Motion of points and lines in the uncalibrated case

In the present paper we address the problem of computing structure and motion, given a set point and/or line correspondences, in a monocular image sequence, when the camera is not calibrated.Considering point correspondences first, we analyse how to parameterize the retinal correspondences, in function of the chosen geometry: Euclidean, affine or projective geometry. The simplest of these parameterizations is called the FQs-representation and is a composite projective representation. The main result is that considering N+1 views in such a monocular image sequence, the retinal correspondences are parameterized by 11 N–4 parameters in the general projective case. Moreover, 3 other parameters are required to work in the affine case and 5 additional parameters in the Euclidean case. These 8 parameters are calibration parameters and must be calculated considering at least 8 external informations or constraints. The method being constructive, all these representations are made explicit.Then, considering line correspondences, we show how the the same parameterizations can be used when we analyse the motion of lines, in the uncalibrated case. The case of three views is extensively studied and a geometrical interpretation is proposed, introducing the notion of trifocal geometry which generalizes the well known epipolar geometry. It is also discussed how to introduce line correspondences, in a framework based on point correspondences, using the same equations.Finally, considering the F Qs-representation, one implementation is proposed as a motion module, taking retinal correspondences as input, and providing and estimation of the 11 N–4 retinal motion parameters. As discussed in this paper, this module can also estimate the 3D depth of the points up to an affine and projective transformation, defined by the 8 parameters identified in the first section. Experimental results are provided.     

Learning state machine-based string edit kernels

During the past few years, several works have been done to derive string kernels from probability distributions. For instance, the Fisher kernel uses a generative model M (e.g. a hidden Markov model) and compares two strings according to how they are generated by M. On the other hand, the marginalized kernels allow the computation of the joint similarity between two instances by summing conditional probabilities. In this paper, we adapt this approach to edit distance-based conditional distributions and we present a way to learn a new string edit kernel. We show that the practical computation of such a kernel between two strings x and x^′ built from an alphabet Σ requires (i) to learn edit probabilities in the form of the parameters of a stochastic state machine and (ii) to calculate an infinite sum over Σ^* by resorting to the intersection of probabilistic automata as done for rational kernels. We show on a handwritten character recognition task that our new kernel outperforms not only the state of the art string kernels and string edit kernels but also the standard edit distance used by a neighborhood-based classifier.     

The undirected capacitated arc routing problem with profits

A profit and a demand are associated with each edge of a set of profitable edges of a given graph. A travel time is associated with each edge of the graph. A fleet of capacitated vehicles is given to serve the profitable edges. A maximum duration of the route of each vehicle is also given. The profit of an edge can be collected by one vehicle only that also serves the demand of the edge. The objective of this problem, which is called the undirected capacitated arc routing problem with profits (UCARPP), is to find a set of routes that satisfy the constraints on the duration of the route and on the capacity of the vehicle and maximize the total collected profit. We propose a branch-and-price algorithm and several heuristics. We can solve exactly instances with up to 97 profitable edges. The best heuristics find the optimal solution on most of instances where it is available.     

Modular architecture of the microfactories for automatic micro-assembly

The construction of a new generation of MEMS which includes micro-assembly steps in the current microfabrication process is a big challenge. It is necessary to develop new production means named micromanufacturing systems in order to perform these new assembly steps. The classical approach called “top-down” which consists in a functional analysis and a definition of the tasks sequences is insufficient for micromanufacturing systems. Indeed, the technical and physical constraints of the microworld (e.g. the adhesion phenomenon) must be taken into account in order to design reliable micromanufacturing systems. A new method of designing micromanufacturing systems is presented in this paper. Our approach combines the general “top-down” approach with a “bottom-up” approach which takes into account technical constraints. The method enables to build a modular architecture for micromanufacturing systems. In order to obtain this modular architecture, we have devised an original identification technique of modules and an association technique of modules. This work has been used to design the controller of an experimental robotic micro-assembly station.     

The use of fuzzy logic and neural networks models for sensory properties prediction from process and structure parameters of knitted fabrics

In a competitive business environment, the textile industrialists intend to propose diversified products according to consumers preference. For this purpose, the integration of sensory attributes in the process parameters choice seems to be a useful alternative. This paper provides fuzzy and neural models for the prediction of sensory properties from production parameters of knitted fabrics. The prediction accuracy of these models was evaluated using both the root mean square error (RMSE) and mean relative percent error (MRPE). The results revealed the models ability to predict tactile sensory attributes based on the production parameters. The comparison of the prediction performances showed that the neural models are slightly powerful than the fuzzy models.     

Action Recognition Using a Bio-Inspired Feedforward Spiking Network

We propose a bio-inspired feedforward spiking network modeling two brain areas dedicated to motion (V1 and MT), and we show how the spiking output can be exploited in a computer vision application: action recognition. In order to analyze spike trains, we consider two characteristics of the neural code: mean firing rate of each neuron and synchrony between neurons. Interestingly, we show that they carry some relevant information for the action recognition application. We compare our results to Jhuang et al. (Proceedings of the 11th international conference on computer vision, pp. 1–8, 2007) on the Weizmann database. As a conclusion, we are convinced that spiking networks represent a powerful alternative framework for real vision applications that will benefit from recent advances in computational neuroscience.     

Choice of the acquisition parameters for frequency estimation of a sine wave by interpolated DFT method

The accuracy of the frequency estimation of a sine wave corrupted by quantization noise using the Interpolated Discrete Fourier Transform (IpDFT) method is affected by systematic and random errors. The systematic errors are independent of quantization noise and are due to the interferences from the image parts of the sine wave spectrum. They depend of the acquisition parameter – integer number of recorded sine wave cycles. The random errors are due to the quantization noise and depend on the acquisition parameter – number of acquired samples. In this paper for the situations in which the IpDFT method uses the maximum side lobe decay windows a condition for the integer part of the number of recorded sine wave cycles is derived to ensure that the systematic errors are very small compared with the quantization errors. Also, a condition for the number of samples is derived to ensure that the absolute error of the normalized frequency due to the random errors is smaller than a desired value with a high confidence level. Carried out simulations confirm the validity of each derived condition.     

FAO’s AVHRR-based Agricultural Stress Index System (ASIS) for global drought monitoring

Agricultural production is highly dependent on climate variability in many parts of the world. In particular, drought may severely reduce crop yields, potentially affecting food availability at local, regional, and global scales. The Food and Agriculture Organization of the United Nations (FAO) operates the Global Early Warning System (GIEWS), which monitors global food supply and demand. One of the key challenges is to obtain synoptic information on a recurrent and timely basis about drought-affected agricultural zones. This is needed to quickly identify areas requiring immediate attention. The Agricultural Stress Index System (ASIS), based on imagery from the Advanced Very High Resolution Radiometer (AVHRR) sensors on board the National Oceanic and Atmospheric Administration (NOAA) and Meteorological Operational Satellite (METOP) satellites, was specifically developed to meet this need. The system is based on a methodology developed by Rojas, Vrieling, and Rembold over the African continent. This approach has been modified and adapted to the global scale to produce an agricultural stress index (ASI) representing, per administrative unit, the percentage of cropland (or pasture) areas affected by drought over the growing season. The vegetation health index (VHI), based on normalized difference vegetation index (NDVI) and temperature anomalies, is used as a drought indicator. A fused time series of AVHRR data from METOP and NOAA was used to produce a consistent time series of VHI at 1 km resolution. Global phenology maps, indicating the number of growing seasons and their start and end dates, were derived from a multi-annual image set of SPOT-Vegetation (1999–2011). The VHI time series and phenology maps were then combined to produce the ASI for the years 1984 to the present. This allowed evaluation of the suitability of the ASIS to identify drought using historical reports and ancillary data. As a result of this analysis, ASIS was positively evaluated to support the FAO early warning system.     

A framework for virtual organization requirements

Virtual organizations (VOs) are formed by an alliance of organizations linked by a partnership for dealing with emerging challenges. Information and communication technologies play a fundamental role facilitating cooperation, communication and collaboration among the VO members. The formal identification and representation of Requirements Engineering (RE) for one organization have been researched to a large extent along with several elicitation techniques. However, these techniques are not adequate for covering the major challenges of RE for VO. We explore the work done in the management and RE fields to propose a model-based framework for eliciting VOs’ requirements. The goal is to analyze the VO from two points of view: border (intra-organizational, inter-organizational and extra-organizational) and abstract (intentional, organizational and operational). This article includes the framework validation with a case study and a transformation process to develop partial Business Process Diagram from the intentional models.