphatase inhibitors.

A Real Time Implementation of the Saliency-Based Model of Visual Attention on a SIMD Architecture

Visual attention is the ability to rapidly detect the visually salient parts of a given scene. Inspired by biological vision, the saliency-based algorithm efficiently models the visual attention process. Due to its complexity, the saliency-based model of visual attention needs, for a real time implementation, higher computation resources than available in conventional processors. This work reports a real time implementation of this attention model on a highly parallel Single Instruction Multiple Data (SIMD) architecture called ProtoEye. Tailored for low-level image processing, ProtoEye consists of a 2D array of mixed analog-digital processing elements (PE). The operations required for visual attention computation are optimally distributed on the analog and digital parts. The analog diffusion network is used to implement the spatial filtering-based transformations such as the conspicuity operator and the competitive normalization of conspicuity maps. Whereas the digital part of Proto-Eye allows the implementation of logical and arithmetical operations, for instance, the integration of the normalized conspicuity maps into the final saliency map. Using 64×64 gray level images, the on ProtoEye implemented attention process operates in real-time. It runs at a frequency of 14 images per second.

An Institutional Personal Learning Environment Enabler

In this paper, we first discuss the concept of Personal Learning Environment (PLE) with respect to higher-education institutions and Virtual Learning Environments (VLEs). This discussion rapidly confronts us to the place of the PLE and self-directed learning and/or training inside the institution. We therefore introduce the concept of institutional PLE enabler, which is expected to stimulate students to create and use their own resources and institutional resources and share them with peers during formal and informal learning activities. Next, we describe a proposal for a federated design and implementation of the PLE enabler across multiple institutions.

Visual motion estimation and prediction: a probabilistic network model for temporal coherence

We develop a theory for the temporal integration of visual motion motivated by psychophysical experiments. The theory proposes that input data are temporally grouped and used to predict and estimate motion flows in the image sequences. Our theory is expressed in terms of the Bayesian generalization of standard Kalman filtering which allows us to solve temporal grouping in conjunction with prediction and estimation. As demonstrated for tracking isolated contours the Bayesian formulation is superior to approaches which use data association as a first stage followed by conventional Kalman filtering. Our computer simulations demonstrate that our theory qualitatively accounts for several psychophysical experiments on motion occlusion and motion outliers.

Possible roles of spontaneous waves and dendritic growth for retinal receptive field development

Several models of cortical development postulate that a Hebbian process fed by spontaneous activity amplifies orientation biases occurring randomly in early wiring, to form orientation selectivity. These models are not applicable to the development of retinal orientation selectivity, since they neglect the polarization of the retinas poorly branched early dendritic trees and the wavelike organization of the retinas early noise. There is now evidence that dendritic polarization and spontaneous waves are key in the development of retinal receptive fields. When models of cortical development are modified to take these factors into account, one obtains a model of retinal development in which early dendritic polarization is the seed of orientation selectivity, while the spatial extent of spontaneous waves controls the spatial profile of receptive fields and their tendency to be isotropic.

A biophysical model for the developmental time course of retinal orientation selectivity.

A quantitative study of the time course of development of the percentage of orientationally selective and isotropic ganglion cells in turtle retina has recently been performed. This study revealed that as soon as ganglion cells start responding to light, a large percentage of them are selective to the orientations of moving visual stimuli. This percentage decreases with age to reach a minimum around hatching, increases dramatically after birth and finally, decreases again following the first month of life to reach adult level. Concomitantly, the percentage of cells responding isotropically to the orientation of elongated stimuli increases monotonically until about 30 days after birth, stabilizing afterwards. To account for both time courses, we propose a biophysical model implementing features ubiquitous to developing vertebrate retinas. These features include early dendritic and synaptic spatial polarization, dendritic growth, and waves of activity generated spontaneously or by visual stimulation sweeping across the inner plexiform layer (IPL). The model also assumes a physiologically plausible Hebbian rule, which includes long-term potentiation and depression. Computer simulations of this model yield good fits of the data. The quality of these fits confirms and extends results from an earlier model using computationally-simple mechanisms, which suggested that early dendritic polarization might be the seed for mature orientation selectivity.